Abstract illustration: rows of rounded blocks on shelves, most of them faded, a few lifting away in bright colour

← ResearchResearch note10 min read

The stock that did not sell

A bundle is judged on what it earns. Over Black Friday it can do a second job: turn slow stock back into cash.

Abstract

Most bundles are judged on one line: what they earned. That misses a second job a bundle can do, which is to move stock that has stopped selling before it has to be marked down or written off. Research on clearance pricing shows that markdowns are costly and teach shoppers to wait, and research on bundling shows both why a package can carry a weaker product and when it backfires. We argue that a merchant should score every bundle on two measures side by side, what it earned and what it recovered, and we set out how to do that with a spreadsheet.

Key findings

  1. 01What you paid for old stock is spent either way, so the right comparison for an aged unit is what it would fetch if it were still on the shelf in January.
  2. 02A study by Coresight Research and Celect put the cost of markdowns to US non-grocery retailers at $300 billion in 2018, or 12% of their sales.
  3. 03Bundling can move a weaker product alongside a stronger one, but experiments show that a cheap addition can lower what shoppers think the whole package is worth.
  4. 04Score each bundle on two measures: what it earned, and the revenue it recovered from stock that was at risk of a markdown.
  5. 05A bundle that earns a little less and clears stale stock can leave the business better off than the one that tops the profit report.

Every store has a shelf it would rather not talk about. Last winter’s candle. The colour nobody chose. The stock is paid for, it takes up space, and each month it is worth a little less.

Bundles are rarely asked to help. A bundle is judged on its bottom line: revenue, profit, the return on whatever discount it carries. By that measure the best bundle is the one built from products that already sell. This note argues that a bundle has a second job, which matters most from Black Friday to Cyber Monday: turning slow stock back into cash. To see whether it has done it, you have to keep a second score.

What stock costs while it sits

The cost of holding stock is called its carrying cost. It has three main parts: the cash tied up, which could have bought something that sells, the storage, and the risk that the product loses value before it sells, because it goes out of season, is replaced, is damaged or passes its date.

The last part grows with time, which is why merchants watch ageing: how long each unit has been in stock. Two plain measures go with it. Sell-through rate is the share of your stock that sold in a period. Shopify’s own reports calculate it as the units sold, divided by the units sold plus the units still in stock.1 Cover is how long the stock you hold would last at the pace it is selling. Shopify reports it in days, as the units left divided by the average sold per day.1 Three hundred candles selling five a week is sixty weeks of cover, which for a winter candle means more than one more winter.

Why the cost price no longer matters

Here is the idea the rest of this note depends on. The money paid for old stock is a sunk cost: it has been spent and no decision made today brings it back. Whether the candle cost $6 or $9 makes no difference to what you should do with it now. The only live question is what it will bring in from here.

Accountants have a rule that says the same thing. Under the international standard IAS 2, stock is carried in the accounts at the lower of its cost and its net realisable value: the price it is expected to sell for, less the cost of selling it. When that value falls below cost, the difference is written down as an expense in the period it happens.2 A shelf is worth what it can be sold for, and no more.

So the honest comparison for an aged unit is not what you paid for it. It is what the unit would fetch if it were still on the shelf in January, after the markdown it would need by then. That figure is its salvage value. Anything a unit sells for above its salvage value is money that would not otherwise have arrived.

The usual way out, and what it costs

The usual way to move slow stock is to cut its price. The scale of that habit is large. A 2019 study by Coresight Research and Celect, built on a survey of more than 200 senior retail decision makers in the United States, estimated that markdowns cost US non-grocery retailers $300 billion in revenue in 2018, about 12% of their sales. Just over half of those surveyed named misjudged inventory as a primary cause.3 Celect sold inventory software, so the study had a commercial interest behind it, and it describes large retailers, not small online stores.

Markdowns are also hard to get right. Stephen Smith and Dale Achabal, whose clearance pricing methods were tested at three large retail chains, describe the problem plainly: the retailer has little time to react, and a cut that is too deep gives away revenue while one that is too shallow leaves stock to be liquidated. The rate of sale also depends on how much of the assortment is left, so a range that has thinned out is harder to clear.4

Then there is what a markdown teaches. Economists use the term strategic customers for shoppers who decide when to buy as well as whether to, and who will wait if they expect the price to fall. Gérard Cachon and Robert Swinney modelled a retailer facing a mix of shoppers who pay full price, bargain hunters who buy only at a discount, and strategic customers who weigh today’s price against the chance of a lower one. When strategic customers are present, the retailer in their model earns less profit, and its best response is to stock less in the first place.5 That is a model, not a field measurement. Its lesson is that a visible discount changes what customers expect next time.

Black Friday is where discounts are most visible. Adobe, which tracks visits to US retail sites, reported that discounts in 2025 peaked on Cyber Monday at 31% off the listed price for electronics, 28% for toys and 25% for apparel.6

The older alternative: sell it with something else

Bundling is at least as old as the markdown. The modern analysis starts with William Adams and Janet Yellen, whose 1976 paper defined bundling as selling goods in packages and showed why it can pay: different customers value different things, and a package lets a seller reach people who would have bought only part of it.7 Stefan Stremersch and Gerard Tellis tidied the field up in 2002. They call a discounted package of separate products price bundling, and selling both the package and its parts mixed bundling. Their first proposition is that price bundling earns more than selling separately when one group of customers prizes the first product and another group the second. They also note that such a package adds no value of its own, so it needs a discount.8

Work that joins bundling to stock is thinner. Kevin McCardle, Kumar Rajaram and Christopher Tang modelled a retailer bundling basic and fashion products. Their conclusion is a caution: whether the bundle pays depends on the demand for each product, the cost of bundling, and how the demands for the products move together.9 Qingning Cao, Kathryn Stecke and Jun Zhang studied the opposite stock problem, a product in short supply, and found that a bundle can widen the market for the less attractive of two products. They also found that when both products are weakly valued, the firm should not bundle at all.10 Two weak products do not make a strong package.

A bundle has one more property that matters here. Its price is one number for several things, and shoppers find a combined price harder to judge than a single one. Michael Kamins, Valerie Folkes and Alexander Fedorikhin use that difficulty to explain their experiments on promotional bundles.11 For a merchant with aged stock, this is the attraction. The candle can leave the building at an effective price below its list price without that lower price ever being printed beside it.

And the Black Friday weekend supplies the traffic. Shopify reported that its merchants took $14.6 billion over the 2025 weekend, from more than 81 million customers.12 Adobe put US online spending for the five days from Thanksgiving to Cyber Monday at $44.2 billion.6 Those totals say nothing about bundles. They suggest that a slow product can be put in front of far more shoppers that weekend than in an ordinary week. That is our argument for the timing, not a published finding.

A markdown tells every shopper what the product is now worth, and a bundle does not.

Where it goes wrong

The evidence against careless bundling is at least as strong. Kimberlee Weaver, Stephen Garcia and Norbert Schwarz ran seven studies on what they named the presenter’s paradox. People who put a package together assume that more is better, and add every favourable item they have. People who judge the package average its parts. A mildly attractive addition therefore pulls the whole down, and in one study presenters spent more money on a product bundle and lowered its perceived value by doing so.13

Aaron Brough and Alexander Chernev found something sharper across five studies. When an inexpensive item is combined with an expensive one, shoppers can be willing to pay less for the pair than for the expensive item alone.14 And the experiments by Kamins and colleagues found that describing one item in a bundle as free lowered what people would pay for that item when it was later sold by itself.11

Three warnings follow. A stale product that looks cheap beside its companions can drag the bundle down. Calling the slow item a free gift damages its price afterwards. And a generous bundle cannibalises: it sells to shoppers who would have paid full price for the popular products anyway, so watch their full-price sales while it runs.

How to track it: two measures, side by side

The usual bundle report shows what a bundle earned. It does not show what it cleared. We suggest three steps, all of which fit in a spreadsheet.

First, give every product a rating for how much you want it gone. Base it on two things already defined: how old the stock is, and how many weeks of cover you hold. Four levels are enough. Then attach to each level your own estimate of the product’s at-risk share: the share of today’s selling price you expect to lose if the unit is still on the shelf in January.

A rating scale a merchant could use. An illustration with made-up round numbers, not a published standard and not a description of how any product works it out.
RatingWhat it meansAt-risk share
HoldFresh stock, selling at the pace you planned0%
NudgeMore cover than you want, no markdown needed yet25%
MoveAged stock that will need a deep markdown in January50%
ClearStock that will be written off: out of date or replaced100%

Second, for every bundle, record what it earned in the ordinary way. Third, record what it recovered.

Now a worked example. A home fragrance store runs two bundles over the Black Friday weekend and sells 200 of each. Bundle A holds three current bestsellers. Bundle B holds two bestsellers and last winter’s candle, which is rated Move. There are 300 of those candles in stock. In the bundle each candle accounts for $16 of the price, and the merchant expects a candle left until January to fetch $8.

Two bundles over one weekend. An illustration with made-up round numbers, not results from any store.
Bundle A: three bestsellersBundle B: two bestsellers and the candle
Bundles sold200200
Revenue$12,000$11,600
Profit (earned)$5,000$4,600
Rated units sold0200 candles
Cash released from slow stock$0$3,200
At-risk sharen/a50%
Recovered revenue$0$1,600
Fig. 1Two bundles, two measuresIllustrative

On the profit report, A wins by $400. But the report charged Bundle B the cost price of 200 candles, money that was spent a year ago whichever bundle ran. Follow the candles instead. Under Bundle A they are still on the shelf in January, where they bring in $1,600. Under Bundle B they brought in $3,200 in November. The difference is the $1,600 of recovered revenue. The bestsellers that Bundle B did not use are still in stock, and will sell at full price. Bundle B gave up $400 of profit to recover $1,600, which leaves the business $1,200 better off.

What to watch

  • •Sell-through of rated products, inside and outside bundles. If the candle sells no faster in the bundle than alone, the bundle is not doing the second job.
  • •Age of stock at sale. The average age of the units a bundle sells should be higher than the store’s average. If it is not, the bundle is selling fresh stock and leaving the old.
  • •Share of at-risk value recovered. The 300 candles carried $2,400 at risk before the weekend (300 units, $16 each, half of it at risk). Bundle B recovered $1,600, or two thirds.
  • •Margin given up per unit recovered. Bundle B gave up $400 to move 200 candles: $2 a candle. The January markdown would have cost $8 a candle. When the first number is higher than the second, mark the product down and leave the bundle alone.

What is not known

We found no published study that measures bundling as a way to clear aged stock in an online store. The case made here joins research on clearance pricing to research on how bundles are valued. It is an inference, and the worked example is arithmetic, not evidence.

Recovered revenue is also only as good as the estimates behind it. The at-risk share is a forecast of what a product will fetch in January. The honest check comes later: compare what the leftover units did sell for with what you assumed, and correct the ratings.

Last, the experiments on perceived value were run in laboratories.1314 They do not say how often a weak addition hurts when a shopper builds a box for themselves and can leave the candle out.

Where our work sits

You can try the idea without risking a weekend. Our bundle simulation, at kitenzo.com/simulation, runs a store through Black Friday. Every product in it carries a move rating, five steps from Fresh to Clear, set by how long the stock has sat. Every run is scored on two targets: profit, and the share of the value at risk in slow stock that the bundle recovered. The game shows what each suggested change does to both, so the trade is visible before you make it. It is a game with invented shoppers, and it proves nothing about your store.

In the product itself, reporting on what each bundle earned is live, and so are side-by-side tests between two versions of a bundle, the sound way to compare a bestsellers box with one that carries a slow product. Advice that takes your stock into account is in the lab. It is not in the product, and we are not going to describe how it is worked out. Until it ships, the two columns are worth keeping by hand: what each bundle earned, and what it recovered.

References

  1. 1Shopify (2026). Inventory reports. Shopify Help Center.
  2. 2IFRS Foundation (2003). IAS 2 Inventories. International Accounting Standards Board.
  3. 3Celect and Coresight Research (2019). Study Finds Markdowns Cost U.S. Retailers $300 Billion in Revenues in 2018. Press release for the report "Revealing the Hidden Costs of Poor Inventory Management", PR Newswire, 6 February 2019.
  4. 4Smith, S. A. and Achabal, D. D. (1998). Clearance Pricing and Inventory Policies for Retail Chains. Management Science, 44(3), 285–300.
  5. 5Cachon, G. P. and Swinney, R. (2009). Purchasing, Pricing, and Quick Response in the Presence of Strategic Consumers. Management Science, 55(3), 497–511.
  6. 6Adobe (2025). Adobe: Cyber Monday Hits Record $14.25 Billion in Online Spending with Over $1 Billion Driven by Buy Now Pay Later. Adobe Newsroom, December 2025.
  7. 7Adams, W. J. and Yellen, J. L. (1976). Commodity Bundling and the Burden of Monopoly. The Quarterly Journal of Economics, 90(3), 475–498.
  8. 8Stremersch, S. and Tellis, G. J. (2002). Strategic Bundling of Products and Prices: A New Synthesis for Marketing. Journal of Marketing, 66(1), 55–72.
  9. 9McCardle, K. F., Rajaram, K. and Tang, C. S. (2007). Bundling Retail Products: Models and Analysis. European Journal of Operational Research, 177(2), 1197–1217.
  10. 10Cao, Q., Stecke, K. E. and Zhang, J. (2015). The Impact of Limited Supply on a Firm’s Bundling Strategy. Production and Operations Management, 24(12), 1931–1944.
  11. 11Kamins, M. A., Folkes, V. S. and Fedorikhin, A. (2009). Promotional Bundles and Consumers’ Price Judgments: When the Best Things in Life Are Not Free. Journal of Consumer Research, 36(4), 660–670.
  12. 12Shopify (2025). Shopify merchants generate record-breaking $14.6 billion in Black Friday Cyber Monday sales. Shopify News, 2 December 2025.
  13. 13Weaver, K., Garcia, S. M. and Schwarz, N. (2012). The Presenter’s Paradox. Journal of Consumer Research, 39(3), 445–460.
  14. 14Brough, A. R. and Chernev, A. (2012). When Opposites Detract: Categorical Reasoning and Subtractive Valuations of Product Combinations. Journal of Consumer Research, 39(2), 399–414.