Abstract illustration of four tall vessels filled to different heights on a teal field

← ResearchResearch note9 min read

The quiet failure

What happens to a six-product bundle when one product sells out, and why nobody is told.

Abstract

What does it cost when one product in a bundle runs out? Two decades of research on stockouts show that shoppers who meet an unavailable item often leave, that the damage spreads to the other items in the same order, and that it lasts beyond the visit. A bundle is more exposed than a single product page, because it needs every one of its components at once, and it is most exposed at peak, when stock moves fastest. The practical conclusion is that a bundle should follow the shelf as it changes, and that the most useful warning is the one that arrives before the product is gone.

Key findings

  1. 01A bundle is only as available as its least available product, so its availability is always lower than that of any single item in it.
  2. 02In the largest published review of out-of-stocks, covering 71,000 shoppers, 31% bought the missing item at another shop and 9% did not buy it at all.
  3. 03A field test at a catalogue retailer found that a stockout on one item made customers more likely to cancel the other items in the same order.
  4. 04Online experiments suggest that revealing a stockout late, after the shopper has tried to buy, is worse than showing it straight away.
  5. 05Hiding sold-out products from bundles is live in Kitenzo, and warnings before a product sells out are not yet shipped.

A bundle is only as available as its least available product. That sounds obvious, and it is the cause of one of the most common ways a good bundle stops working.

One product in a six-product bundle sells out. On a page that was built once and left alone, one of three things happens. The bundle keeps selling something you cannot ship. Or it breaks. Or the shopper builds most of a box and finds out at the last step. In every case the page looks exactly as it did the day before. The path through it now has a dead end, and no report says why the bundle you spent weeks on has stopped converting.

This note looks at what published research says about that moment: what shoppers do when something is missing, what it costs, and why a bundle is the place in a store where the problem bites hardest.

Fig. 1Days of stock left, by product in the bundleIllustrative
Beans41d
Grinder3d
Dripper28d
Kettle55d
Dashed line: time left to swap it outHealthyRunning out

What shoppers do when something is missing

The reference point is a 2002 study by Thomas Gruen, Daniel Corsten and Sundar Bharadwaj, written for the grocery industry’s trade bodies and paid for by Procter & Gamble. It pooled 52 studies of out-of-stocks covering 661 shops, 32 product categories, 29 countries and 71,000 shoppers. Its headline figure has been quoted ever since: on average, 8.3% of items were not on the shelf when a shopper looked for them. Most retailers fell between 5% and 10%, and for fast-selling and promoted products the rate regularly passed 10%.1

The more useful part is what the shoppers did next. The study sorts their responses into five, averaged across eight categories.

What shoppers did when an item was out of stock. Worldwide averages across eight grocery categories, from surveys of shoppers in physical shops.
ResponseShare of shoppersWhat the retailer keeps
Bought the item at another shop31% 1Nothing
Substituted a different brand26% 1A sale, often a cheaper one
Substituted the same brand19% 1A sale, often a smaller one
Delayed the purchase15% 1A later sale, if they come back
Did not buy the item9% 1Nothing

Four in ten of those shoppers were lost to the retailer outright, and the authors put the typical cost at about 4% of sales.1 Corsten and Gruen gave the finding its lasting title two years later in Harvard Business Review: "Stock-Outs Cause Walkouts".2 The report adds a detail that matters for what follows. When one item is missing, shoppers tend to substitute or wait. When several are missing, the odds that they go to another shop rise.1

These numbers describe supermarkets at the turn of the century, where a substitute sits a few inches along the shelf. They should not be read as a forecast for an online store. The pattern has held up better than the percentages: a missing item sends a real share of buyers somewhere else.

It depends on how much they wanted it

Gavan Fitzsimons ran four laboratory experiments to find out why some stockouts hurt more than others. The answer was commitment. When the missing item was one the shopper had already been weighing up or preferred, they reported lower satisfaction with the whole decision and were more likely to switch shops on later trips. When it was an item they had little attachment to, the stockout could even help, by making the choice easier.3

Put that beside a bundle builder. By the time a shopper has put four products in a box, they have invested in the box. A product that turns out to be missing at step five is not one they glanced at in passing. It is part of something they had decided to buy, which is the condition Fitzsimons found to be most damaging.

The damage spreads through the order

The strongest evidence comes from a field test at a mail-order catalogue, reported by Eric Anderson, Gavan Fitzsimons and Duncan Simester in 2006. They followed more than 22,000 customers who ordered during the test and then tracked what those customers bought for 13 months afterwards.4

Items that were in stock when ordered ended as a completed sale 86% of the time. Items that were out of stock did so 62% of the time, and 32% of them were cancelled, against 4% for items in stock. Then the finding that matters most here. When one item in an order was out of stock, the other items in the same order were less likely to complete as well: 71.2%, against 83.1% when everything was available. The authors’ explanation is that products bought together often belong together. If the sheet is unavailable, the customer has less need for the pillowcases. Customers who met stockouts also ordered less in the following year.4

One more result is worth a merchant’s attention. The retailer tried several ways of softening the news, and offering a discount to persuade customers to wait for the item was the least profitable of the responses tested.4 The authors caution that the sizes of these effects belong to one catalogue and will differ elsewhere. The direction is the part they expect to travel.

Online, and whether to show the gap

Two studies move the question online. Xiaoqing Jing and Michael Lewis analysed customer transactions at an online grocer and found that stockout rates had a dramatic but nonlinear effect on profitability, in both the short and the long run, and that different kinds of customer reacted differently. Their encouraging conclusion was that much of the benefit could be won with small reductions in the stockout rate.5

Els Breugelmans, Katia Campo and Els Gijsbrechts ran an online grocery experiment on a question every bundle page has to answer: what should the shopper see? They compared showing the stockout straight away, revealing it only after the shopper tried to buy, and showing it together with a suggested replacement. Revealing it late was the worst. It created confusion, sharpened the sense of loss and made shoppers less likely to buy in the category at all. Suggesting a replacement made substitution easier and substantially raised the chance that the suggested item was chosen, unless the suggestion was a more expensive product, which shoppers treated with suspicion.6

The worst place to learn that a product is gone is the last step of a box you have already built.

Why a bundle is more exposed than a product page

A product page depends on one item. A bundle depends on several at once, and that changes the arithmetic. Think of the bundle as a chain. It holds only while every link holds.

The research above explains why each of those gaps is expensive. The shopper is committed, in Fitzsimons’ sense, because they chose the bundle or built it.3 The products are complements, like the sheets and pillowcases, so one missing piece weakens the case for the rest.4 And on a page that does not follow stock, the news arrives late, which is the policy that performed worst in the online experiment.6

There is also the matter of who finds out. A sold-out product page says so, and the merchant sees it. A bundle with one missing component still loads. Its photographs are still good. The only sign is a conversion rate that drifts down, and that has a dozen other possible causes.

Why it happens at peak

Stock runs out when things sell quickly, and the 2002 study had already found the highest out-of-stock rates among promoted and fast-moving products.1 For most online stores the fastest days of the year are the four from Black Friday to Cyber Monday. Shopify reported that its merchants took $14.6 billion over that weekend in 2025, up 27% on the year before, with sales peaking at $5.1 million a minute just after midday Eastern time on Black Friday.7 Those figures are Shopify’s own, unaudited and covering every merchant on the platform. They say nothing about bundles in particular. They do show how much demand arrives in very few hours.

Now work a bundle through it. A merchant builds a six-product gift box in October and promotes it for Black Friday. The box sells well, which means it draws down six products at once. Some of those products are also selling on their own. The one with the least stock, perhaps a seasonal candle bought in a single batch, runs out on Saturday morning. Nobody is watching, because everybody is packing orders. From that moment the best-promoted page in the store is offering something that cannot be sent in full, during the hours when it has the most visitors it will have all year.

Nothing here needs bad management. It needs only a bundle that was fixed in October and a shelf that kept moving.

What is not known

We have found no published study that measures stockouts inside product bundles specifically. The argument in this note joins findings from grocery shelves, a catalogue and online grocers to a piece of arithmetic. We think the join is sound, and it is still an inference.

The large numbers in circulation deserve care. The analyst firm IHL Group estimates that inventory distortion, its term for out-of-stocks and overstocks together, cost retailers $1.73 trillion worldwide in 2025, of which about $1.16 trillion was sales lost to out-of-stocks.8 It is the most quoted figure of its kind. It is also an analyst’s estimate, the method is not set out in the public summary, and we read that summary on the site of a software company that uses it in its marketing. It is fair to take from it that the problem is large. It is not a number to plan against.

Nor do general benchmarks fill the gap. The Baymard Institute puts the average rate of cart abandonment at about 70%, averaged over 50 studies, and its survey of reasons lists extra costs, slow delivery and forced account creation.9 Being out of stock is not among the reasons it lists. That tells us abandonment is common, and nothing about how much of it a missing bundle component explains.

Last, the best response is not settled. Hiding an unavailable product keeps the path clear, but a shopper who came for that product may wonder where it went. The online experiment favoured openness and a sensible suggested replacement over concealment that ends in a late surprise.6 It did not test a bundle, where the alternative to hiding one product may be losing the whole box.

What it means for a bundle, and where our work sits

The first fix is to follow the shelf. Kitenzo bundles follow Shopify inventory. Sold-out products can be hidden from a bundle, and so can whole bundles, across the store or one bundle at a time. Shoppers do not build a box that cannot be sent, and they do not meet the gap at the last step. That is live today, and it deals with the late surprise, the outcome the research treats most harshly.

The better fix is to know before it happens. A product rarely sells out without notice. There is usually a stretch of days in which it is plainly going to, and in which a swap would be easy: a replacement chosen calmly, of similar price, before the weekend and not during it. The evidence on suggested replacements points the same way.6 We are working on bundles that notice that stretch, say so, and offer the swap for the merchant to accept or decline. We are deliberately not going to describe how the warning is worked out.

That second part is still in the lab. Early warnings and suggested replacements are not in the product, and we would rather say so than imply otherwise. Until they are, the arithmetic above is worth doing by hand before any peak: list the components of your best bundle, find the one with the least cover, and decide now what takes its place.

References

  1. 1Gruen, T. W., Corsten, D. S. and Bharadwaj, S. (2002). Retail Out-of-Stocks: A Worldwide Examination of Extent, Causes and Consumer Responses. Grocery Manufacturers of America, Food Marketing Institute and CIES.
  2. 2Corsten, D. and Gruen, T. (2004). Stock-Outs Cause Walkouts. Harvard Business Review, May 2004.
  3. 3Fitzsimons, G. J. (2000). Consumer Response to Stockouts. Journal of Consumer Research, 27(2), 249–266.
  4. 4Anderson, E. T., Fitzsimons, G. J. and Simester, D. (2006). Measuring and Mitigating the Costs of Stockouts. Management Science, 52(11), 1751–1763.
  5. 5Jing, X. and Lewis, M. (2011). Stockouts in Online Retailing. Journal of Marketing Research, 48(2), 342–354.
  6. 6Breugelmans, E., Campo, K. and Gijsbrechts, E. (2006). Opportunities for Active Stock-Out Management in Online Stores: The Impact of the Stock-Out Policy on Online Stock-Out Reactions. Journal of Retailing, 82(3), 215–228.
  7. 7Shopify (2025). Shopify merchants generate record-breaking $14.6 billion in Black Friday Cyber Monday sales. Shopify News, 2 December 2025.
  8. 8IHL Group (2025). Inventory Distortion Study 2025, as summarised in "The $1.73 Trillion Wake-Up Call: Why Retail Planning Must Change". Summary published by Board International.
  9. 9Baymard Institute (2026). Cart Abandonment Rate: 50 Statistics & Why People Abandon (2026). Baymard Institute.