5AgendaReview what we know about retail out-of-stocks from our researchPresent 7 general areas that need to be addressed:Most areas focus on dataShow how measurement can direct us to solutions by revealing the root causesDemonstrate our approach to reducing out-of-stocks.5
6Why should we pay attention to OOS? Lost Sales & Margin For ManufacturerLost Sales And Margin For RetailerDomino Effect On CategoriesDissatisfied CustomersHere are some findings from our 2002 study. These generated the interest (and another research grant) for the current study!Let’s find out why.6
7Two Studies 2002 GMA/FMI/CIES Study Current 2007 Study Worldwide Extent of OOSShopper Reaction When Faced with OOSRoot CausesCurrent 2007 StudyAimed at SolutionsPreliminary Report Completed in JunePublished in September7
8First Study’s Objectives Examine extent of Out of StocksExamine cause of Out of StocksExamine consumer response to Out of Stocks…on a worldwide basis, with the objectives:1. to present an updated and accurate “map” of facts surrounding retail out-of-stocks in the Fast Moving Consumer Goods (FMCG) industry,to examine out-of-stocks worldwide, examining rationale for similarities and differences,
9Research Project Inputs: 52 Studies 16 previously published academic and industry studies36 studies proprietary to this reportCovering:Number of retail outlets examined: 661Number of FMCG categories included: 32Number of consumers surveyed world-wide: 71,000Number of countries represented: 29This was a huge study! So what did we find out…
15EXTENT VARIES BY CATEGORY… Reliable data from three or more studies
16EXTENT VARIES BY DAY OF WEEK Reflects expected patterns due to shopping and deliveries
17More than half are OOS more than 24 hours! 3/25/2017EXTENT: DURATIONMore than half are OOS more than 24 hours!55% OF OOS LAST LONGER THAN 24 HOURS – USUAL REPLENISHMENT TIME
18Background: Extent Interpretation and Implications In spite of heavy investments to improve supply chains, worldwide OOS levels still average 8%, or from the shopper’s perspective, for every 13 items a shopper plans to purchase, one will be OOS.For promoted items, OOS levels average 16%, which translates to one OOS item for every 7 promoted items a shopper plans to purchase.Thus, in an industry heavily dependent upon promotions, the impact of one-seventh of promotional dollars is reduced.Sales velocity always affects the rate of OOS.18
19Q: HOW MUCH HAVE OOS RATES CHANGED? Background: ExtentQ: HOW MUCH HAVE OOS RATES CHANGED?Coca-Cola Research Council 1996 study = 8.2% (USA only)Our GMA/FMI/CIES 2002 study = 8.3% (Worldwide; 7.9% USA)A: Not Much.But… there are so many new kinds of technology in scanning systems, databases, computer assisted ordering (CAO) systems, etc…19
20WHY HAVEN’T OOS RATES CHANGED? Background: ExtentWHY HAVEN’T OOS RATES CHANGED?Technology improvements have been offset by process complexitySKU proliferationPromotional proliferationStore level assortmentStore level planogrammingRetailers face increased pressure to keep labor costs downInteresting side note: Isn’t this the way life with technology works in general? We get something new, and then we think of new ways to make life more complicated.20
21Substitute – same brand Substitute – different brand SHOPPER RESPONSETHERE ARE 5 SHOPPER REACTIONS WHEN FACED WITH AN OOS:Do not purchasePurchase elsewhereSubstitute – same brandSubstitute – different brandDelay purchaseIt’s also important to mention here that in addition to knowing how consumers respond to OOS, we also know how consumers respond to assortment reductions.21
23SHOPPER RESPONSE REGIONS Note differences in brand substitution across regions!23
24SHOPPER RESPONSE Varies Greatly by Category Range of store switch varies from 13% to 40%Fem Hygiene buys at another store 3 times more than Towels24
25SHOPPER RESPONSE Grocery Store Featured Category Source: ECR-UK 2005
26SHOPPER RESPONSE Drug Store Featured Category Source: ECR-UK 2005
27Multiple Out-of-Stocks In a Single Shopping Trip Can Cause the Shopper to Leave the Store Source: GS1 Columbia, “Diagnosis Report,” 200727
28QUIZ QUESTION: EFFECT OF MULTIPLE OOS IN A SINGLE TRIP With an average OOS level (8%) and a shopper purchasing 40 items – statistically what % of trips will he/she be completely satisfied (i.e. no OOS)?A. 4%B. 24%C. 44%D. 64%E. Can’t tell from theinformation given
30LIKELIHOOD OF 100% CUSTOMER SATISFACTION If a retailer can cut the OOS rate in half, the potential for 100% satisfaction skyrockets!Thanks to Synchra Systems, Inc. for this chart!
31IMPLICATIONS: RETAILER SALES LOSSES DUE TO OOS ARE ABOUT 4 PERCENT Sales Losses are similar worldwide, but vary greatly among categories31
32Calculating a company’s lost sales from OOS: OOS Rate _______%xCategory AvgLost Sales _______%Total Category/Organization Sales $_____=Sales Lost to OOS $_____Example:Avg OOS rate 8%XMFR Avg Loss 30%Category Sales $1B=Lost sales $24,000,000Typical Retailer Sales Loss/$1B total sales is about $32 million32
33FINDINGS: IMPLICATIONS The implications of our findings suggest that the cost of out-of-stocks to retailers is greater than what has been reported in previous studies.Our findings show that a typical retailer loses about 4 percent of sales due to having items out-of-stock. A loss of sales of 4 percent translates into a earnings per share loss of about $0.012 (1.2 cents) for the average firm in the grocery retailing sector where the average earnings per share is about $0.25 (25 cents) per year.
34Motivation – Additional Costs ManufacturersRetailersOOS Lowers Impact of Promotions and Trade Promotion FundsOOS Distorts True Store Demand, thus Forecasting, Category Management and Related Efforts are Less Accurate and EffectiveOOS Increases Overall Costs of Relationship with Retailer (Increased Post-Audit Activity, Irregular Ordering)OOS Distorts True Shopper Demand thus Decreases Forecasting and Ordering AccuracyOperational Costs are Increased through Personnel Looking for OOS Items, Providing “Rain Checks” to Shoppers, Unplanned Restocking, etc. (could be $1.0 Million for 100 stores)OperationalDirect Loss of Brand Loyalty and Brand EquityOOS Encourages Trial of Competitor BrandsLowered Overall Effectiveness of Sales Team ResourcesDirect Loss of Store LoyaltyDecreased Customer SatisfactionOOS Encourages Trial of Competitors’ StoresPermanent Shopper Loss Rate is Undocumented, but Annual Cost is US$1 Million per Every 200 ShoppersTwo notes for this slide:First, the impact is much larger than lost sales alone.Second, in the retailer operation quadrant, note that in the report we present a method to estimate the personnel costs.Strategic
35Costs of Addressing OOS in Store For Retailers:Labor Spent Satisfying Shopper OOS Questions $800/week/store for an Average Food StoreAbout $4.1million annually – 100 storesFor Shoppers:Shoppers spend >20% of the Average Shopping Trip Waiting for an Answer
36Let’s Examine the Causes of Out-of-Stocks Where does the breakdown occur?Supply chain?Retailer ordering?Retailer merchandising?Uneven consumer demand?36
37Lowering OOS Begins With Understanding the Causes of OOS Retail store ordering and forecasting causes (about ½ of OOS)Retail store shelving and replenishment practices where the product is at the store but not on the shelf (about ¼ of OOS)Combined upstream causes (about ¼ of OOS)This leads us to focus on the two major retail components: store and shelfCredit: Gruen, Corsten, and Bharadwaj 200270-75 percent of out-of-stocks are a direct result of retail store practices
38Summary of Findings of OOS Causes 3/25/2017UPSTREAM CAUSES OF OOSSummary of Findings of OOS Causes(Worldwide)Same as previous slide but details the the upstream causes.Other Cause4%Store OrderingRetail HQ or13%Manufacturer14%Distribution CenterFIRST STUDY TO IDENTIFY SHELVING AS ONE OF THE KEY CAUSES10%Store Forecasting34%Store Shelving25%
39LET’S SUMMARIZE THE CAUSES: Store Forecasting – 35%Ineffective algorithmsLong forecasting cyclesStore Ordering – 13%Late order / no orderInappropriate replenishment intervalsStore Stocking – 25%Inadequate or poorly allocated shelf spaceShelf stocking frequencyCongested backroomWarehousing – 10%Poor ordering policiesData accuracy issuesManagement Errors – 14%Last-minute price / promotion decisionsInaccurate or obsolete product informationManufacturer Availability – 4%Packaging, raw material or ingredient allocationCapacity issuesCan skip this slide if time is short.39
40So… We know the extent, consumer responses, basic causes. The implication of doing nothing is huge.Some retailers are actively addressing OOS and making progress.Given the number of potential remedies, it appears that fixing one or more OOS root causes should be a fairly easy task for retailers.However, knowing where to begin and which remedy will produce the most efficient and effective results relative to the invested resources remains a key barrier to implementation.What is next?40
417 Key Areas that Impact OOS We have to understand product flow.We have to understand measurement of OOSDue to OOS (and lots of other reasons), sales and demand are differentMost of the time, inventory data is inaccurateShelf space is often inadequate for the fast moversIt helps when stores comply with plansKeeping shelves and back room straight really matters41
421. We have to understand product flow (i.e., to the shopper) 42
43There Aren’t That Many Fast Moving Items New analyses of POS data provide us with a clearer picture of product movement across time. The conclusion: a relatively small number of items constitute the majority of the store’s total sales.
44Product Movement—Lower Volume Stores A lower volume grocery store with 50,000 items will sell:5,000 items in a typical dayIn a typical week, the yellow line crosses 80% at 15%, or 7,500 itemsChart provided by Standard Analytics44
45We have to understand SKU’s sales velocity and variability Measurement & FocusWe have to understand SKU’s sales velocity and variability…and focus on the ones that matter45
462. We Have to Understand Measurement of OOS, How Measurement Points to Root Causes, and How Root Cause Understanding Points to Solutions46
47OOS Measurement Method 1 Manual Audit ApproachLabor Intensive, costly to use ongoingBelieved by EmployeesData IntensiveError ProneNote that both methods can point to root causes of OOS, and these root causes can point to solutions.Note that several companies have developed algorithms to estimate OOS. We mention two of them in the presentation.47
48Manual Audit Example: Root Cause Percentages Note that this was a manual audit of 600 random items across 125 stores.1. Start with replenishment (15%)—determined by how many OOS items have inventory in the back room?2. After the 15% replenishment has been identified, then how many of the remaining have PI > 0? In this case, PI inaccuracy is the leading cause (42%) of OOS.3. And so on, identifying 12% due to ad, etc.
49OOS Measurement Method 2 Perpetual Inventory SystemWhen on-hands = 0 (or less), then item is OOSMany retailers already have PI systemOn hand data is badIs itself a cause of many OOS49
50OOS Measurement Method 3 Point of Sale Data Approach>85% Accurate (even false positives have benefit)Gives value to lost salesCalculates durationExtensive ReportingCostly to set up, cheap to run ongoingTwo Partner VendorsData VenturesStandard Analytics
51POS Data Estimation Example 3/25/2017POS Data Estimation ExampleExample 1:(3 lost sales)Example 2:(4 lost sales)Works best with fast moving items.The algorithm determines each item’s velocity (using 52 week history)Item expected velocity varies as the store velocity varies and price of the item variesWhen an item’s purchase cycle (expected velocity) is interrupted, that item is deemed “Out-of-Stock”51
52Example: Top 100 OOS Items by Store This report helps:Identify items with consistent Out-of-Stocks.Identify day and time of OOS Events.Understand the extent that promotional activity drives Out-of-Stocks.Identify items that need modification of delivery schedules.Also: Use POS data to examine frequency attributes to show patterns52
53Q: What Else Can We Do with POS Generated OOS Data? A: Find Patterns of OOSPromo Velocity Underestimated OOS Correlate with Promotion ScheduleWeekend Sales Underestimated Item Consistently OOS on WeekendsInadequate Shelf Space Short Duration OOS (< 1 day), Additional Supply is Clearly On-handDistribution Center OOS Relatively Long Duration OOS with High Correlation in Geographically Close Stores
54Sample Assessment Patterns Pattern 1: Promo Velocity UnderestimatedStore A, Fresh Express American Salad 12 ozProblem Corrected in JanuarySimple map here—note when the OOS pattern has been identified and fixed.Copyright Standard Analytics, LLC All rights reserved.
55What does this pattern indicate? Tell the audience that this is the egg category before showing the information at the bottom.This store needs to add shelf space or check shelf stock and restock the large eggs shelf more frequently55Copyright Standard Analytics, LLC All rights reserved.
56What does this pattern indicate? This store probably has an inadequate replenishment schedule for fast moving PLAIN PITA BREAD.The item is usually OOS by Thursday, and back on Friday evening.It is usually OOS again by Saturday or Sunday, and replenished by Tuesday.It looks like there are 2 deliveries a week, but 4 or more are needed.56Copyright Standard Analytics, LLC All rights reserved.
57What Does This Pattern Indicate? Problem: Item sells nearly every day - very few zero-sales daysAlmost daily stock-outs - demand is usually not met.Full sales units /day, average sales 21 units / day.Occasional multi-day stock-outs.Solution: Increase daily supply by approx. 60%; Check shelf 3x per day.57Copyright Standard Analytics, LLC All rights reserved.
583. OOS Disguises Actual Demand 47% of OOS due to poor forecastsLost sales is unobserved because most customers, who do not find the product that they intended to buy, make a decision to not buy, buy elsewhere, or substitute, without registering the non-purchase of the intended item with the store.Forecasting models do not include estimations of lost sales and simply forecast future demand on the basis of historical sales.Researchers have attempted to estimate demand with unobserved sales. All models conclude that that lost sales can be substantial and that it is strongly influenced by average demand and demand uncertainty.58
644. OOS Linked to Inventory Accuracy Issue 1: Product Data AccuracyData inaccuracy in retailers’ inventory databases comes from a variety of causes including:Merging previously independent databases; this happens due to corporate mergers, and also through the joining of previously separate systems.Accuracy with product replacements, new items that don’t get in the database correctly, and purging information on discontinued items.Manufacturers introduce temporary product changes, such as bonus packs, where a new UPC/GTIN code that follows the bonus pack, but then reverts back to the old UPC/GTIN.
65Product Data Accuracy Small differences can have a large effect. Third-party vendors, such as 1SYNCH, have evolved to facilitate data improvements. The effects of data alignment on lowering OOS can be substantial as the following two pilot studies reported by Capgemini/GCI 2005 show:In Latin America (Mexico, Guatemala, and Columbia), Procter & Gamble and several retail customers reduced purchase order errors from 3.6% to 0.8%, and this resulted in a decrease in OOS items from 8% to 3%.
664. OOS Linked to Inventory Accuracy Issue 2: Perpetual Inventory (PI) AccuracyStudy (USA drug store chain):Out Of Stock Rates were Benchmarked by In- Store Shelf Audits:4.1 % OOS Where OOS Matched P. I. (i.e., P.I. = 0)8.9% OOS Where OOS Do Not Match P.I. (i.e., P.I. >0)
67PI Accuracy Observations 45.4% of the time there was no variance18.8% of the time there was +/- 1 unit10% of the time there +/- 2 unitAsk, why is this so low?Improper scanningShrinkProduct lost in back roomStore stocker ability to make on-hand adjustments
68PI Accuracy For Items In One Location vs. Multiple Locations Our first step was to revisit our database. Again, we have now have over 20,000 store level items in which we can query on understand casual relationships. Based upon our hypothesis, we wanted to understand what the drivers to PI accuracy was. We started with items in only one location, meaning they were not on any end cap, profit planner, or POG in multiple locations. What we found was both compelling and concerning;If you look at items in only one location, the accuracy of the perpetual inventory jumps from an average of only 45% accurate +/- one to 52%. Similar improvements occur if you across the accuracy for +/- 1 where the accuracy increases 9% to 73% and +/2 where the accuracy increases another 9% to 83%.
69Steps to Improving Inventory Accuracy Focus Store Inventory Counts on:Physical OOSNegative On-HandsZero On-HandsOther Directed Items (e.g. high shrink, fast movers)Eliminated all other CountingReduced total cycle counts and improved accuracyResults:Increased PI Accuracy 19%Reduced Labor Costs in Inventory Accuracy by 50%Note that PI can be increased with lower effort due to focus.Note how critical this is for slow movers, because PI off by 1 or 2 delays orders of slow movers.Less critical for fast movers.69
705. Peak Demand Planograms 91% of the SKUs are Allocated Shelf Space Based on case packoutMany retailers use a “Red Dot” program (a work-around)86% of the inventory on shelf is in excess of 7 days supplyCutting slowest movers to provide shelf space for the fastest may prove cost effective.70
756. Planogram ComplianceTo what degree does compliance with POG link to levels of OOS?POG Compliance involves:DistributionSpaceArrangementItem ShelfBrand ArrangementSKU Level ArrangementWhole thing starts with understanding POG compliance.We developed the best practice for measuring POG compliance, which is summarized on the following slide.75
80POG Compliance Study Summary All Categories Showed Statistically Significant Relationships between Planogram Compliance and On-Shelf Availability (effect is 1% : 0.1%)With High Compliance the Benefit is Rather Small80
817. Item ManagementThere is a key need to keep items straight on the shelfDon’t cover holesDon’t hide productKeep shelf tags accurateThere is a key need to effectively get merchandise from the backroom to the shelfTest of shelf accuracy on OOS levels shows strong results81
82Item ManagementSales losses reduced by ~40% in Test stores with disciplined stocking practices versus Control82
84Recap and ConclusionsGet the product to the store, then get the product on to the shelf.Get it right on the shelvesIdentify and address the fast moversFor store formats with faster moving items, use POS estimation and look for patternsFor store formats with slower moving items, work on PI accuracyIn all cases get item data correct through data synchronization.
85Solve high risk products with Store OOS solutions Recommended ApproachMeasure & AssessHigh OOS risk products (fast movers)High OOS storesShelf versus Store OOSSolve high risk products with Store OOS solutionsSolve high OOS stores with Shelf OOS Solutions85
86Finding the Genie to Grant Our Wish: How to Solve Out of Stocks Measurement and AssessmentRoot Cause IdentificationApply SolutionsContinual ImprovementSee the whole picture and address what you can86
87Professor of Marketing University of Colorado, Colorado Springs, USA Contact for additional information:Thomas W. Gruen, Ph.D.Professor of MarketingUniversity of Colorado, Colorado Springs, USAFor a PDF copy of the 2002 study, you can download directly from:Also check the website for announcements on 2007 Report87