INFORMATION MANAGEMENT FOR SUPERMARKET CHAIN PRODUCT MIX DECISIONS: A SIMULATION EXPERIMENT Thesis fer the Degree of Ph. D. MICHIGAN STATE UNIVERSITY JOHN FREDERICK GRASHOF 1958 (Hm. This is to certify that the thesis entitled INFORMATION MANAGEMENT FOR SUPERMARKET CHAIN PRODUCT MIX DECISIONS: A SIMULATION EXPERIMENT presented by JOHN FREDERICK GRASHOF has been accepted towards fulfillment of the requirements for Ph . D degree inBlISINESS . 4L Major professor Datew 0-169 ABSTRACT INFORMATION MANAGEMENT FOR SUPERMARKET CHAIN PRODUCT MIX DECISIONS: A SIMULATION EXPERIMENT by John Frederick Grashof In recent years the number and complexity of product rnix decisions in supermarket chains have increased rapidly. 'The increase is a result of increases in the rate of intro— o Homsm, Honda cameo op Bond 30: mnowwo Cdamoacm condom ohopm on» yo xoopm on» 0» Bond 0:» UU< .0 mOVHGE US“ 500d acamaoov on» mopmdadbo dodmaoou on moxoe pap soon on» mopcdaubo Homsm_ e Hohpp mango op Bond So: whowmo Guamoacm chopm cnpm Mo xOOpM 0:9 on Head on» 664 scamaoow a nexus can send one J Hchdn Gauze op Bond So: whommo swamp cm HHH pdfiHom HH nuance H pdEHOL scapduod posuonm Mom mpdahom coamdooo obdpdnhopam conga “Him mmbon 65 of the item. He merely relays his informatbn on the item and his evaluation with respect to the desirability of the item to the buying committee. The buying committee then considers the item and, using the information presented by the buyer, the buyer's recommendation, and its own exper- ience, makes a decision on the item. The same decision alternatives as available to the buyer under the first format are available to the buying committee. The third format for addition decision procedures is illustrated in Format III of Figure 3-1. Although the most complicated of the three, the third format is essentially a combination of the first two formats. The procedure is initiated with a new item presen- tation to the chain by a salesman just as are Formats I and II. Following the presentation the buyer analyzes and eval- uates the item. 3 Based on his analysis and evaluation the buyer, as in the first format, makes a decision. However, rather than accept or reject, as in the first format, the decision is to reject or not-reject. Thus, while the buyer has the authore ity to reject items and thereby keep them off the shelves of the chain's stores, he does not have the authority to add items. Those items the buyer does not reject are presented to the buying committee. Following discussion and analysis the buying committee makes a decision among the same alter- natives as presented above. Thus, in chains using the third format, the buyer can reject but can not accept items. The 66 buying committee can not only reject items, but can also accept items. Table 3-2 summarizes the three decision formats. TABLE 3—2: The three decision formats source of the possible the decision decision alternatives Format I Buyer Accept, Reject, Hold Format II Buying Committee Accept, Reject, Hold Format III 1) Buyer 1) Reject, Not-reject 2) Buying 2) Accept, Reject, Committee Hold While not directly relevant to the buying process, the composition of the buying committee mentioned above is of some interest. There is no set number of persons on the buying committee, with the size varying from as few as three persons to as many as ten. Despite the variety in the number of persons, there are always persons with three types of specialty present. The three specialties are: 1) a supervisor or member of middle management, 2) a buyer, and 3) a field representative or merchandiser. The first type, the manager, will be the head of purchases and/or his assistant or, in some cases, the head grocery buyer. The fun- ction of the manager is to direct the meeting, insuring that decisions are made and that such decisions are made carefully. The functions of the buyer(s) arettp present new items, to help in the evaluation of the items, and to vote and thus aid in the decision. While in most chains all buyers are in attendance for the entire meeting, there are some cases 67 where a buyer attends only long enough to present and vote on his new items. In these cases the buyer participates only in decisions which affect directly the items carried in the product families for which he is responsible. The third specialty represented on the buying commit- tee is sales. The member may be called a field representative, a merchandiser, or a sales manager but his duties are the same. He will have knowledge of the chain's stores and the problems faced by the stores. In particular, he will be aware of the condition of the various departments in the stores. For example, he must be aware if items have been added to a department without increasing the shelf space allocated to the department and the department is, therefore, overcrowded. Further, just as the buyer concentrates on how and what to buy, the third member of the committee concentrates on how and what to sell. He must be sensitive to the likes and dis— likes of the chain's customers. He must be aware of which pro- duct areas are increasing in consumer acceptance, which areas are losing consumer acceptance, and what items have been scheduled for special promotions. Based on his background, the third member of the committee acts as a sales advisor, as well as participating in the discussions and voting. The Criteria Employed Despite the difference in the format of the buying pro- cess indicated above, the criteria used by all chains is nearly the same. While the emphasis may vary slightly from chain to chain, the factors evaluated in reaching a decision on a new item offer do not vary. 68 There is little question that the most important cri- teria in all chains is demonstrated consumer demand for an item. If a salesman can prove to the chain that his item will sell and sell well, expanding the market for the product family, then that salesman has assured the acceptance of his item. However, only rarely, if ever, can the salesman prove consumer demand for his item. After all, the item is sup- posedly new and thus would not have been in the market where it could prove itself. Therefore, in most instances, the item offered is evaluated on the basis of two other gen- eral considerations. The first general consideration is an attempt to apply the consumer demand criteria mentioned above. It can be roughly stated as "What is our (the chain's) estimate of the consumer demand and sales for this item?" Since the chain can not know beforehand what the demand is going to be, several other criteria are used; criteria that are supposed to be indicative of consumer demand. The first and foremost of the secondary criteria is the promotional program of the supplier. If the new item is being supported by a large, expensive promotional pro- gram, the chain will give a high estimate to the expected level of movement. The chain will be even more favorably impressed if, in addition to a strong national advertising program, a strong advertising program supported by couponing and/or sampling is to be conducted in the chain's local market. Further, chains respond to guaranteed advertising programs rather than programs dependent upon distribution. 69 A second factor evaluated as part of the estimate of consumer demand for the new item is test market data. A great many new items, particularly those that are product innovations, are test marketed prior to national introduc- tion. The results of such test markets can then be used to estimate the extent to which the new item expanded the demand for a product family rather than merely switch customers within a product family. The data, though gen- erally not directly transferable, can also be used to estimate sales in the chain's local market area. The third factor used by chains to estimate consumer demand for a new item is the sales history of competing items. As part of the analysis and evaluation of a new item, the buyer extracts from the firm's records the move- ment history of items in the same product family which the chain now carries. The movement history is then examined to help estimate the expected demand for the new item. Particular attention is given to trends in the sale of the product family and to shifts in sales patterns within the product family. The fourth factor affecting a chain's estimate of the consumer demand for a new item is the attitude of competi- tion as reflected by the competitive chains that have added the item to their product lines. If all of a chains com- petition in a market are carrying an item, that chain is likely to believe it made a mistake in not accepting the item before, that consumer demand for the item does exist, and that they should add it to the list of items now stocked. 70 The second general consideration chains use in eval- uating new items is the effect of the item on the mix of items carried by the chains. If an item is completely new and different, then a chain may well add the item, even if they are unsure of the level of consumer demand for the item. The rationale for such a decision is the desire of chains to prosent a wide range of items to the consumer. The factors considered when chains evaluate an item under the general consideration of product mix are: Unit cost Unit retail Unit size Number of items with which it competes Similarities with competing items Differences with competing items Sales of competing items \Jown¥1un)4 VVVVVVV There are several other factors, in addition to the above, which chains consider when evaluating a new item. One of the factors is the reputation of the firm offering the product. The chains consider the dependability and reliability of the firm with respect to service, and also consider the firm's reputation for ethical business dealings and successful introduction of new items. A second factor considered is the gross margin per cent that the manufacturer suggests. Chains, of course, would like all items to have a high gross margin but realize that for some items a high gross margin is impossible. Further, the gross margin on an item tends to decrease with time rather than increase. Thus, a new item which comes in with a low gross margin is not attractive to the chains. Third on the list of specific factors considered in evaluating new item offerings are the introductory deals and 71 the promotional allowances available. The opportunity exists for chains to increase profits greatly through careful evalua- tion of the introductory offers. For example, it might be profitable for a chain to add an item on a promotional basis only, selling the item at a very low price, because of the introductory deal available. Similarly, the net profit accruing to a chain from carrying an item can be increased if the manufacturer offers a generous promotional allowance. Chains consider such promotional profit during their evaluation of the new items offered. A fourth factor considered is the quality of the pro- duct's handling characteristics. Due to the tremendous expense incurred in handling goods, chains are extremely sensitive to the physical characteristics of goods, and to the method of packing and delivering the goods. Square packages that are easy to stack are much more appreciated by chains than odd shaped bottles. Further, chains are concerned as to whether the case pack seems reasonable given the anticipated level of sales for an item. Chains do not like to have items that sell only 5 units a week come in cases of 48 units. 0 Most chains today are becoming sophisticated enough in the technique of materials handling to appreciate the advantages of palletized shipping. Some even go so far as to insist on palletized truck or rail car loads and extract a penalty from manufacturers not complying. CloSely related to the aspect of physical handling characteristics are freight allowances and/or back haul 72 privileges. Freight allowances are straightforward and need no further discussion. Back haul privileges, on the other hand, are not as usual. Chains generally maintain fleets of trucks for warehouse to store distribution. The delivery of goods to a store is a one way trip with the trucks returning empty. In the instances where empty returning trucks pass within a reasonable distance of a manufacturer's plant or storage ware- house there is a tremendous financial advantage to the chain if the trucks can pick up the goods ordered from the manufacturer and haul the goods back to the chain warehouse. The use of the chain's own trucks in such cases saves the freight on the goods. Some manufacturers allow back haul while others don't. At present the legality of back haul privileges is being questioned by the Federal Trade Commission.1 However, until ruled illegal, back haul privileges will be a factor in the purchase decisions of chains. Information Available on New Items The following few paragraphs will discuss briefly the information presented to the buyer by the salesman. There is other information used in the addition decision but such infor- mation is internal to the firm and will be discussed later. The salesman uses three vehicles to get information to the chain buyer. The first of the three vehicles is the verbal 1For a discussion of the present poSition of the Federal Trade Commission with respect to "back haul" privileges, see "F.T.C.: Allowances for "Back-Haul' Orders are Possibly Illegal," The Marketing News, Semi—monthly newsletter of the American Marketing Association, Volume 1, #7, Febuary 15, 1968, p. 1. 73 presentation of the salesman to the buyer during the interview. The salesman repeats what is presented in writing and answers a few questions. The fifteen minutes generally aldowddifor interviews do not provide much time for involved sales present- ations. The second form of communication between the salesman and the buyer is the brochure presented to the buyer by the salesman. Ranging from nothing (the brochures are optional) through one typewritten sheet to extensive multipage promotional pieces prepared by the manufacturer's advertising agency, the brochures again reiterate what is on the "new item form." However, the information is presented via diagrams, charts, pictures, and phraseology designed to "sell" the item. The third vehicle for the flow of information from the salesman to the buyer is the "new item form". A "new item form" is an information sheet developed by a Chain which the salesman must fill out prior to his interview with the buyer. The salesman then presents the new item form generally with samples of the new item to the buyer during the interview. There are spaces on the form for every piece of information the chain deems important and the salesman can supply. The exact layout of the new item form varies from chain to chain. However, there is little or no variation in the information requested. Seven general categories of information are requested, each requiring several specific facts. The seven general categories are listed in Table 3—3. Appendix C is an example of a typical "new item form" and illustrates the specific pieces of information requested. 74 TABLE 3-3: The seven general categories on new item forms category information I Description of the new item II The types and amounts of guar- antees on the item III The types and amounts of al- lowances on the item IV The types, amounts, and sched- ules of advertising and sales promotion efforts V The competitors stocking the item and the retail price VI Shipping information VII Store handling information The Deletion Decision The disucsSion will now turn to item deletion deci— sions, explaining how items are identified for possible deletion, the source of the deletion decision, and what criteria are used. Following the discussion of the item deletion decision, the nature and sources of internal infor— mation will be discussed. The order of presentation results from the fact that the internal information is used in both addition and deletion decisions. Thus, the discussion logic- ally follows the discussion of both decisions. How Items Are Identified for Possible Deletion The problem of identification of items that should be deleted, or at least considered for deletion, is one of the most serious facing chains today.2 The squeeze on shelf space 2Stated by executives of several chains during per- sonal interviews in August 1967. 75 brought on by the increasing number of new items means that the unprofitable items now stocked must be identified and removed from the shelves of the chain's stores. Supermarket chains have two procedures for the iden- tification of items that should be considered for deletion. The first procedure requires that for each new item added an item will be dropped, preferably in the same product family. Thus, when a buyer presents a new item he is expected to sug— gest an item to be deleted. Obviously, since the number of items carried by chains is increasing, the rule is not adhered 1 to 100 per cent. One chain executive estimated that an item was dropped between 25 per cent and 50 per cent of the times a new item was added.3 The second procedure for identifying items to be deleted is a periodic review of all items a chain carrys. The review is accomplished in a variety of ways by chains. One method for the periodic review is for the buyer for each product family to regularly review all items in his section and identify those items he feels should be dropped. The examination may take place weekly or periodically, or it may be continuous with the buyer rotating the product family he examines. A second method is for the head buyer and/or his assis— tant to examine all items carried by the chain. Such a review is generally on a periodic basis with the evaluafion based on a summary report of each item's performance during the period. 3Suggested during an interview with an executive of a major chain, August 1967. 76 Items with levels of performance below certain acceptable lev- els are identified and considered for deletion. The Source of the Deletion Decision There are two general approaches to the deletion deci- sion. (In. Figure 3-2 the "Format" refers to the Format used for the buying decision as illustrated in Figure 3-1.) Respon— sibility for the deletion decision depends on the procedure followed by the chain for addition decisions. For chains fol- lowing the first format (a buyer completely responsible for the addition of new items) the buyer is completely respon- sible for the deletion decision. While he may consult with his superior concerning the decision, the buyer, since he is generally responsible for the profitability of his product families, assumes the ultimate responsibility for deletion decisions. In chains with a buying committee (Formats II and III under the buying decision) the deletion decision is made by the committee. As noted above, when a buyer makes a presen- tation of a new item he is expected to suggest an item for deletion. In making its decision on the new item the com- mittee will consider the possibility of deleting an item. The deletion decision, however, is not tied to the addifion of a new item, but rather is a separate decision, made by a sep- arate vote of the committee, with a majority of the comittee carrying the decision. In addition to presenting new items to the committee, the buyer will also, on occasion, present items he feels should be deleted. Such presentations may be made even if the buyer 77 cocaupoa on send one a Suva 0:» cacpom noumaooc names 1 Bond on» mopupaobo ooppaeao wcahsm cocacpoh ma Bond 0:8 l Bond on» :«upom Godmdooc on moxde.psn Bond on» mopdsadbo nohsm . Sodboh Bond 30: a mo soap tdstbo mo cad» pd scapoaou mom copmomwSm EopH Sop“ nodmdooo on» gong a nexus I one son: on» moposacbo Hohsm 4‘ cacoahom sweeps» codpoaoo canammoa Mom codmdpona ma sopH zodboa cacoanom smacks» noapoaoo oHdemon How vwmmdpnocd ma aopH HHH a HH essence weasam scapoaou posoona pom mpdfihom nodmaooc obdpcnhopao 039 H pdfihoh wdamzm “Nlm mmDUHm 78 is not suggesting a new item to replace the one considered for deletion. The buyer and the committee know that an item that is not generating sufficient profits may be deleted at any time. Part of the rationale behind such a decision is that new items are being constantly added to all product fam- ilies and therefore, shelf space is always needed. The Criteria for Deletion Decisions An important criterion for the addition of a new pro- duct is demonstrated consumer demand which results in a hgih rate of sales for the item. It is, therefore, not surprising that the most important criterion for item deletion is a low rate of sales or lack of movement. The prime factor in the identification of items for possible deletion, as well as the most important criterion used in the decision is lack of dem- onstrated consumer demand for an item as indicated by a low rate of sales for the item. Another important criterion is the level of the gross margin of the item. Any item with a particularly low level of gross margin, especially when compaired with other items in the same product family, will probably be considered for deletion. The two factors mentioned above can be combined to provide a third criterion - gross margin dollars generated per unit time. Although generally secondary to movement the gross margin dollars generated per unit time is important because it permits compairsons between dissimilar items in a product family, and in some cases across product families. 79 For all three of the above criteria many chains con- sider the trend more important than the absolute level at any point in time. For example, if movement is low but seems to be increasing then the item probably would not be dropped. However, if the movement shows a decline, then even an item with a fairly good level of movement might be considered for deletion. A fourth criterion used by chains is not as easily applied as those mentioned above. The chains, in their desire to maintain variety on the store's shelves, will hesitate to delete one-of—a-kind items. Most chains feel an obligation to carry as broad a product mix as possible. Thus, an item with slow movement might be "saved" if there are no substi4, tutes for the item. However, items in the one—of-a-kind cat- egory are continually appraised and should movement fall to an extremely low level the item would be dropped. Occasionally an item might be dropped due to a need for shelf space in its product category. Should a chain find an item must be added, probably due to consumer demand for the item, and also find that there was no space on the shelf for the item, the chain might make room for the item by deleting an item now stocked. The item dropped would be the least attractive item, based on the criteria discussed above. The Nature.Sources and Flows of Internal Information The buyer's card is an important source of internal information in most supermarket chains. The cards are a set of 5 X 8 index cards, one for each item carried, maintained 80 by and for the buyer. Not only do the buyers' cards provide a record of the activity and actions on an item but also are a primary source of decision making information. Essentially the buyer's card is'a perpetual inven— tory record which is updated weekly for sales and as neces— sary for receipts of an item. The card contains a record of the total cases shipped from the chain's warehouse to the stores each week, the number of cases ordered and received, and the date of each orderand receipt. In addition, the card generally contains notations as to the nature and ex- tent of promotions on items, the current cost and retail, ineluding any changes. The card also contains basic product information such as supplier, size, case pack, case cube, pal- let count, and minimum order quantity. From the data contained on the buyer's card, the buyer extracts the information he uses t0 make addition and deletion decisions. While the information used most often is the move- ment of the item, other information is also used. As Part of the analysis of new item offers, buyers generally make compar- ison tables of competing items. Figure 3-3, an example of such a table, shows the information used in the comparison. All the information contained in the table can be extracted from the buyer's card. Buyers' cards are maintained manually, either by the buyer or his secretary. In addition several chains are storing the Same information within their data processing system. At least one chain has gone one step further and reduced the 81 FIGURE 3-3: A table for item comparison ITEM PACK SIZE UNIT UNIT GROSS SALES COST RETAIL MARGIN LAST AVERAGE 'PERIOD PERIOD duplication of information by eliminating the buyers' cards. The cards' functions are now performed by reports of the data pI‘Oc e ssing system. Electronic data processing or management information Systems are making important advances with respect to con- tI‘OZLling the flow of items through supermarket chains. A EOOd example of such advances is the "short and expedite re— POI't". Generated daily, the short and expedite report lists SuCh items as those for which store orders could not be filled due to inadequate stock in the chain's warehouse. Based on the report, buyers can take the necessary steps to correct the deficient inventory problem as are indicated necessary by inVestigation of the problem. Very similar to the short and expedite report is the I‘ee'eivings report. Each day data processing prepares a list Of all items received in the chain warehouse that day. The list can then be used to eliminate items from the short and expedite report that were received after the report had been generated. The report can also be used to evaluate vendors. 82 _ While very useful from a control pdint of view, the above reports are not product mix decision making information. The application of data processing and information system technology for decision making in supermarket chains is gener- ally. behind the control applications. A series of weekly and period summary reports are generated for both control and decision making. The reports are activity summaries for eaCh item containing the basic item data as well as summaries of the cases received and shipped, dollar sales, gross margin per cent, gross margin dollars generated, the inventory level in days Supply, and the number of inventory turns. While most of the {information is used to evaluate and control performance, much 0f it also can be used in support of product decisions. Even though the summary reports mentioned above, and one or two other similar reports can be used in decision making, the main purpose of the reports is control. Buyers make little, if any, direct use of data processing or information systems for product addition and deletion decisions. For example, most Chain buyers are aware that certain items sell better in some stores than in others due to the particular customer character— iStics. Yet few chains evaluate the mix of items ordered by each store. The technology is available to maintain records of ship- ments to stores on a per item per store basis. The information made available form such records may be immensely useful in evaluating both new items offered and items being considered for deletion. For example, take a hypothetical 100 store chain. The buyer for beans has suggested for deletion chili beans whiCh are Currently moving at the rate of 20 cases per week. The rate 83 of movement is an average of 1/5 case per store per week. Due to low movement the item might be deleted. Analysis of ware- house-to-store shipments might show, however, that the item was ordered by only ten stores, each store selling an average of two cases per week. The chain-wide average, in the above case , is quite misleading and the chain would probably retain the item for the stores that have the high rate of sales. Re la ted Uses of Computer Based Information Systems Many chains have made, in the planning stages at least, improvements in their information systems designed to provide more and better information for decision making. Chains, during the introduction of data processing systems, first applied the systems to routine problems such as pauroll, and accounts receivable. With the routine problems solved, chains are now turning to more sophisticated uses of data processing. One of the uses chains are investigating is computer- iZed evaluation of the customer mix of each store. The re- sults of such evaluation can be used to identify items that have appeal to ethnic, racial, religious, or social groups. Another example of the proposed use of computer based information systems is the evaluation of promotional deals and announced price changes. When a promotional deal is being discontinued or the cost of an item is being increased it is profitable for chains to stock up on the item at the pr'eSent lower price. At some point, however, the costs of StoI'age become greater than the savings resulting from the lower purchase price. The computer can easily compare the 8% the savings and the costs and identify, based on the dollar costs and savings, the optimum quantity to be purchased. A third use to whihh chains are comtemplating putting their information systems is the evaluation of in-store pro- motions. End-aisle and similar displays are not only costly to set up and maintain but also take up valuable space. Chains realize the importance of evaluating the costs and returns of such displays and then using the information in planning 0 the r promotions . My, The previous paragraphs have discussed the research deSign used to gather data on the present product addition and deletion decision procedures used by supermarket chains. Based on the data gathered, a case study of the decision pro- cess was developed. The case study illustrated the three general formats for the product addition decision and the two general formats for the deletion decision. In addition, the case study iden— tifi ed the criteria now used in the decisions and the infor- Ination available to support the decision. The information presented above has been used as a baSis for suggesting improvements in the decision criteria and the flow of information. The following chapters discuss the methodology used to identify and test the suggested improvements, the results of the tests, and the implica- ti011$ of the research for chain management. CHAPTER IV RESEARCH DESIGN: THE SIMULATION EXPERIMENTS Introduction Phase II of the research project consists of a series (If eez Initial trend factor Developed from the sales for the forecasting histories of 76 pet food routine. items over a 52 week period in an actual chain. '15?) Initial seasonal factor Same as 14). for the forecasting routine. ‘165) The lead time for each Set by researcher. supplier. 1 7’) Identification of the Extracted from the records items carried by each of a national chain. supplier. .153) Item data including: Extracted from the buyers' case pack, case cost, cards for the items under case retail, handling study. cost, initial inventory. W . . . . be different for each store 1n the chain if so deSlred. l’n 88 icicliigan area. In particular, the number of items carried by the: chain and the characteristics of the items including :mlplnlier, cost, retail, and case pack, were extracted from the; (order book of one chain. The market share of each of the itxanls was derived from the movement history of the items. The maximum shelf quantity of each item was determined by observa- tir>r1 of the actual quantities on the shelf in a store. The data needed for the exponentially smoothed seasonal arufl. "trend adjusted sales forecast were developed from the weekly sales records of one chain. The sales history of sev- enty-six pet food items for a period of fifty-two weeks was decomposed into trend and seasonal parts through a linear re- gression analysis on the computer of the University of Detroit. rl‘he program returns not only the slope of the regression line (‘tllez trend factor) but also the residuals (the differences bertaneen the observed value and the values computed from the trend equation). The residuals are, therefore, a seasonal fafiltmor (assuming zero random error) for each of the fifty-two “Emelcs. The fifty-two week year was then transformed into thirteen four-week periods, in order to smooth out the effect OI‘ Efuch factors as weekly promotions. The seasonal factor for ea<311 of the periods was assumed to be the arithmetic average 0f. 1ihe seasonal factors of the four weeks that constitute the pel7i43d. The trend factor determined by the linear decomposi- ti(311 of the sales history data is 0.36612. The seasonal factors for each of the thirteen periods are presented in Table u-a. The handling costs associated with the items in the 89 TABLE H-2: Seasonal indices for pet food average seasonal period residual index 1 0.70196 100.70196 2 0.16838 100.16838 3 0.11836 100.11836 4 2.82060 102.82060 5 -6.39289 93.60711 7 -H.72980 95.27020 8 9.78266 109.78266 9 9.5%183 109.5M183 1o -o.5h183 99.u5817 11 -0.59371 90.H0629 12 -9.38381 90.61619 13 0.96765 100.96765 SOURCE: Calculated from the weekly item sales histories for fifty-two weeks of seventy-six pet food items carried by a chain operating in Detroit, Michigan. .Eszinnulation were estimated from the costs compiled by LIVI<3Kinsey and Company.1 While not developed specifically for <51<=>g foods, the McKinsey data are the most recent estimates <:>-1T' the costs of handling dry grocery items. The costs are ‘5‘ <2<3epted by the industry and can be applied to items other ‘tzfilfilzin.the.ones specifically studied by McKinsey. The factors of case pack, type of container, and size <:>:IT' container were used to apply the McKinsey data to dog fQOd. For example, canned dog food items packed in cases <:>;1SI twenty-four were assigned the handling cost of canned :ETEJT’Wulit ($0.79). Canned dog food items packed in cases of :E? <:>‘3F'ty-eight were assigned the handling cost of canned soup (2;. 1"The Economics of Food Distributors," McKinsey- QQ\§ literal Foods Study (White Plains, New York: General Foods :1?;poration, October, 1963). 90 ($0.74) as developed by McKinsey. Estimates for the other dog food items were developed from the costs of items similar in package size and case pack studied by McKinsey. Table l+—3 presents the various types of dog food items, the comparable item studied by McKinsey, and the handling cost. Exclusive of the input and output sections, the CHAINSIM program consists of five activity routines. See Appendix D for a flow chart of the CHAINSIM program. The first activity routine is generation of store-to-customer sales. Using Monte Carlo simulation techniques, the program generates a preset number of customers per day for each store.2 A random number between 0 and 100 is generated by a random number generator. The random number (customer) is then matched with a particular 1 tem. A cumulative market share is calculated by sequentially adding the market share of each item. The cumulative market Share is the device used by the routine to match the random number with an item. To illustrate, suppose the random number generated was 20 . 2347. The computer compares the random number with the ma Iket share of the first item. If the market share of the fi I'st item is less than 20.23’+7 the computer would add the market share of the second item to that of the first. Again the computer would compare the random number with the market Share, using as market share the cumulative total of the \ 2Monte Carlo simulation techniques are experimental J:;.:S:_E:Ltrlpling techniques that can be used to model processes 1(:h are essentially probabilistic. For an introduction 1the concept of Monte Carlo simulations see Ronald E. Green, Quantitative Methods in Marketi ( Ink and Paul E. n Englewood Cliffs, N.J.: Prentice—Hall, Inc., 1967) pp.89-95. 91 TABLE h-3: The various types of pet food packages and the McKinsey handling costs associated with the items item comparable McKinsey handling cost item - (dollars) 1 pound can canned fruit 0.79 (Case pack 2H) (Case pack #8) canned soup 0.7% ‘1 pound 10 ounce coffee 0.86 canned dog food :36 ounce package flour 0.51 of semi-moist C72 ounce package detergent 0.77 of semi-moist 22 pound box detergent 0.77 5; pound bag flour 0.51 1 0 pound bag detergent 0.77 225 pound bag cereal 1.17 ma rket shares of the' first and second items. If the cumula- 't:F£i.Wre total market share is greater than the random number (: :22(3,23h7) then the computer would recognize the second item EEI'ES the selection of the customer in question. If the cumula- CtzijL“fe market share was less than 20.23%7, the computer would Q a iculate a new cumulative market share by adding the market Share of the third item to the cumulative market share of the £13? st and second. The procedure outlined above continues 1u1;t:lft3lil the cumulative market share is equal to or greater than ‘tzjth‘Ee random number. When the cumulative market share becomes Squel to or greater than the random number generated, the <2:<:) InclIDuter uses as the item to be purchased the last item added 92 'tc> obtain the cumulative market share. Once the computer has identified the item the customer raj.3hes to purchase, the store shelf inventory is checked to (ieetermine if the item is available on the store's shelves. 12f the store has the item on the shelf the purchase is re— czorded and the computer recycles to the next customer. If the item selected by the customer is not on the sshelf of the store, the computer generates a second random :rlumber. The number generated is compared with a preset .Iulnnber indicating the probability that the customer will siccept a second choice item. If the customer will accept a ssecond choice item, the second choice item is identified IJnsing the original selection of the customer. For every item (:airried, a second choice item is read as input data with the =3'torage address of the item based on the number of the first '<2]noice item. Again the shelf inventory is checked, now for ‘tSlne second choice item. The program recycles to the next <:=Ilstomer if the item is not on the shelf. If the item is ‘Ei“failable it records the purchase and then recycles to the ‘1:1:f the item and the quantity of the item now on the store's =Eilielves. If one or more cases of the item will fit on the sl'ielves, the item and prOper quantity are placed on the store The computer then proceeds to the next item until all If the avail- Q I‘der. :i-'13r both addition and deletion decisions. The inputs to the BUYSIM program are presented in Table1+5. VVi.th.the information given in the table the computer auto- nfléftically calculates the values necessary to rank the items according to seven decision criteria. If the program is being used to evaluate new items, the J?911?st step is to estimate the level of unit sales for the new j‘13enn. The basis of the estimate is the average weekly unit 53531435 of the currently stocked items selected for comparison. 51313£3 estimate is then adjusted by the test market data, and 98 TABLE h-5: Data input to BUYSIM number description source 1) Item number and name Set by researcher 2) Case pack Records of a national chain 3) Case cost Records of a national chain 4) Case retail Records of a national chain 5) Gross margin dollars Item H minus item 3. per case 6) Gross margin per cent Records of a national chain 7) Handling cost per case Adapted from McKinsey data 8) Four weeks unit sales Records of a national chain 9? Test market data Set by researcher 107 Rating of the intro- Set by researcher ductory program 117 Rating of the national Set by researcher advertising program 123 Rating of the local Set by researcher advertising program 13? Rating of the com— Set by researcher petitors'reaction 1%? List of items for Set by researcher comparison 15) Number of items to be Set by researcher ranked 16? Number of new items to Set by researcher be ranked *Information needed only if the program is to be used to rank new items. the ratings of the introductory program, the national adver- tising program, the local advertising program, and competitive reaction to the item. Appendix F presents an example of a "New Item Evaluation Form" that would be filled out by a buyer to provide the necessary input to the program. Follow- ing calculation, the estimated movement of the new item is used by the computer to calculate the information necessary to rank the items. When the program is not being used to evaluate new 99 item offers, but rather is being used to rank only items now stocked, the above section of the program is skipped. The computer proceeds directly to the calculation of the values for the criteria. Of the seven criteria used in ranking the items, five must be calculated. Table 4-6 lists the criteria used in the BUYSIM program. TABLE H-6: Criteria used to rank items in the BUYSIM program number criteria 1) Movement (in units) per week 2)* Gross dollar sales per week 3) Gross margin per cent )* Gross margin dollars generated per week 5)* Dollar contribution per week 6)* Net profit generated per week 7)* Weighted summary ranking *The values for the criteria must be calculated by the program using the rate of movement and the item characteristics. Once the necessary values for the above criteria are calculated, the BUYSIM program calls a subroutine to sort and rank the items in descending order. The items under consideration are first ranked according to each of the first six criteria listed above. Then, a weighted summary value is calculated for each item. The rank each item received accord- ing to each criteria is subtracted from the total number of items plus one. The values for each item are then added across all criteria. When the summary values have been calculated the items are ranked according to the waighted summary values. 100 The output of the BUYSIM program is a table listing each item evaluated. (See Appendix G for examples of the output table of BUYSIM.) For each item evaluated the table indicates the value and the ranking of the item according to the criteria, and the weighted summary ranking. SPACALLO: Linear Programming Allocation of Shelf Space The SPACALLO program uses the optimization character- istics of linear programming to allocate available shelf space to individual items, given a predetermined management goal. The program is not intended to solve space allocation problems, but rather, to illustrate the impact of alternative decision criteria on item evaluation. Rather than design a linear program routine specifically for the present research, the computer's library linear programming routine is used. The input to the routine consists of item data, a set of constraints and an Objective function. The item data needed are: The width of one facing of the item. The number of units in one facing. The number of facings per case of the item. The relationship between a change of one facing and the sales of the item.)+ rwm—s The constraints imposed on the solution are: 1) A maximum number of cases of each item that may be stocked. 2) The total linear shelf feet required for the items may not exceed a preset maximum. 1+While thenais little concrete evidence that additional facings of an item result in higher sales volume for that item, many supermarket operators feel that additional space will in fact sell more of an item. Research in the area provides con— flicting results. See for example Keith K. Cox, The Relation- ship Between Shelf Space and Product Sales (Austin, Texas, 196%) 101 The criterion (objective) function may be changed for various runs of the program. In particular, the program may be used to allocate shelf space so as to maximize: Unit sales Dollar sales Gross margin per cent Gross margin dollars Dollar contribution \nrwm —* vvvvv Thus, the criterion function must be adopted to the particular problem being solved. As output, the routine will specify the number of facings that should be given to each item. The output will also in- clude the value of the objective criteria. The Simulation Experiments Introduction The sets of experiments described in the following paragraphs are designed to provide the data necessary to test the hypotheses listed in Chapter One. The description is presented in two sections with each section designed to pro- vide data on a particular aspect of the hypotheses. The two sections are: 1) Identification of the Effect of Alternative Decision Criteria on Item Evaluation. L+(cont) However, a research report by the U.S.Department of Agriculture did find that for seventeen canned fruits and vegetables each additional facing increased sales by ten per cent. The ten per cent value reported by the Department of Agriculture was used as the sales response to additional space coefficient for the present research. For some items, such as twenty-five pound bags of dry dog food, the coefficient was adjusted. See Hans Pauli and R. W. Hoecker, Better Utilization of Selling Space in Food Stores: Part I. Relation of Size of Shelf Display to Sales of Canned Fruits and Vegetables, U.S.D.A. IMarketing and Facilities Bureau, Marketing Research Report #30 (Washington, D.C.: U.S.Government Printing Office, 1952). 102 2) Sensitivity Analysis of the Effect of Minor Fluctuations in the Input Data on Item Deci- sions. Identification of the effect of alternative decision criteria on item evaluation The first set of experiments is designed to identify the impact of the use of various decision criteria. The hypothesis to be tested by the results of the first set of experiments is: H01: The ranking of each item in a set of items will not change when the criteria used for the ranking is changed. Both the BUYSIM and the CHAINSIM programs, as well as an analysis routine, are used in the experiments. The first step is to categorize the dog food items and place each item in one of four product families. The four product families are: Canned ration type dog food Canned all-meat type dog food Dry (meal) type dog food Semi-moist dog food :UJIU —* VVVV The next step in the experiment is to make four runs of the BUYSIM program, using as input for each run all the items in one product family. After each run the rankings of items 'using the alternative decision criteria are compared. If the (ramparison of the rankings shows that the rankings vary using the alternative decision criteria then the results will have provided evidence that the particular items selected by a chain.are dependent on the criteria used in the selection process. 103 Item data are available on fifty-two dog food items stocked by the Detroit Division of a national food chain. Table H-7 presents the product family categorization of the fifty-two items. TABLE H-7: Product family categorization of 52 dog food items . total number number of items Product family of items used in CHAINSIM* Canned ration type 12 9 dog food All meat and gourmet 1h 10 Semi-moist dog food 12 9 Dry meal dog food 1% 10 Total 52 39 *Approximately 75% of the total number of items. A computer program was written to analyze the results of the rankings of the items by the BUYSIM program. The problem computes the sum of the absolute differences in the ranking of each item by each pair of criteria. The pair of criteria with the smallest sum of absolute differences in rankings is defined as the pair that is most similar. The pair with the highest sum of absolute differences is defined as the least similar pair. Further analysis of the rankings is conducted using Kendall's Coefficient of Concordance. Kendall's statistic answers the question, "How much do these rank orders tend to agree, or show 'concordance'?"S The statistic is a ratio of 5William L. Hays, Statistics for Psychologists (New York: Holt Rinehart and Winston, 1963) pp. 656-658. 104 the observed amount of variance in the rank sums and the maximum possible variance in the rank sums. The output from the second step in the experiment, the item rankings generated by BUYSIM, is used as the input data for the third step. The top seventy—five per cent of the items in each of the four product families, as determined by the item rankings, are used as the items stocked by the stores in the CHAINSIM program. Since the output of BUYSIM provides seven rankings of the items, seven different sets of items are generated to make up seven different departments. Thus, seven runs of the CHAINSIM routine must be made. In order to provide for comparability amongythe results of the seven runs of CHAINSIM, the same store characteristics and input data on the items are used. Further, the same set of customers are generated by using the same number to initial- ize the random number generator. Therefore, any differences observed in the output of the CHAINSIM routine from the seven runs must be due to the different set of items stocked in the department. A second experiment designed to show the effect of alternative decision criteria uses the SPACALLO program. The hypothesis tested by the results of the experiments with the SPACALLO program is: H02: The per cent of total available shelf space allocated to individual items by a linear program allocation routine will not vary when the objective function is changed from one to another of the fol- lowing criteria: a) Maximize unit sales b) Maximize dollar sales 105 c) Maximize gross margin per cent d) Maximize gross margin dollars e) Maximize dollar contribution A dog food department with sixty linear feet of shelf space is assumed. The mix of items stocked and a minimum space allocation to each item is also assumed. The space required for the minimum allocation of all products is forty linear feet or two-thirds of the total available space. The linear programming routine then determines the optimum allo- ,cation of the remaining twenty linear feet of space in the department. The amount of shelf space allocated to each item by the five runs, as specified by the hypothesis, is then compared. If the space allocations vary, the variations provide further evidence that alternative decision criteria, as manifested in various management goals, result in the emphasis (or selec- tion) of different items. In addition, evidence is provided on the specific effect of various management goals on the operating results of a chain. The sensitivity of item rankings to variations in the input data During the interviews with chain executives reported in Chapter Three of the dissertation, several of the executives indicated that they felt highly quantitative item evaluation or item evaluation by computer was impractical. Their reason was that price and the shelf space devoted to individual items changed so frequently that accuracy could not bezachieved. To guide an investigation into the correctness of opinion of the chain executives the following general hypothesis was 106 formulated: H03: Sensitivity analysis will show that an item's ranking by the BUYSIM routine will not change when the item character- istics of price and handling cost are changed. To provide greater structure to the research, the following more specific hypotheses were formulated. HO3-a: The ranking of an item by the BUYSIM routine will not change when the price of the item is increased by five, ten, and fifteen per cent. HO3-b: The ranking of an item by the BUYSIM routine will not change when the handling cost of the item is increased by five, ten and fifteen per cent. HO3—c: The ranking of an item by the BUYSIM routine will not change when the price of the item is decreased by five, ten and fifteen per cent. H03_d: The ranking of an item by the BUYSIM routine will not change when the handling cost of the item is decreased by five, ten, and fifteen per cent. The hypotheses are tested by selecting an item in one of the product families and submitting twelve runs of the BUYSIM routine to the computer. Two product families, the canned ration type dog food family and the canned all meat family, have been selected for examination so that results could be obtained for more than one set of goods. For each product family the item selected was that item given the median weighted summary ranking by BUYSIM. The median item was selected so that the ranking of the item when the data is changed could either increase or decrease. Table H-8 indicates the set of experiments carried out to 107 test the sensitivity of the rankings to variations in the input data. TABLE h-8: Experiments performed as part of the sensitivity analysis number experiment 1) Price increased 5% 2) Price increased 10% 3) Price increased 15% 4) Handling cost increased 5% 5) Handling cost increased 10% 6) Handling cost increased 1 o 7) Price decreased 5% 8) Price decreased 10% 9) Price decreased 15% 10) Handling cost decreased 5% 11) Handling cost decreased 10 12) Handling cost decreased 15% Evaluation of the hypotheses Each of the hypotheses listed above is evaluated using the results of the computer simulation experiments. The first hypothesis is rejected if more than three items are assigned different ranks by each pair of criteria. The second hypothesis is rejected if the space allocated to more than ten items changes significantly when the objective function is changed. Minor variations in the space allocated to an item should be ignored because the total space used by the department under each of the objectives may be different. The subhypotheses under hypothesis three are evaluated as pairs - one pair relating to price variations and one pair relating to handling cost variations. The pairs of hypotheses 108 are rejected when the rank of an item changes according to three or more criteria for each ten per cent variation in price or handling cost. Definitions 1) 2) 3) h) 5) 6) 7) 8) 9) 10) Computer model - A mathematical model or simulation which is specifically designed to utilize an electronic computer in the solution or operation of the model. Decision Criteria - The factors evaluated in reaching a decision. Direct Profit - The amount of money remaining after the cost of goods sold and the direct expenses-have been subtracted from dollar sales. Handling Costs - All costs associated with the physical movement of an item through the physical distribution system of a chain. Included are the warehouse costs, the cost of shelf space, and ringing-up and bagging the item. ~ Indirect Costs — All costs associated with the sale of an item which cannot be traced directly to the item. Such costs include the operating expenses of the central headquarters and the division headquarters.» Information Management - All activities related to selecting the kinds and amounts of information neces- sary at the several decision points within a firm and providing for the steps necessary to make the required information available. Information System - An organized structure composed of data collection, transmission, and analysis devices and the personnel through which data is collected, analyzed, and turned into timely, relevant information for decision making. Product Mix - The assortment of goods that is presented to the customers of a store. Product Mix Decision - Any decision which relates to the content of the product mix. There are essentially two decisions which affect the content of the product mix of a store: 1) the decision to add a new item and 2) the decision to delete an item now stocked. Quantitative Decision Criteria - Decision criteria for which quantitative or numeriCal decision rules can be developed. 109 11) Simulation — The description of a real world system or organism through the use of mathematical models of the actions of the components of the system and the inter- action of the components. Generally, though not neces- sarily, such models are designed for operation on an electronic computer. 12) Subjective Decision Criteria - Decision criteria for which quantitative or numerical decision rules cannot be developed. 13) Supermarket - A complete, departmentalized food store with sales of over one million dollars per year and at least the dry grocery section completely self-service. 1h) Supermarket Chain - A group of eleven or more super- markets operating regionally under the same management. Summary The computer programs and the experiments described in the above paragraphs had two major purposes. The first purpose was to illustrate that simulation could be used effectively to study the operations of supermarket chains and that computer technology could be used in the everyday operations of a chain. The second major purpose was to investigate and develop evidence concerning the effect of alternative decision criteria on the evaluation of items. The investigation included identi- fication of the effect of the use of seven alternative decision criteria, the effect of the criteria on the amount of space optimally allocated to individual items, and preliminary evid— ence on the effect of item selection on the operating results of a chain. Also included was a set of experiments designed to test the sensitivity of item rankings to variations in the input data. The results of the experiments are presented in the following chapter. CHAPTER V FINDINGS. Introduction The results of the experiments with the computer rou- tines are discussed in the following paragraphs. The dis- cussion is centered on the hypotheses presented earlier and presents the results obtained from each of the experiments. Tables are presented in the text and in the Appendix to facilitate the discussion. Findings Relative to Hypothesis I The first hypothesis H01: The ranking of each item in a set of . items will not change when the criterion used for the ranking is changed. was investigated through the use of the BUYSIM program. Each item in four product families was ranked according to seven criteria. The criteria used are listed in Table H-5 and the rankings appear in Appendix G. Examination of the item rankings presented in Appendix G shows that the rank of items does change when the criteria used for the evaluation are changed. One measure of the ex— tent of the effect of the variation of the criteria on the ranking of items is the number of times each pair of criteria agree on the rank assigned to the items. While the measure does not consider the variation in rakings, but rather consid- ers only the fact of agreement or non—agreement, the measure 11O 111 does provide for some useful insight into the criteria. Table 5-1 presents data on the number of times each possible pair of criteria agreed on the ranking of all fifty—two items. For example, the first pair of criteria, movement and dollar sales, gave the same rank to thirteen of the fifty-two items. TABLE 5—1: The 21 possible combinations of the seven decision criteria and the number of times each pair agreed on the rankings of the 52 items number number pair of times Movement and Dollar Sales Movement and Gross Margin Per Cent Movement and Gross Margin Dollars Movement and Dollar Contribution Movement and Net Profit Movement and Summary Dollar Sales and Gross Margin Per Cent Dollar Sales and Gross Margin Dollars Dollar Sales and Dollar Contribution Dollar Sales and Net Profit . Dollar Sales and Summary Gross Margin Per Cent and Gross Margin Dollars Gross Margin Per Cent and Dollar Contribution Gross Margin Per Cent and Net Profit Gross Margin Per Cent and Summary Gross Margin Dollars and Dollar Contribution Gross Margin Dollars and Net Profit Gross Margin Dollars and Summary Dollar Contribution and Net Profit Dollar Contribution and Summary Net Profit and Summary _3 _\ LUUU _s_s._s N—‘meflomrwmd VVVVVVVVVVVV 4—4—8 ..5 Oan—F’ Lu VVV V 4—8—8 0\ [\DWUIODOUI W000“ fl O\—*l\)\OO—‘—‘l\)fl “DNA—8.4 40003“ VVVVV .05 Average Number of Agreements The data presented in Table 5-1 show that the number of times a pair of criteria agreed on the ranking of the items is low. The highest number of agreements was eighteen as scored 112 by gross margin dollars and the summary with dollar contribu- tion and net profit second with fifteen agreements. Two of the pairs, dollar sales with gross margin per cent and gross margin dollars with net profit, scored zero agreements. The average number of time a pair agreed on the rankings of items was 6.05 out of a possible score of fifty-two. An overall estmiate of the level of agreement among the 1 criteria is provide by Kendall's coefficient of concordance. The statistic W = observed variance of the rank sums maximum possible variance of the rank sums was computed for each of the four product groups. The values of the statistic, and the average value, are presented in Table 5-2 0 TABLE 5-2: Kendall's coefficient of con- cordance for four product families product value of the family statistic Canned Ration 0.26h735 All Meat 0.243776 Semi Moist 0.0H21OO Dry Meal 0.106883 Average 0.16 37 The values of the Kendall statistic are not subject to a test of significance by comparison with table values as would be the case with a statistic such as Chi Square. However, some interpretation is possible. The statistic, by definition, can assume values between zero and one. If there is no'agreement at all among the criteria the value of the 1Wm. L. Hays, Statistics for Ps cholo ists (New York: Holt, Rinehart and Winston, 1963) pp. 656—65 113 statistic would be zero and if the criteria agreed completely the value of the statistic would be one. The low values in the Table indicate, therefore, a low level of agreement or "concordance". Although the agreement or non—agreement between pairs of criteria is interesting and useful, a better understanding of the alternative criteria requires analysis of the amount of variation in the rankings. To analyze the amount of var- iation in the rankings of the items by the various criteria a computer program was written to compute the sum of the absolute differences in the rankings of the items according to the alternative criteria. For example, the first item scored a rank of five according to movement and one accord- ing to gross dollar sales. The absolute difference in the ranks is four. The second item scored a nine according to movement and a nine according to gross dollar sales for an absolute difference of zero. The sum of the two absolute differences is then four. The absolute difference in the rank is computed for all items and for all pairs of criteria. The sum is computed by adding the absolute difference in the rank given each item by each pair of criteria. The reshlts of the calculation of the sums of the absolute differences in the ranks of the items are presented in Table 5—3. The results can be used to identify the degree of similarity in the rankings of the items. Similarity, for the present research, is defined in terms of the sum of the absolute difference in the rankings. 111+ om.om a: mm mm o: , saoasam one oaaoao cmz Arm om.mm om am a: om sasaaam one soapaoaaosoo asaaoo low 00.mm :e 1m 0m 0m pflmonm poz can Coaudnampqoo pmaaoc Ame 00.mm :m me em we mamaadm cam mmmaaom sampmz mmono Awe oo.mo. om mm as mm second ooz eds maoaaoo cameo: mmoso Ase 00.mm on On 00 mm Coauznappnoo pwaaoo cam mmmaaoc Cameo: mmomo Ame mm.am em mm am on sasaaam oam ammo soc mamas: moose Ame ms.sm em as am am pauoam poz one odmo soc mamas: mmoao Ase mm.mm a: mm mm mm QOHpSQHHpcoo Hmaaoo cam pcoo pom cflwmmz mmono Ame mm.0: mm mm Pm mm mnmaaoo camps: mmOHo cum pcmo pom Camps: mmOpo Ame oo.ma mm mm am am mamasom out moaom amaaoo Ace 00.5m :0 co we we paconm poz cam moamm Hmaaoo A0? 00.05 cm mo mm 1: noapznfiapcoo mmaaoo cum moamm Hmaaoa Am 0m.mm 0: we mm :m mnmaaoo Gamma: mmopo cam moamm Hmaaom Aw mm.:© mm mm Fm mm pcoo mom Cameo: mmOHo cum moamm amaaoc Am ms.om em of mm on sumESSm cam pco8o>oz A0 ms.es mo mm mm mo caeoam ooz one osoaoeoz Am mm.mo as a: om om soacooaaosoo amaaoa cam aaoaoeoz A: mm.m: mu 0m mm mm mnwaaoo Gawnmz mmonu cam pcoao>oz Am 00.0w mm P: mm mm ammo mom cameo: mmono cum pcmao>oz Am mm.mm mm mm oe mm moamm aoaaoo can esososoz Ar mowaaawm Home pmaoa pmoa coapmn Rama mo coapnanommc armada noncona J mac uaaom Ham cmcdmo Hams mo owmno>m menopano nowmaomc mo name manammoo memo an sawawm podconm m SH ampa some ao>am mama esp ca mesonommflc opzaomnm mnp mo 85m one "Mum Mame“ 115 The lower the sum the greater the similarity attributed to the pair of criteria. According to the data presented in Table 5-3 the pair of criteria showing the greatestsimilarity is "dollar contribution and net profit" with an average sum of absolute differences over the four product categories of 22.00. The pair of criteria showing the least similarity is ”movement and net profit” with an average sum of absolute differences Of 71.75. The results of the analysis of the sum of the absolute differences in rankings agree with the results of the analysis of the number of times each pair of criteria agreed on the rank— ing of items. In both cases dollar contribution and net profit are among the most similar and in both cases movement and net profit are among the least similar. The findings show the effect of alternative criteria on the evaluation of an item. Further, the findings indicate the amount of similarity between pairs of criteria.~ Findings Relative to Hypothesis II The second technique employed in the research to identify the effect of alternative decision criteria is based on the concept that the selection of the decision criteria depends on the goals that management sets for a firm. The maximiza- tion feature of linear programming was used to test the effect of alternative management goals, on the allocation of shelf space to items. The experiment was designed to provide evidence to test the hypothesis that: H02: The per cent of total available shelf space 116 allocated to individual items by a linear . programming allocation routine will not vary when the objective function is changed from one to another of the following manage— ment goals: a) Maximize unit sales b) Maximize dollar sales c) Maximize gross margin per cent d) Maximize gross margin dollars e) Maximize dollar contribution The results of the experiment are presented in Table 5-4. In the Table the values which are underlined indicate values which are significantly different from the value in the preceding column. Thus, the values represent significant differences in the per cent of shelf space allocated to items as the basis of allocation was changed. For example, when the maximization goal was changed from unit sales to dollar sales the space allocated shifted only 0.05 per cent which is in- significant. However, when the maximization goal was changed from dollar sales to gross margin per cent the space allocated changed significantly from 1.h2 per cent to 0.3% per cent. Table 5-5 presents a summary of the changes indicating the pairs of criteria which produce significantly different shelf space allocations. As indicated in the Table, shifting from one to another of six pairs of maximization goals caused a significant change in the per cent of space allocated. How- ever, for four of the pairs a shift from one to another produced no significant change in the per cent of shelf space allocated to the item. 117 TABLE 5—h: The per cent of shelf space allocated to items by a linear programming allocation of shelf space using five maximization objectives maiiElEaii2r_cbiaariias_______ m c: c: 5:: (D H H O m e1 no on -a a) (U H F-1 +-> 73 m as a o B U) F—c G) H Fin-1 E 89 as as item :31 .-1 o F—c 01-1 .-1 1:: description 5 8 :3 g :3 8 8 8 030h1 KLRAT DG FOOD 0.52 0.5% 0.51 0.53 0.52 03051 KLRAT STEW PK 1.37 1.h2 0.3% 0.35 0.35 03061 KLRAT DG FOOD 1.h8 1.55 0.36 1.18 0.38 05031 VET DG FD LIVR 0.52 1. 6 1.01 1.93 1.92 050h1 VET DG FD CHKN 1.88 1.96 1.86 1.95 1.92 05051 VET DG FD 11b 0. 2 0.5% 1.86 0.53 0.52 05061 VET DG FD 1-10 1. 8 1.55 . 6 1.5 1.51 070h1 RIVAL DG FD 0. 2 0.5% 1. 6 1.95 1.92 07051 RIVAL DG FD BEEF 1. 8 1.55 1.E7 1.5a 1.51 08011 DASH DOG FOOD 2.20 2.29 2.17 2.28 2.2M 11011 STREAK DG FOOD 0.52 0.5% 0.51 0.53 1.92 12011 STRONGHEART DG FD 0.52 0.5M 0.51 0.53 1.92 07012 RIVAL BURG + GVY 1.88 1.96 0.51 1.95 1.92 07022 RIVAL MXD GRILLE 1.88 1.96 0.51 1.95 1.92 07032 RIVAL CKN CROQ 1.88 1.96 0. 1 1.95 0.52 09012 ALPO LAMB 1.88 1.96 1. 1.95 1.92 09022 ALPO SCRAMBLE 1.88 1.96 1.86 1.95 1.92 09032 ALPO MTBL + GVY 1.88 0.5% 1.86 0.53 1.92 090u2 ALPO HORSMT DG FD 0.52 0.5 1.86 0.53 1.92 09052 ALPO LIVER DG FD 0.52 0.54 1.86 0.53 1.92 09062 ALPO CNK BF 0.52 0.5% 1.86 0.53 0.52 09072 ALPO CHICKEN 0.52 0.5% 1.86 0. 1.92 09082 ALPO CHPD BF 1.88 1.92 0. 1 .1.35 0.52 10012 KAL KAN CHKN 1.88 1.96 1. 6 1.95 1.92 10022 KAL KAN CNK BF 1.88 0.5% 1.86 1.95 1.92 10032 KAL KAN STEw 1.88 1.96 1.86 0.53 0.52 02033 GAINES PRIME 2 2.99 3.13 2.96 3.10 1.69 020h3 GAINES PRIME A 5.39 3.13 2.96 3.10 3.06 02053 PRIME VARIETY 2.43 2.53 2.39 2.51 2.H8 02063 GAINES BURG 72 2.78 2.90 2.75 2.88 2.8M 02093 GAINES BURG 36 1.85 1.93 1.83 1.92 1.89 2103 G BURG LIVER 1.85 1.93 1.83 1.92 1.89 02113 G BURG CHKN 1.85 1.93 1.83 1.92 1 89 03013 KLRAT SPEC CUTS 2.79 2.91 2.76 2.90 0.82 03033 KLRAT BURG 36 3.68 3.80 3.59 3.76 3 72 03023 KLRAT BURG 72 2.57 2.68 2.5% 2.65 2.61 06013 TOP CHOICE 36 3.1% 3.27 3.10 3.75 3.20 118 TABLE 5-4: continued maximization opiectives m G G Q m -H -H o m H b0 b0 H (D (U F-c F4 4-) r1 co m4: m S 0 21C 2:m ,0 U) H (D F-q 54-1-1 cu 020 U) to CU F-o item :1 :1 8 :4 8: :45; description :15: 8 (:3 0‘3 538 88 06023 TOP CHOICE 72 2.99 3.13 2.96 3.10 3.06 01014 PUR PUP CHOW 2.06 2.16 2.04 2.14 2.11 01024 PUR DOG CHOW 2 2.90 3.02 2.86 3.00 2.96 01034 PUR DOG CHOW 5 2.28 2.38 2.26 2.36 2.33 01044 PUR DOG CHOW 10 2.28 2.38 2.26 2.36 2.33. 01054 PUR DOG CHOW 25 2.73 2.85 2.71 2.83 2.80 02014 GAINES DG BITS 1.48 1.55 1.47 1.54 1.51 02024 GAINES DOG BISC 1.48 1.55 1.47 1.54 1.51 02074 GAINES MEAL 1.82 1.90 1.80 1.89 1.86 02084 GAINES DOG FOOD 2.73 2.85 2.71 2.83 2.80 04014 GRAVY TRAIN 5 2.28 2.38 2.26 2.36 2.33 04024 GRAVY TRAIN 10 2.28 2.38 2.26 2.36 2.33 04034 GRAVY TRAIN 25 2.73 2.85 2.71 2.83 2.80 05014 VETS DG FD 5 2.28 2.38 2.26 2.36 2.33 05024 VETS DG FD 25 2.73 2.85 2.71 2.83 2.80 Note: Minor variations in the per cent of the shelf space allocated to an item by the various criteria should be ignored as they result from minor differences in the total space occupied by the department. The values in the Table are expressed as the per cent of linear shelf space available in the department. Findings Relative to Hypothesis III The results of the experiments conducted to test the third hypothesis H03: Sensitivity analysis will show that an item's ranking by the BUYSIM routine will not change when the item characteristics of price and handling cost are varied. are presented in Table 5-6. The Table presents the rankings 119 TABLE 5-5: The alternative pairs of maximization objectives and whether or not each pair indicated signif- icant changes in the space allocated to items* gross gross gross uniti dollar margin margin dollar sales sales per cent dollars contribution ggigs X NO YES NO YES 8:32: d°llar x YES NO YES gross margin per cent X YES NO gross margin dgllars X YES dollar X contribution * "YES" indicates that the per cent of space allocated to items changed significantly when the maximization objective was changed from oneip the other of the pair. "NO" indicates that the per cent of space allocated to items did not change when the maximization objective was changed. of a selected item according to the seven criteria as the input coefficients are changed by five, ten and fifteen per cent. The data presented in Table 5-6 indicates conflicting results. When the price of an item is altered by as little as five per cent the ranking of the item changes significantly. However, when the handling costs of the item are altered very little change in the ranking of the item results. Part of the explanation is that a change in the price affects all the other factors except movement. However, a change in the handling costs changes only the dollar contribution and the net profit. Even for dollar contribution and net profit, the ranking of the item was affected more by the variations in the price of the item than by variations in the handling cost .3oa op swan some coxcmu ohm cocawpno msnv m03aw> one .manopano Nam pmnam map Ham mmonow muaaadm cam mac mafia macaw wo gonads map acne moanopano some on wca ncnooom aopa 20mm mo scan on» wnflpomnanm an copmadoamo ma sumaadm copnmao3 mna** .coom moo pmoa Ham mom wwmc can mo mnwamo m gonads one cam coom woo Coapmm cchwo_nom meme map mopmq ammo a no ads mam * 120 s o c or w a me m m a me m me m me.e e o o m m s we m m a m_ m me m oe.c s o o a m b we m m a me m me m mo.e s. m o m m a me m. m a me m me m 00.? s o m s m a me m m a m_ m m_ m mm.o s o m s m m N, m m a me m me m om.o s o m s m m N, m m a me m me m mm.o omoo maaaoaom Hoaamaao mo ammo Mom m e a a a a s a m a me m me m me._ m e m e a e w P a a me m me m oe.e s a m a a a or e m m we m me m mo.e u o o w m a ma m m a me m m_ m oo.H a o o oe m or N, m m_ 6. me a me m no.0 me o s Fe. m er a, 6 ac or me m me m oa.o me me s S or me i me i: me T m 9 m mmd m e m e m a m e m e m a *m *e moans #003 you #003 you sow: Hod ammo moo moamw Hanamano **mnwaadm copmmoqom Goapdn mnmaaoc Gamnma umaaoc pcmam>oa mo copnmaoz pacona nanpcoo Campos mmonm mmoew pcoo pom pod Hmaaoc mmonm coanm> mam mpnoaoammooo panda ozp non3.manopano Go>om on mcficnooow coom moo comp umma Ham cam noapwn cocqmo cocooaom m mo mcfixcmn one "01m mam030m .poxpma pfiompoo map CH Hm>oa ooflnm 30H zaawcoapmooxo one mo coapoCdm m ohm mpamomm pom o>apmmoc one umpoz * mmxde mmommm Mbobm0| omoOBMI emoMOJl mmorm5l ooomm©| wooomwl ®Nobbml HHmomQ Hmz mmmzmmxm 90mm se.m0: mm.mmm mm.mmm 0o.m:: mo.m_: .ms.0m: um.m0: IHQzH cmeeooqqe . aHaomo oza name 1mm>0 09 onebm om.mmm- ms.am- mo.ao- mm.mom- aa.eam- em.aom- ma._se- -Hmazoo Hommam me.mmo am.mm: oo.ssa mm.ams mm.oso mm.oos aw.mom memoo ozHamzam om.osm mm.smm mm.mmm om.ema aa.am: eo.eam ma.mam zHomaz mmomo oqom me.emom :m.0umm 0m.w00m 0m.mm0: Nn.0omm 10.0FF: 0o.:oom moooo m0 Hmoo mo.emo: om.ummm mm.mamm om.msaa be.omea mo.soma ms.wmoa macaw mmomo mnmaadm pwmona Goapsn mamaaoc pcoo moawm pnoao>oa cmpnwao3 pom. Jaupqoo Cameos moo pmaaoc mmaaoc mmomw cameos mmOpw mmonm menopano *mwnopano Coamaooc Go>om on mnacnooom copooammw mucoapnmaoc mom spH>Hpow copmasaam Mo mooanma Goopuanp co me coanmm How upcoaopmpm mafipmumao "s1m mamas szzH9 umpmymCmm pawamumpm mcflummmao mnp mo oaaamxm C< >zoawpm qutu»<_w wznha , ...1 iii- .;-..- I mo.v¢ , ::I.. nu--lsinmo.vv onhazomm cz< az~m.hxw>ale. : -;;:n oo.n~ ocans mp<_mm 4HF ea aouxwm 121+ Summary Chapter Five has presented and briefly discussed the results of the experiments conducted using the simulation programs. The first experiment used the_BUYSIM program to rank pet food items according to seven alternative criteria. The rankings and the analysis of the rankings indicate that the alternative criteria do have a significant effect on the evaluation or ranking of an item. The results of the rankings of the items were then sub- mitted as input to the CHAINSIM routine to determine the effect of the different rankings resulting from the use of the various criteria on the operating results of the chain. Again, the results varied indicating that different criteria which result in the selection of different items will cause variation in the profit accruing to the chain. , Based on the concept that the goals of management and the decision criteria employed by a chain are interrelated, the affect of alternative management goals on shelf space alloca- tion was investigated. The results showed clearly that manage- ment goals, and therefore the decision criteria used, do af- fect the emphasis placed on various items. The final set of experiments was designed to evaluate the sensitivity of the ranking of an item to variation in the price and handling cost of the item. The results showed that the ranking of an item was extremely sensitive to variations in the price. However, minor variations in the handling cost of the item had little or no effect. 125 The conclusions and implications of the results presented above are discussed in Chapter Six. In addition, Suggestions for further research, based on the present project, are presented. CHAPTER VI CONCLUSIONS AND IMPLICATIONS Introduction The final chapter of the dissertation is subdivided into three sections. The first section presents the con- clusions on the hypotheses that result from the findings presented in Chapter Five. The second section of the Chapter discusses the implications of the research. The particular elements of the product mix decision are dis- cussed at length. The last section of the Chapter presents some suggestions for further research based on the present project. Conclusions Relative to Hypotheses . The first hypothesis was: H01: The ranking of each item in a set of items will not change when the criterion used for the ranking is changed. The results of the ranking of the items in four product families presented in Appendix G and the data presented in Table 5-1 provide sufficient evidence to reject the hypothesis. As indicated in Table 5-1, the highest number of items ranked the same by any pair of criteria was eighteen. Thus, thirty- four items were given different ranks by the pair of criteria which showed the highest level of agreement. The criteria established for the rejection of the hypothesis was that if more than three items were assigned different ranks by each 126 127 pair of criteria the hypothesis would be rejected. Since the fewest number of differences was thirty-four the hypothesis is rejected and the alternative hypothesis HA1: The ranking of each item in a set of items will, in general, change when the criterion used for the ranking is changed is accepted. The results of the allocation of available shelf space using the SPACALLO linear programming routine were used to test the hypothesis that H02: The per cent of total available shelf space allocated to individual items by a linear program allocation routine will not vary when the objective function is changed from one to another of the fol- lowing criteria: a) Maximize unit sales b) Maximize dollar sales c) Maximize gross margin per cent d) Maximize gross margin dollars e) Maximize dollar contribution. The data used to test the hypothesis are presented in Table 5-5 and summarized in Table 5-6. The hypothesis is rejected when the ranking of more than ten items changed significantly from one criterion to another. Table 6—1 lists the possible pairs of criteria and indicates whether or not the hypothesis can be rejected for the pair. As in- dicated in Table 6-1 the hypothesis can be rejected for six pairs of criteria but is accepted for four pairs of criteria. The conclusion is then, that some pairs of criteria give similar rankings to items while other pairs of criteria give different rankings to items. 128 TABLE 6-1: Summary of the results of the tests of Hypothesis II: The pairs of maximization goals and accept- ance or rejection of the hypothesis. pair . . r number description of pair acgzggggggno 1 Unit sales and dollar sales Accept 2 Unit sales and gross margin per cent Reject 3 Unit sales and gross margin dollars Accept M Unit sales and dollar contribution Reject 5 Dollar sales and gross margin per Cent Reject 6 Dollar sales and gross marin dollars Accept 7 Dollar sales and dollar contribution Reject 8 Gross margin per cent and gross Reject margin dollars 9 Gross margin per cent and dollar Accept contribution 1O Gross margin dollars and dollar Reject contribution The third hypotheSis was subdivided into four sub- hypotheses. The four are: HO3-a‘ The ranking of an item by the BUYSIM routine will not change when the price of the item is increased by five, ten, and fifteen per cent. HO3-b: The ranking of an item by the BUYSIM routine will not change when the handling costs of the item are in- creased by five, ten, and fifteen per cent. HO3-c‘ The ranking of an item by the BUYSIM routine will not change when the price of the items is decreased by five, ten, and fifteen per cent. H03_d: The ranking of an item by the BUYSIM routine will not change when the handling costs of the item are de- creased by five, ten, and fifteen per cent. The data presented in Table 5—7 was used to test the sub- hypotheses. Based on the criteria established in Chapter 129 Four, the two subhypotheses related to variations in the price of an item are rejected. However, the two subhypotheses related to variations in the handling costs are accepted. General Conclusions In addition to the specific conclusions regarding the hypotheses presented above, several general conclusions may be drawn from the present research. The present research has focused on the evaluation of the alternative criteria for product addition and deletion decisions. The project has investigated the decision process in the literature, the industry, and through experiments with computer simulations. Based on the investigation conclusions can be drawn 00 the sources of the information, the location of the decision, and the criteria to be used. The most useful source of product mix decision infor- mation are the records of the chain. In particular, detailed records of the sales, costs, and profits associated with the sale of each item provide the best basis for item decisions. For new items the records of the chain can be used to provide data on comparable items. For the new item, the source of the information must be the new item form completed by the manu- facturer. The new item form is presently used but can be improved by providing more specific detailed information on the promotional programs of the manufacturer. The research has identified three possible locations of the product decision within the structure of the firm; namely 1) the buyer only, 2) all decisions made by the buying 130 committee, and 3) a combination of the buyer and the buying committee making the decision. When the three alternative formats are evaluated according to the criteria of efficiency and control, the third format, a combination of the buyer and the buying commitee, appears to be the most useful. The alternative criteria available to the chain for product mix decisions fall into two categories - quantitative and qualitative or subjective. The most appropriate quanti- tative decision criteria appears to be the net profit per unit time generated by the item, assuming that the goal of the firm is the maximization of net profit. The second most useful criteria appears to be the direct dollar contribution of the item to the chain. The subjective criteria used are "newness" for new items and the role of the item in the mix of items carried by the chain for other items. Both of the subjective criteria lack research support for their use but appear to be highly thought of by supermarket executives. The use of such criteria can only be for items that would be rejected using the quantitative criteria and even then should be donewith care. A final general conclusion is that supermarket chain management could make a great deal more use of computer technology in the decision process. The present research has suggested the use of a computer to perform the item evaluation tasks generally performed by the buyers and the buying commit- tee. Once the computer has made the evaluation the buyer and/ or the buying committee could use the evaluations to make the product decision. By removing the computation from the hands 131 of the buyer, such a system would reduce the subjectivity in the evaluation and also provide more time for creative manage- ment. Major Conclusions The most important conclusion of the present research is that the item addition and deletion decisions of super- market chains are often hastily made using inappropriate, incomplete, and sometimes inaccurate information. The deci- sions regarding the mix of items carried by the chain are among the most important decisions a chain makes. Yet the decisions are often made in two or three minutes and at least one buyer has said that, "On practically all products I can ."1 The results decide which way I feel within 30 seconds.. of the present research have shown that the criteria used by chains may not lead to the greatest profit for the chain. The research has further shown why and how additional information can be used. The overall decision process is illustrated in Figure 6-1. In the Figure the inputs to the decision process, the criteria used, and the bases for performance evaluation are presented. The sections following Figure 6-1 present detailed con- clusions concerning the elements of the decision process based on the present research, the literature, and industry data. 1A comment of a buyer for a large chain as reported in Neil H. Borden, Jr., Acceptance of New Food Products by Super- markets (Boston: Division of Research, Harvard Business. School, 1968) p. 20% 132 oneaapamm m.mmmpaozH amoezm>zH zo zmmamm tz axe 2H zmHH mug mo mgom zmeH mma mo mmmzzmz wquHm<4H<>< moaam spfiafinmaamm m.amfiaaasm I muod©0pm m_pmflamasm I anommpmo podooam I mHHz mozmHmmmxm mpmoo wGHHUCmm I me0 mod Campos mmoau Hampom ammo pmoo mmmo on8o>oz maosaomm ozHHmmz©o mmHOpm I omzonmmm3 I mHZHHHHHmmzoo cofiummsm cam QOH I mqo< Hmooq I mcfimfipmm>©< HmQOHpmz I zfiamo I. modo ommo ammo 9mm :HmHmZ mmomo pmoo mmmo. pmoo mmmo xomm mmmo I moHHmHmmeooz Fl IIIIIL mmm moaam est mam zmpemm azmzamm>zH fiAI . amoezm>zH zQ zmpemm zHoz mmm wmdflfipcoo "file mmDpo 13h Elements of the Product Mix:Decision Types of information required The specific items of information required for decision making in a chain depends on the criteria employed in the deci— sion process. Whether the item is a new item or one currently stocked is also important in determining the information needed. Table 6-2 is a summary of the data needed for the evaluation of items using the criteria now employed by chains. TABLE 6-2: The data needed for product mix decisions new items items currently stocked Movement of comparable items* Unit cost of the item Test market data on the item* Unit retail price of the Data on the introductory pro- item gram of the supplier* Unit sales of the item by Data on the natioal advertising day or week program of the supplier* Unit handling cost of the Data on the local advertising item A program of the supplier* Unit overhead expenses Data on the reactions of our of the item competitors to the item* Unit cost of the item Unit retail price of the item Unit handling cost of the item Unit overhead expenses of the item * These data are required to provide an estimate of the rate of unit sales of the new item. It is possible to estimate unit sales even though a chain does not have all of these data, but the more accurate data available the more accurate will be the estimate of the movement of the new item. The quality of the data indicated above could be improved considerably. Test market data should be documented, perhaps by a cooperative agency developed by the chains. Further, test market data should be translated into a chains local mar- keting area. For example, data on a test market conducted in 135 Syracuse, New York must be translated to the Los Angeles area before the results of the test market would be meaningful to a chain operating in the Los Angeles area. Another area in which the quality of the information could be improved is the data on the promotional programs of the suppliers. At present the data is presented in the form of schedules of advertisements in newspapers and magazines and the schedules and number of shows or spots on television and radio. While interesting, such information is not really useful. The important facts about an advertising campaign are not the individual parts that make-up the campaign but rather the total impact of the campaign. Such data as the range, frequency, and gross rating points of the campaign would be much more valuable than the fact that the supplier was going to run a series of nine advertisements in four consumer mag- azines. Sources of the information required The data on the movement of the comparable items can be extracted from the records of the chain after the buyer has decided which items will compete with the new item to be eval- uated. The data on the handling costs and the indirect or overhead expenses should be available from a table that has been previously developed by the chain. For each type of item or each set of item characteristics a chain should develop standard overhead costs. Such costs would then be available when a new item was being evaluated. 136 All the data needed for the evaluation of items now stocked by the chain are available from the records of the chain. The data other than the handling costs and the esti- mated indirect expenses would be available from the buyers' cards. The data on handling costs and the indirect expen- ses should be available from the tables mentioned above. The structure of the decisiongprocess As presented in Chapter III of the dissertation, chains follow oneof three alternative formats for the decision pro— cess. The three formats are: 1) The buyer makes all product addition and deletion decisions. 2) The buyer evaluates the items but makes no decision. Rather, all decisions are made by a buying committee. 3) The buyer evaluates all items and makes a decision to reject all items clearly unacceptable to the chain. All accep- tance and deletion decisions are.made by the buying committee. Based on the material in Chapter III the conclusion of the present research is that the third format would be best for most chains. The selection is based on the results of the evaluation of the control and efficiency of the decision pro- cess under the three formats. A buyer makes all decisions The first format is quite efficient since only one per- son is involved in the decision process. Further, the format provides for a high degree of flexibility and the decision maker has the ability to respand quickly when necessary. How- ever, there is little provision for control in the decision 137 process. The experience and ability of only one person, the buyer, is brought to bear on the decision. No matter how good the buyer is, more experience and ability can be brought to bear on the decision by a committee. Further, a single per- son is more likely to be affected by emotional reactions to a product and/or a salesman and less likely to make rational decisions than is a group of experienced supermarket personnel. A committee makes all the decisions The second format, where all item decisions are made by a committee, sacrifices efficiency for what could be a great deal more control. Unfortunately, the control aspect is often missing and the procedure introduces delay into the decision process. The lack of efficiency stems from the fact that all items must be reviewed by the committee. There.are a number of items offered to a chain that could be rejected by the buyer using a preset group of decision rules. Screening out such items before presentation to the buying committee would greatly reduce the number of decisions that the committee has to make. If the process of evaluation and decision on all items by the committee resulted in a very high degree of control or on extremely good decisions, the lack of efficiency could be justified. Unfortunately, the quality of the decisions may, in some instances, be decreased by the requirement that all items must be reviewed by the committee. The number of new items and items suggested for deletion and must be evaluated each week by a chain is large. Since the buying committee is , ‘ ll] ill 138 in session for only a limited amount of time each week, the amount of time spend on each decision is partially a function of the number of items to be evaluated. If the number of decisions were reduced, each decision could receive more time. A combination decision process The third decision format provides for efficiency by having the buyer screen out such items as are clearly unac- ceptable to the chain. The buyer only presents to the committee those items he feels are acceptable or about which he is in doubt. The committee then decides the fate of the items. Both the control and the efficiency of the thkd format can be improved through the use of an item evaluation sheet which would be duplicated and distributed to all members of the committee prior to the meeting of the committee. Such a sheet would force the buyer to collect all relevant data and summar- ize his position which would saved meeting time. The sheet would contain a record of all items evaluated by the buyer, the information used in the evaluation, and the results of the evaluation. For those items rejected, the buyer would present his reasons. If the other members of the commit- tee disagreed with the buyer or for some other reason wanted the item discussed at the meeting, that item could be presented at the meeting. For those items not rejected by the buyer, the other members of the committee could study the data presented and formulate the questions or points to be discussed at the 139 meeting. Not only would such a procedure result in more ef— ficient use of meeting time, bUt more importantly, the items discussed would receive a closer scrutiny. The use of a pre-meeting fact sheet on items to be discussed will be of value only if all members of the com- mittee seriously consider the information on the sheet prior to the meeting. Unfortunately, in some chains where such an information sheet is now used, the membersof the committee do not always examine the sheet prior to the meeting. In such cases not only is any possible benefit of the sheet lost, but the cost involved in reproducing and distributing the sheet is wasted. Objective criteria for product mixidecisions There are three distinct sets of criteria for product mix decisions that are often used by chain management. The first set consists of individual pieces of information which can be used to evaluate items. Included in the set are: 1) Movement (For new items the value would be an estimate of movement based on the movement of comparable items, the intro- ductory programs of the supplier, and the promotional programs which will support the items.) 2) Dollar sales per period of time 3) Unit gross margin dollars 5) Gross margin dollars generated per unit time 6) Direct dollar contribution per unit time. 7) Unit dollar contribution 8) Net profit dollars generated per unit time 9) Unit net profit dollars 1)+O Without knowledge of the objective of management, a statement of the best decision criteria is impossible. Under the assumption that the objective of the chain is the maximiza- tion of net profit, the best criterion is item net profit per unit time. Net profit is calculated by subtracting the direct expenses of carrying an item and a share of the indirect expenses allocated to the item on some basis selected by man- agement from the gross margin dollars generated by the item. Since chains must select among alternative items to decide in which to invest limited shelf space and capital, the selec- tion of items using net profit will maximize the net profit of the chain. Were a chain to have an objective other than maximum net profit, the decision criteria would have to be modified so as to agree with the objective. A few general statements regarding decision criteria are possible however. First, any criteria used should pro- vide for a measure of both rate of sale and return to the chain. Unless a chain has as its objective to establish and maintain the highest rate of unit sales possible, the criterion of movement would not be appropriate. Movement, in and of itself, does not give any indication of the value of an item to a chain. Similarly, a measure of per unit return of an item tells very little about the actual value of an items to a chain. Without some indication of the rate of sales of an item the total return to the chain is not available. Thus, any criterion which presents only a measure of the rate of 141 sales or the per unit profit cannot be an effective criterion for product mix decision making. Measures of performance as criteria The second set of criteria aVailable to supermarket chain managers consists of a group of measures of performance. Measures of performance are essentially indices or ratios developed from the operating results of the chain. The measures of performance that have been suggested for item evaluation are: Return on assets employed Return on inventory investment Stock turnover Direct product profit per unit space Net profit per unit space \ntxunaa VVVVV The first two measures of performance are attempts to apply standard return on investment criteria to item eval- uation. The four indicators of return which would be appro- priate are: 8 Gross dollar sales Gross margin dollars generated Direct profit dollars generated Net profit generated rwme VVVV The selection of the indicator of return to be used by a chain depends on the objectives of the chain. If the objective is the maximization of net profit, then the return which should be used is net profit.: The major problem with the use of return on assets employed or inventory investment is that an accurate measure of assets employed or inventory investment is difficult to develop. For most situations the most appropriate measure of assets employed would be the cost of the store level 1)+2 facilities devoted to the item. The cost of the assets employ- ed then becomes the same as the space cost of the item and the measure becomes equivalent to number 5 - net profit per unit space. The use of return on inventory investment as a measure of the performance has an additional drawback. That is, the indicated level of performance depends not only on the level of performance of the item but also on the level of perform- ance of the chain with respect to inventory control. Thus, were a chain to use return on inventory investment to evaluate items the level of inventory investment used should be either a mimimum level of inventory required to support adequately the item or a "standard" level of inventory based on the rate of sale of the item. The use of either of the two suggested indicators of inventory investment would remove the perform- ance of the chain's inventory control system from the measure or performance. Thus, the measure would more accurately measure the level of performace of the item. Stock turnover is a very useful measure of performance for the inventory control system but is not appropriate for measuring the level of performance of an item. The components of stock turnover are the movement of the item and the average level of inventory. As pointed out above, movement, without a measure of the per unit profit of an item, is not a good criterion. Further, the average level of inventory is not a good measure of the performance of an item since the inventory level is dependent on the inventory control system not the item's performance. 11+3 Direct product profit per unit space and net profit per unit space are the two most useful measures of performance. Both direct product profit and net profit are based on the rate of sales of the item and a measure of the return to the chain. Further, since one of the major constraints on the operations of a chain is the fixed amount of store space available, the use of space in the evaluation of items is quite appropriate. The major weakness of measures of performance for item evaluation is that such measures are relative rather than ab— solute measures. The measures or ratios developed are meaning- ful only when compared with similar measures for other items. Thus, measures of performance are useful for comparing items but are not appropriate forevaluating individual items. Newness as a criterion for product mix decisiong ' In addition to the quantitative criteria discussed above, chains often employ two highly subjective criteria in product mix decisions. The first subjective criterion deals with the concept of the "newness" of the item and is used in the evaluation of new item offers. The use of the criterion is based on the assumption that chains must present an image of being modern and up-to-date to consumers. Thus, an item which was the first of its kind on the market would probably be accepted by many chains, particularly if the item were being supported by a high level of consumer promotion. It would seem that if a chain has as its objective maximization of profits, all items, including new items, IIIII should be evaluated by the same criteria. New items do occupy shelf space that could be used by other items and thus incur the same costs for shelf space. In addition, the new item will incur other costs related to adding the item to the prod- uct line of the chain. Thus, rather than a lower level of net profit, new items probably should be responsible for a higher level of profit. ‘The role of an item ip the mix of products carried by the chain a§ a griteriop ‘ The second important subjective criteria used by chains is the role of an item in the total mix of items stocked by the stores of the chain. The basis for the use of the criter- ion is the assumption that if a customer is not satisfied by the mix of items carried by a chain she will switch to another store. A chain would, therefore, lose the profit of the customer on all items she might purchase. Chains use the above assumption as justification for adding otherwise unac- ceptable items. 8 Unfortunately, thenais little or no research evidence to support or disprove the assumption that a consumer will switch stores if her present store does not carry an item she desires. If a consumer would not switch simply because an item she desired was not available there would seem to be little justification for adding an unprofitable item, even if the item were the only one of its kind. Cost and profit allocation problems and procedure; One of the most difficult problems connected with the evaluation of items for product mix decisions is the effect 1H5 of stocking one item on the sale and resulting profit of other items stocked by the store. The problem is closely related to the decision to stock an item to enhance the attractiveness of the store to consumers. As indicated above, many chains stock items which are deemed necessary to present a broad range of items to consumers. Further, some very slow moving items are stocked because there is a very low level but consistent demand for the items. When evaluating the pro- fit of such items it is necessary to consider not only the profit generated by the item, but also the profit generated by other items sold to the consumers who purchase the slow moving items. One approach to the determination of the indirect profit generated by an item would be to develop a measure of the total profit of the "market basket" of items purchased by consumers who purchase each item. The total profit could then be multiplied by the probability that the consumer would switch stores if the item she desired was not available. The result would be the profit lost from not stocking the item. The profit gained by stocking the item would be the total profit of the "market basket" of goods purchased less the profit generated if the item were not stocked. Included in the measure must be some provision for the fact that some consumers would accept a substitute item rather than switch stores. The development pf item handling costs There are two difficult problems associated with the development of the direct product profit per item and the net 1M6 profit per item. The first of the problems is the devel- opment of the handling costs of items. McKinsey and Company have demonstrated that it is feasible to develop the dirept costs associated with stocking an item. The development of handling costs is neither simple nor inexpensive. The most accurate approach to the problem is through the use of engin- eering studies conducted by industrial engineers. Through the use of data compiled by time and motion studies, indus- trial engineers can develop accurate measures of the cost of handling individual items. The use of the engineering study approach is quite expensive. However, the results of the re- search presented in the dissertation indicate that extreme accuracy in handling costs is not necessary for effective evaluation of items. Therefore, less expensive estimating procedures can be used to develop estimates of the handling costs of individual items. A second factor which serves to reduce the problem is that the handling costs of many items will be the same. The problem is to identify the factors which cause the handling costs of items to be different.2 Once the factors have been identified, the costs of groups of items can be developed. Among the factors which should be considered are: 1) Case pack 2) Type of container 2For a discussion of the approach see: Frank H. Mossman, Distribution Cost and Revenue Analysi : A new Approach, Bureau of Business and Economic Research, College of Business and Public Service, Michigan State University, East Lansing, Michigan, 1962. 1H7 3) Size of container H) Cubic space occupied by one unit 5) Rate of unit sales of the item For example, the handling costs of most canned soup would be the same since the case pack, size of container, and other factors are the same. Thus, handling costs need to be devel- oped for only one type of soup. The development of indirect expenseg The development of net profit on items requires an estimate of the indirect expenses that should be charged to each item. The estimate can be developed by allocating the total indirect expenses of the chain to each item carried. There are several bases of allocation available to a chain including: 1) Per cent of unit sales 2) Per cent of dollar sales 3) Per cent of available space occupied by the item . Little or no research evidence is available to aid in selecting the allocation factor. However, the most suitable basis of allocation appears to be per cent of available space. An important factor in the cost of selling an item is the cost of the space occupied by the item. Thus, the space occupied by the item appears to be quite appropriate. A second basis that would seem appropriate is the per cent of unit sales. The number of units of a particular item that must be handled has a significant effect on the cost of the item. Further, the per cent of units sold would be much easier to develop and use than the per cent of shelf space occupied by the item. Therefore, unless research should show 148 that the per cent space is a much better basis of allocation than the per cent of unit sales, per cent of unit sales could be used. Implications The general implication of the above conclusions is that in order to improve their profit position chains can and must reevaluate the product mix decision procedure used. Not only must the decision process itself be studied, but the criteria used and the information available to the decision maker must be evaluated. The following paragraphs discuss the elements of the product mix decision process and indicate the implications of the present research on the various elements. Elements of the Product Mix Depision‘ T a d rc f i f r at o e uired The present research has concluded that the decision criteria used by chains often do not lead to the most profit- able selection of items. A prerequisite to the application of improved decision criteria is the availability of improved information. Improvements shoud be made in both the external and internal information a chain has available. The fixterpal lnformation One of the areas in which chains should improve the external information used is the information provided on new items. Specific detailed information should be demanded on the range, frequency, and gross rating points of national and 119 local advertising. The market expansion and share of market data resulting from test markets should be translated into the local market area of a chain. Further, chains should, on a cooperative basis, set up an agency to document test market results. Chains should also become familiar with other types of external data. In particular, chains should have available data on the trends in consumers'purchasing patterns, incomes, and population shifts in age distribution and location. Such information would be most useful if broken into racial, religious, ethnic and socio-economic groupings. The above information is available from several sources. The United States government, through the census and other studies, develops a great deal of information. Further, chains should make more use of the several trade associations, univer- sities, and independent research groups which publish the results of research. For example, Sales Management's Survey of Buying Poper contains detailed information on the expendi- ture patterns of the American consumer. lnternal informatiop In order to provide better internal information for product mix decisions chains should examine the nature and flows of information within the structure of the firm. Chains should first strengthen and improve the existing information. For example, warehouse shipment data should, as soon as feasible, be replaced with sales data generated at store level. Store level data would be more accurate for shorter periods of 150 time, such as a week. Further, the evaluation of such activities as special promotions, end-aisle displays, and advertising would be improved. While the hardware to provide such data is available, it is expensive. Therefore, it may be some time befor store level collection of sales data is a reality. During the interim, chains should have warehouse shipment data by item on a weekly basis for store groups. The stores of a chain should be grouped so that stores serving similar racial, religious, ethnic and social groups are identified and may be treated as a group. Another type of internal information that should be developedby chains relates to the development of measures of profit. The research has shown that the criterion of gross margin does not necesSarily lead to maximum profits. The indication of the research is that dollar contribution and/ or net profit are better criteria. To calculate dollar contribution chains need data on the direct expenses (handling costs) of items. Chains should develop tables of the handling costs of items. The most promising approach to the task is the development of coSts according to item characteristics such as case pack, weight, cub, item size, type of package, shape of package, value of the item, and the rate of sales. The information can be developed through engineering studies, but the research has shown that estimating procedures can be used wherever possible. To develop net profit chains must also have data on the amount of indirect expenses that should be charged to an 151 item. Using the same item characteristics as for direct expen- ses, chains should develop tables of the indirect expenses that should be charged to items. The most important problem in developing such a table is the decision regarding the basis of allocation regarding the proper allocation of expenses to items. Chains should evaluate the alternative bases of alloc- tion available and select the basis most appropriate for their chain. In addition to the above information chains can use data processing to develop useful information for decision making. The application of data processing to product mix decision making are discussed later. The structure of the product mix decision Chains should reevaluate the structure of the decision process within the chain organization. Two criteria should be used in the evaluation. The first, control, while closely related to the criteria used, is also affected by the struc- ture of the decision process. The aspect of control deals with: 1) The ability of the process to screen out new items that should not be stocked. 2) The ability of the process to provide for the addition of those new items that should be stocked. 3) The ability of the process to identify and delete those items that should be deleted from the list of items now carried by the chain. The second criteria is efficiency. There are two factors which should be considered; 1) the ability to react and make decisions when necessary and 2) the cost of the process in terms of both dollars expended and man—hours required. 152 The decision structure should be able to provide for the accurate evaluation of items, reacting as fast as necessary with aslittle cost and manpower expended as possible. The research concluded that a combination of the buyer making reject - non- reject decisions and the buying committee making reject - accept decisions would be appropriate for most chains. However, each chain will have to consider its own situation, including the experience of the personnel available, to determine the structure most appropriate for its particular situation. The criteria used for the roduct mix decis on There is little question that one of the most import- ant factors in the product mix decision is the set of cri- teria used for the decision. The present research has shown that the various criteria available to chains result in the selection of different sets of items. The rate of intro- duction of new items continues to increase so chains will have to make more and more decisions. Thus, the criteria used by the chain will become even more important. Chains must continually evaluate and reevaluate the criteria used for product mix decisions. As the squeeze on shelf space increases and as the expenses of the chains increase, the pressure on profits will increase. To guard its profit position, a chain will have to make better deci- sions regarding which items to stock. Further, the selection of items to promote either actively through advertising or passively through increased shelf space, will become more important. 153 Chains should evaluate the criteria used for both the selection of new items and the deletion of current items should be examined. In addition, chains should also examine the criteria used for the selection of items to promote. The evaluation of the criteria should be based on the goals and objectives of the chain. Once the goals have been specified, the cahins are in a position to select the criteria that 1 will best lead to the attainment of the goals. The Application of Computers to the Decision Process Simulation One of the most important application of computers is based on the ability of researchers to develop models of supermarket shain operations and utilize electronic computers to simulate the activities of the chain. Such simulations have a wide variety of uses. One important use is research into the various factors surrounding the product mix decision. For example, such simulations can be used effectively to test the alternative results of the selection of items using various decision criteria. The simulation can also be used to project the effect of various promotions on the sales and profits of the chain. Simulation can also be used in routine productdecision making. For example, in the evaluation of alternative items, a simulation can be used to evaluate the strength of alter— native product mixes. By simulating the results of using Various possible mixes of items chains can better evaluate alternative configurations of the mix of products offered to 154 the consumer. One step beyond the simulation of the product evaluation process is the simulation of the total product mix decision process. While the identification of the proper decision rules for such a simulation is difficult, the problem can be solved by modeling the decision process used by the chain. Howard and Morgenroth have described the development of such a model of the executive decision process.3 Rather than attempt to develop ideal decision rules, such a model programs the logic of the decision as now accomplished by the decision maker. The model, when built, can be tested by comparing the output of the decision model with the actual decisions made by the executives. Experimentation with the model can then provide for insights into the decision process, the effect of variations in the decision process and could be used to make the actual decisions for the chain. Routine and Exception Reporting One of the most valuable uses of a computer in product mix decision making is the evaluation of items. As indicated by the BUYSIM program developed as part of the present re- search, a computer can be used to evaluate items according to a variety of criteria. Not only would such evaluation provide more information for product selection decisions through evaluation on a set of criteria rather than one criterion, but the use of a computer would also relieve the buyer of the ‘2 3John A. Howard and William M. Morgenroth, "Information I’I‘ocessing Model of Executive Decision," Mana ement Science, )RDlume 1%, No. 7, March, 1968, pp. H16-H28. 155 routine of the evaluation. Thus, the buyer would have more time to devote to the aspects of the decision process that cannot, at present, be computerized. For example, the buyer could devote more time to the investigation of the specific characteristics of the purchasers of individual items and the customer mix of individual stores. A second important use of the computer is the reporting or identification of items that should be considered for deletion. A chain could set up a series of decision rules which a computer could use to evaluate each item. Once a per- iod, probably at the end, the evaluation program would evaluate each item carried by the chain. On an exception basis the computer would identify, based on the preset decision rules, the items which should be considered for deletion. A computer can also be used in other ways to support the chain buyer.1+ For example, in most chains the buyer re- ceives each day a short and expedite report and a list of the items received in the warehouse. When he receives the reports the buyer must compare the lists to determine the items on the short and expedite report that were received in the warehouse and are, therefore, no longer out of stock. A computer could easily compare the two lists and delete from the short and expedite report such items as were received in the warehouse. Further, for the items remaining on the short and expedite report, the computer could list not only the item, but how long “See Appendix A for a comprehensive list of the possible uses of a computer to support the buyer. 156 the item had been out of stock, the date of the next expected shipment, and which, if any, item could be substituted for the out of stock item. The performance of new items once accepted by the chain is of extreme importance. Even items that appear to be excel- lent may fail after a brief initial period. A computer can be used effectively to monitor the performance of new items. For a period of three to six months after a new item is added to the mix of items carried by the chain the computer could evaluate the new item to determine whether the level of per- formance was meeting a minimum acceptable level. On an exception basis the computer could report the items that were not performing as expected. The Product Mix Decision and Retail Information Systems A retail information system encompasses all the informa- tion necessary to plan operate and control a retail business. Included in such a system would be essential operating data such as payroll, accounts receivable, accounts payable, and product control data. In particular, the product control information includes all the information on the flow of items through the chain system. The process starts with the development of a sales forecast for each item. From the sales forecast an order quantity is determined and an order generated. The infor— mation system also monitors the inventory level of each item in the chain warehouse and on the store shelves. The system accounts for all merchandise received as well as the sales of the items via warehouse shipments to the stores. 157 Ideally, the retail information system of a chain would provide the necessary information for planning such activities as promotions. For example, the information system should provide the necessary information for manage- ment to identify the most effective types of promotions. Further, the system should provide data on the most effect- ive mix of items to be promoted. One of the most important segments of the retail information system is the segment devoted to the planning and evaluation of the mis of items carried by the chain. The segment provides the information necessary to make decisions concerning the addition and deletion of items from the mix of products stocked by the chain. Table 6-2 lists the data needed to evaluate items. The data is the input to the system. To thedata must be added the constraints or decision rules under which the chain wishes to operate. The combination of the data and the decision rules is the data which the information system uses to evaluate items. The output of the product mix decision segment of the information system consists of two types of reports. The first type of report is the result of the item evaluation process. Data on a specific set of items is input to the evaluation process, the items are evaluated and the process reports the results of the evaluation. The report could either be in the form of a ranking of the items selected for evaluation or in the form of specific values for a set of criteria. Perhaps the process would be most useful if both the values and the rank were reported. 158 The second type of report consists of an exception report indicating the items now carried by the chain that should be considered for deletion. The report would be generated at periodic intervals, probably to coincide with the fiscal periods of the chain. Included in the report would be not only the identifica- tion of the item but also the information needed to make the deletion decision. One possible set of such information would include: 1) The movement of the item 2) The gross margin per cent of the item 3) The direct profit and net profit generated by the item In addition to the information on the item to be considered for deletion, the report would also contain the same infor- mation on the items that compete with the item. Such infor- mation would provide a basis of comparison to further aid in - making the deletion decision. The primary advantage of the development of a product mix decision information system is the such a system should lead to greater control of the mix of items carried by the chain. The improved control will have several advantages. One of the advantages is that the mix of products carried by the chain will better meet the needs of consumers. For example, one development in the information system will be the inclusion of data on the matching of items to customer charact- eristics. Items which appeal to specific racial, ethnic or religious groups will be identified and stocked only in stores where the customer mix contins persons with the appropriate characteristics. 159 A second advantage will be that chains will better evaluate the items stocked by the stores of the chain with respect to the contribution of the item to the profit of the chain. Items which do not contribute to the profit of the chain will be identified and delected. Such deletion will be particularly important for items which have a relatively- low rate of sales. Both of the above factors contribute to the most important advantage resulting from the development of an effective product mix information system. That advantage is the competitive advantage a chain can achieve over the other chains that do not develop such an information system. Not only will the short run profits of the chain be increased by the elimination of items which do not contribute to profits but more importantly the long run profits of the chain will be increased. The increase will result from-the fact that the mix of items offered by the chain will be superior to the mix of items offered by other chains. Thus, the chain will attract and keep the customers of other chains in addition to Satisfying its present customers better. Integration of information systems for management declsion making During the development of the information system for product mix decision a chain must be aware of the need to have the product mix decision information system be compat- able with the total information system of the chain. -Prior to developing the product mix decision information system a chain should delineate the parameters of the total information 160 system. The primary factor to be considered is the configur— ation of contents of a central data bank which would contain all the basic data needed by the chain. While the product mix decision information system would utilize only a limited amount of the data stored in the data bank, the development of the system must not preclude the compilation and storage of any data necessary for the total information system. In particular, the above requirement implies that the format of the storage of all data be developed. Further, the data necessary for the total system should be stored even though the total information system is not fully developed. Suggestions for Further Research As a result of the present research several areas for further research can be identified. The first such area is the effect of decision criteria on the operating results of the chain. The results of the seven runs of the CHAINSIM program reported in Chapter Five indicate the use of alterna- tive decision criteria can affect the operating results of the chain. The results indicate that additional research to further identify and better delineate the effects of alterna- tive criteria could lead to a better understanding of the alternative criteria. The increased understanding should lead to a more informed selection of the decision criteria that a chain might use. A second area for further research is concerned with the best criteria for the evaluation of the mix of items carried 161 by a chain. The present research has identified the fact that the mix of items carried by the chain has an effect on the profit of the chain. However, the project did not focus on the question of the evaluation of the mix of items to be carried by a chain. In addition to evaluating items on an individual basis, the research pointed out that chains do evaluate, albeit subjectively, the role of individual items in the total mix of items offered to the consumer. Such analysis and evaluation should be removed from the realm of subjective evaluation and placed in the providence of informed, analytical evaluation. The implication of the suggestion is that the cross-elasticities of demand for each item with all other items carried by a chain must be identified. While such a task is probably beyond the immediate ability of researchers, there are methods available to begin the task. A third area in which a great deal of important research could be conducted deals with the identification of the factors which cause a consumer to remain loyal to a store and the factors which cause a consumer to switch stores. The primary justification of many chains for the addition and/or retention of slow moving or unprofitable items is that such items are necessary to present a broad range of products so as to retain consumers. Research'evidence to either support or refute the assumption would be immensely.valuable. Closely related to the third area is the effect of the gain or loss of a consumer on the profit of a chain. If one is to evaluate accurately the consequences of a decision not to stock an item, then one must know the effect on the profit 162 of not stocking the item. The effect On the profit of the chain will be the expected value of the loss-of a customer times the number of customers that would be lost by not stocking the item. The data on the loss of profit from the loss of a customer would be useful also in evaluating the effect of out of stock conditions which might cause a customer to shop elsewhere. Another area for research is the identification of the best method for estimating the sales of a new item. One of the suggestions based on the results of the present re- search was that chains should evaluate new items with the same criteria as items that are currently stocked. In order to do such evaluation chains must be able to identify the factors that are important in estimating the movement of new items. Further, chains_muSt have the knowledge necessary to apply the estimating technique and know how the information should be used. A sixth area for further research is the development of handling cost on items. McKinsey and Company have made a beginning with the development of the handling costs for six- teen items in the dry grocery area. However, the dry grocery section of a supermarket may contain several thousand items. The development of handling costs on all items is a necessary prerequisite to the application of net profit as an accurate criteria for item evaluation. It is unlikely that handling costs will have to be developed for every item in a store since, for many groups of items, the handling costs of all items in the group will 163 be the same. For example, the handling costs of all canned soups are probably the same. Research is needed to identify the items which may be grouped together because of similar handling costs. A seventh area for important research is the determina- tion of the most appropriate basis of allocation of indirect expenses to items. The allocation is necessary to develop the net profit of an item. Yet, as pointed out earlier, no research based information is available to guide chains in the selection of a basis of allocation. A final area for particularly useful research deals with the sensitivity of item evaluations to variations or imperfec— tions in the data. The present project showed that for the items evaluated variations in price were significant in terms of their effect on the ranking of an item while varia- tions in the handling costs of the item did not appear to be as important. Further research into the effect of variations in the data on the evaluation of items could not only aid in the development of better dataIfor item decisions but could also help further isolate the impact of alternative criteria on item rankings. 16% APPENDIX A Applications of Computers in the Food Industry Data Processing Applications Sales invoicing (store shipments) Sales Analysis Accounts Receivable . Inventory Adjustments Analysis Retail "going in gross" Purchase Order Writing Quarterly Velocity Turnover Report Summary of Item Analysis, Inventory and Sales Accounts Payable Financial Statements Retail Bill-out Control Accounting for Retail Stores I—JprLIJ-bS‘UQ “36’ £140 0‘” Delivery Analysis . Comsper Fleet Unit Rated Delivery Payload Driver Performance Backhauls and Inbound Receiving on Company Trucks Budgets and Variance Comparisons macaw Warehouse Performance a. Cost per Case b. Tons and Cases per Man-hour c. Projected Tonnage and Case Movement by Selection Area Payrolls a. Retail - Corporate b. Retail - Independent c. Concentration of Purchases by Retailers - Semi-annual Other Applications a. Net Profit b. Net Profit by Product Item Decision Formulations Where to Store Merchandise in Warehouse Scheduling of Inbound Truck Receiving Weight and Cube on Outbound Deliveries Minimal Economic Level to Stock of an Item Minimal Supplier Case Allowance to Order and Stock Excess Stock When to Discontinue an Item Return on Investment Profit Planning, Volume—Cost Analysis Rebate and Allowance Analysis to Retailers l-“S'O’QH: OQOO’D 10. 11. 12. 13. 165 Item Movement Analysis a. Seasonal Order Patterns b. Specialty Orders c. Effect of Advertising on the Sale of Items d. Discontinued Item Analysis and Review New Stores a. Prebudget and Labor Control Analysis b. Item Movement by Various Store Vblume Categories 1) Most Profitable Store 2) "Loss" Stores 3) New Stores Automatic Distribution a. Weekly Advertised Items b. New Items c. New Stores Inventory a. Daily Adjustments Analysis b. Physical Counts vs. IBM Control Counts c. Warehouse, Office, and Delivery Errors d. Central Billing on Retail Shipments Budget Variance From Actual Warehouse Retail (corporate) Dividends Stockholdings Proxy votes and Stockholder Votes (DD—IOU!” Advertising Income Suppliers Retailers (independent) Property, Equipment, and Depreciation Prebudget of Retail Meat, Produce, Bakery, and Other Perishables -- Sales and Labor 920 0’9, Sales per Man—hour Warehouse (wholesale) Retail . Rating Advertising Income to Purchases by Suppliers Merchandising Cost/Sell Audit and Equalization (Warehouse Profit Control) QIOU‘QD ILI. 15. 16. 17. 18. 19. k 166 Merchandise Variances a. Count and Recount b. "Cents Off" Deals c. Advances d. Declines e. Off Label Deals Inbound Freight Costs a. Freight Register b. Expected to Arrive Date Compared to Actual c. Inbound Freight and Routing Cost d. Freight Claims Period End Accounting a. Unmatched Receivings b. Unmatched Invoices c. Fixed Entries (Rent, Depreciation, Etc.) Marketing Analysis a. Sales Quotas by Sales Counselors, Stores and Territory b. Sales Comparison With Past Performance and Quaotas c. EVTOP Verification of Quantities to Buy Out-of-stock Report a. Total Cases and Dollars Ordered but not Shipped b. "Out" by Reason Code Retailer Returns - Allowances a. Cases, Dollars and Returns 1 b. Warehouse Scratch-off Analysis 1Charles P. Kreichelt and Michael J. Roach, The Role of Data Processin in the Food Industr , Food Marketing Paper #3 (lflast Lansing, Michigan: Food Marketing Program, Michigan State University, 1967) Mimeographed 167 APPENDIX B Interview Format: Supermarket Chain Interviews INFORMATION MANAGEMENT FOR SUPER MARKET CHAIN PRODUCT MIX DECISIONS: A SIMULATION EXPERIMENT OPERATING EXECUTIVE'S FIELD INTERVIEW FORMAT CHAIN ORIGINAL CONTACT ADDRESS PERSON INTERVIEW TITLE DATE FUNCTION y TIME START THE INTERVIEW WITH A PREFACE COVERING THE FOLLOWING: 1) Who I am and where I am studying. 2) General statement of my dissertation topic area and research methodology. (Give a copy of summary statement.) 3) Why I am interviewing chains. A) What I hope to gain from the interviews. 1) OPERATING EXECUTIVE The first thing I would like to do is draw a flow chart of your organization's decision process for NATIONAL BRAND manufacturer's products. (note: Show the example "Information System".) TIME PRODUCT DAY OFFERING BY 1 SUPPLIER ZOHH< m>¢m T .. Bobmomm mmHm ho mmombom mme zommmm mmom mmmb mo ZOHmHomm mmm mme mo mam¢qH¢>< mH 9mm: mo onmHomQ mm¢z mWQZDz mmombom ZOHB¢SmOm2H H< m>¢m EmBH mmem mme zommmm mmoo QmmD mo onmHUmm MQHmm Hm¢ m>m QmmD mo ZOHmHomm «HmmeHmo Ema mmm<2 om3 Aa_zoov mozemm me<>Hma amsmoIzoqe 2H mezHoa MQ<2 onmHomm mamZDz Mooqm 3. Now I would like you to forget about corporate restrictions, time and money limitations, or even what you think is feasible or realistic. I want you to draw a flow-chart of your ideal product mix decision system. In other words, if you could have any information that you wanted, and could organize it in any way that you wanted, how would you make product mix decisions? TIME DAY 1 ZOHH< m>m mo onmHomm mQ<2 mamZDz QmmD ¢HmmHHmo mma mmm< ZMHH mmadwbqswm mma mg: zommmm mmoa mmmb mo ZOHmHomm we; mmmvgz mo mmombom ZOHHSAmoE/HH a; I READ CHAIN EMBLEM” INITIALIZE VARIABLES READ ITEM DATA SET YEAR = ONE SET PERIOD = ONE SET WEEK = :NE 0 B 65 CHAINSIM (Continued) Page 2 of 9 I.) 6 <9 ONE +1 SET STORE NUMBER = ONE SET COST. NUMBER = ONE GENERATE A RANDOM NUMBER MATCH RANDOM NO. WITH ITEM IN STOCK ON THE STORE'S SHELVES E Q IDENTIFY SECOND CHOICE ITEM CHAINS IM (Cont inued) Page 3‘of 9 PG?" CUSTOMER ACCEPT SECOND CHOICE ITEM ? IS SECOND CHOICE ITEM AVAILABLE ON THE STORE'S SHELVES RECORD THE PURCHASE AND UPDATE THE SHELF INVENTORY OF THE ITEM CHAINSIM (Continued) IS TODAY A STORE ORDm‘ DAY FOR THE STORE SET ITEM STO E NUMBER + l ONE CASE OF THE ITEM FIT ON THE SHELF CALCULATE AND RECORD ORDER QUANTITY Page 4 of 9 9O CHAINSIM (Continued) SET ITEM NUMBER = ONE 1, ITEM # = # + STORE ORDER THE ITEM ? CALCULATE TOTAL QUANTITY ORDERED HAVE ALL ITEMS Page 5 of 9 CHAINSIM (Continued) SET ITEM # E l Page 6 of 9 ea I4 iii-3 IS THE WAREHOUSE INVENTORY ENOUGH TO FILL ALL ORDERS COMPUTE QUANTITY ALLOCATED TO EACH «HI:- ,STORE SET STORE STORE # = 1 #=# , + l "“‘—“’1 A UPDATE STORE AND WAREHOUSE INVENTORIES STORES BEEN CHECKED e» WISE CHAINSIM (Continued) Page 7 of 9 HAVE . ALL ITEMS NO BEEN CHECKED ? GENERATE THE DAILY REPORTS TODAY THE END OF A WEEK ? NO SET ITEM # COMPUTE A SALES FORECAST FOR THE ITEM CHAINSIM (Continued) Page 8 of 9 ; COMPARE SALES (;E) FORECAST WITH EXPECTED LEVEL OF INVENTORY THE INVENTORY). Y’S FORECAST ITEM# =#+ COMPUTE ORDER QUANTITY wEEK# =#+ PRINT WEEKLY REPORTS I. <96) CHAINSIM (Continued) ‘R NT PERIOD REPORTS INCLU DING A PERIOD OPERATING ‘ Illa i“ Q HAVE ENOUGH YEARS BEEN SIMULATED ? YES c so») Page 9 of 9 PERIOD #=# +1 YEAR# NO a=#+—-—-—— 1 191 APPENDIX E Flow-Chart of BUYSIM APPENDIX E FIDW-CHARI‘ OF "BUYSIM" Page 1 of 7 CW 3 1 SET DATA GROUP NUMBER EQUAL l READ INPUT DATA BUYSIM - Continued ESTIMATED I’DVEFILE‘ITI‘ OF THE NEW ITEM 8 THE AVERAGE. i/IOVEIVENT OF THE ITEA‘IIS NOW CARRIED ti 1 ESTII‘IATED MOVE‘IIICEIYI‘ = ESTIMATED MOVEJVLENT X TEST MARI’CE‘I‘ EXPANSION OF SALES I) ESTIMATED MOVEMENT =- ESTIMATED MOVEMENT X NEW ITEM'S SHARE OF 'I'EST MARKET SALES ESTIMATED MOVEMENT = ESTIMATED MOVEMENT X THE RATING OF THE INTRODUCTORY PROGRAM ESTIMATED MOI/’EIVEI‘TT = ESTEVIATED MOVEMENT X 'IHE RATING OF THE NATIONAL ADVERTISING BUYSIM - Continued Page 3 of 7 C99? ESTII‘v’IATED MOVEMENT ESTII-‘IATED MOVEP'IENT X THE RATING OF THE LOCAL ADVERTISING ESTIFv’LATED MOVEMENT = ESTIMATED MOVEMENT X THE RATING OF THE REACTION OF COMPETITORS NEW ITEM = NEW ITEM + l ARE ALL MOVEMENT ESTIMATES COMPLETED SELECT FIRST ITEM GROSS DOLLAR SALES = RETAIL PRICE X MOVEMENT BUYSIM - Continued 1 GROSS MARGIN DOLLARS PER WEEK = PER UNIT GROSS MARGIN DOLLARS X MOVEMENT DOLLAR CONTRIBUTION = GROSS MARGIN DOLLARS PER WEEK - HANDLING COSTS l NET PROFIT PER WEEK 8 DOLLAR CONTRIBUTION - INDIRECT EXPENSES PER WEEK NO HAVE DATA FOR ALL ITEMS BEEN CALCULATED CALL SORT - MOVEMENT Pageuof'? BUYSIM - Continued " CALL SORT - CROSS MARGIN PER CENT I CALL SORT - GROSS MARGIN DOLLARS PER WEEK TION PER WEEK CALL SORT - NET PROFIT PER WEEK CALL SORT - DOLLAR CONTRIBU- 1 SET CRITERIA = 1 SELECT FIRST ITEM 1% VALUE (ITEM) = NUMBER OF ITEMS +' 1 - RANK(ITEM,CRI‘IERIA) l PageSof? BUYSIM - Continued Page 6 of 7 ITEM= ITEM+ l CALL SORT - VALUE PRINT READINGS FUR TABLE GROUP PRINT ITEM NUMBER 8 RANKINGS BY GROUP ITEM AND BY NUMBER L l CRITERIA GROUPS ME I) Cm» D BUYSIM - Continued SUBROUTINE SOKP STARI‘ Page 7 of 7 SELECT FIRST REVERSE'IHERANKINGS OF'IHE'IWOITEIVB 199 APPENDIX F NEW ITEM EVALUATION FORM ITEM NAME CASE COST CASE RETAIL SUPPLIER ITEM NO. CASE PACK MANUFACTURER GROSS MARGIN 3 SIZE TEST MARKET DATA EXPANSION OF MARKET SHARE OF MARKET INTRODUCTORY PROGRAM ACTIVITY PROGRAM RATING 1. Couponing___ 1. Excellent (1.20) 2. Sampling 2. Strong (1.10) 3. Cents-off___ 3. Average (1.00) h. 2-for A. Weak (0.90) 5. Other 5. Poor/None (0.80) NATIONAL ADVERTISING GUARANTEED MEDIA GROSS RATING C. RATING t YES NO POINTS 1. Excellent 1.30 1.00 TV 2. Strong 1.20 0.90 Radio 3. Good 1.10 0.80 Magazines 4. Average 1.00 0.70 Newspapers 5. Fair 0.90 0.60 Other 6. Weak 0.80 0.50 TOTAL 7. PoonNone 0.70 O.MO LOCAL ADVERTISING GUARANTEED MEDIA TOTAL SPOTS PER WK. RATING YES NO TV 1. Excellent 1.30 1.10 Radio 2. Strong 1.20 1.00 Newspaper 3. Good 1.10 0.90 Other 4. Average 1.00 0.80 5. Fair 0.90 0.70 SUMMARY: Reach ; Frequency . 6. Weak 0.80 0.60 7. Poor/None 0.70 0.50 COMPETITIVE REACTION NUMBER OF COMPETITORS CARRYING 1.___(0.90); 2.___(O.95); 3.___(1.00); A.___(1.OS); 5.___(1.1O) ITEMS FOR COMPARISON ITEM DESCRIPTION CASE PACK CASE COST CASE RETAIL MOVEMENT 200 APPENDIX G Item Rankings According to Seven Criteria for Four TypeSCOf Dog Food m m hoomoi ¢ amoo m oooomd ¢ -. n omo¢m~ m o.m¢m~ mo hadmrmzouhm -ON~ N h o~.N0I m wm.o N Nnno¢~ N ~N. m oo.o- a o.N~oo mean no xdwwhm «acuu m m ehohml o hNom o ~n.¢¢ Cu 0‘. w ¢¢uom¢ w oonmN 000m non Imco uuooo N~ - NN.¢~—I dd mmocmn Nu odon Na 00. o mm.mam Nd oomoo~ mm on no a<>~¢ amaso a N~ on.oo~l Nu chaos: h o¢oum _— poo N ¢o.oo- N oow0mm on no 4<>~a ~¢o~o ~ ~ «Good! 5 éco —— OOoMM m w—. N— omoem— o— ooN-N o~u~ an no hm) uoomo 0 Cu mmomoi o— Nsaon ¢ ouoo~u N Na. ¢ o—oomh c oooh- mad Cu no hm> umomo m m ~Qo~MI o doann o~ h0.o¢ k N~o du hooumN s o.NowN zxru am no hm) ~¢omo a .0 omommt o ~¢u¢u w mvoo¢ h N~o o~ moom¢m o o.mNmm ~>_4 am we hm) umomo m e mm.¢ml m meood o doohm c m~o h enocmm - oonoN 000m mo bdaax ~00mo ¢ N ¢m.ONI N NeodN m Nwouo a NN. o oo.~—¢ o ooN¢—N xa Imhm huoz Gupta—w: h_uo¢c hwz «(Jase ZMMMMM mmoao mwcac zn~h<¢ Om224u I m02—x241 twp— o~ ¢~ - Nd NA oz_xz<¢ >¢oz Imbm z~4 do htmxox >>as 42b: m;az<«um no mid; Quad Oadd oaa< Quad Quad ca4< oaa< Dada can szu J<>~¢ 4—«0 ex: 4<>—¢ >>d+cd3m 4<>~¢ cho~ NNoo~ NuOOu Nmowo thoo Nocoo Nmooo N¢ooo Nmooo NNoeo Nuooo Nac~o NNo~o N~o~o ZS—haudumma Imb— - ¢~ ad Cu dd DZ~¥zmdzzjm owbro~wx N— fld o m «A _3 ON.mmNI NuosN_I dmom¢N| oo.ool Onamddl «moonl emoNo~I ¢womml omamNI ~N.moml ~N.¢~I nmoaoal so.0¢u mo.mNI wa2 sun pr10 c N o~ “a «a Md N— odomocd mcomNn Nooomnd co.hcm 00o~m~ coomcm— smcm¢m ~moN¢m mo.mmN oncommm mo.~o~d mm.~¢n~ ~¢.cmN oconm wwadm «<44co mmcmc wQZ—xz¢ m.NnN~ o.~mo oo¢mo m.ON¢ N.o—m oommmu o.moo 0.0N~ .m»_z:. b2m2w>or mN Om m nu no mhw> no mbw> mN 7~>¢¢o ma 7~¢yb >>¢wo m 7—dwb >>dd mane o modNo m o Nm.MN! m Nm.cu od no.mN N no. N do.N¢o m o.odw 0m wxnm mmzddm MOONO ¢ dd m¢.mdd1 N deoodl ¢ oo.oo m 00. N NN.¢mnd o o.mNN NN nwsn mw7d¢u nOONo o o mm.mou Nd neoNNI o wo.~m N mo. m mn.NMN N o.NNo Ndmd«<> mzdwa mmONo m Nd om.N¢dI m No.NdI N m¢.NN w co. d NN.momd m o.me S midyd mwzddw M¢ONC N d 0N.wdl N no.nm m om.NN d ¢d. o do.¢Nm m o.moo N mrddn mw7d¢o MMONO ozde<¢ rum: awn xwwx gun xwwz «we Nzwu «Nd mwdqm dmdd73d 7Cdedmummo twdd >dCt omNIudwr Nducxa Nmz waddco 7dw~