Only Two Left in Stock
Scarcity works when it carries a social signal, and which cue works depends on what is being sold. A 1975 cookie jar and a meta-analysis of 131 studies ask different questions about it, and only one of them has a product-type answer.
Key takeaways
- Worchel, Lee and Adewole found in 1975 that cookies in scarce supply were rated more desirable than cookies in abundant supply.
- The detail that travels least is the most important: cookies that became scarce beat cookies that were always scarce, and scarcity caused by demand beat scarcity caused by accident.
- A meta-analysis of 416 effect sizes from 131 studies found scarcity cues generally raise value, desirability and purchase intentions.
- The moderator is an interaction with product type: demand-based scarcity most effective for utilitarian products, supply-based for experiences, time-based for high-involvement products.
- Most commercial scarcity cues are unverifiable, and the persuasive power comes from being a social signal supplied by an interested party.
In 1975 a psychologist put cookies in a jar and asked people what they thought of them. Some saw a jar with ten. Some saw a jar with two. The ones who saw two rated the cookies as more desirable, and consumer marketing has been built on that sentence ever since.
The sentence is also, as it happens, the weakest thing in this article. The badge on your screen tonight is the strong version of the finding or the weak one depending on how it is worded, and almost nobody selling you something knows which they have used.
Stephen Worchel
The study is Worchel, Jerry Lee and Akanbi Adewole, and the detail inside it is far more interesting than the summary that travels.
They did not simply compare scarce with abundant. They varied how the scarcity came about. Cookies that started abundant and became scarce were rated more valuable than cookies that had been scarce all along. And cookies that were scarce because of high demand were rated higher than cookies that were scarce because of an accident.
The mirror image held too. Cookies that had been constantly abundant were rated higher than cookies that started scarce and became plentiful.
So the effect turns on what the quantity implies rather than on the quantity. A thing running out because other people want it is being told to you by other people. A thing that is simply rare is not telling you anything.
The meta-analysis
That was fifty years ago and on cookies, so the sensible question is whether it holds at commercial scale.
Belinda Barton, Natalina Zlatevska and Harmen Oppewal pooled 416 effect sizes from 131 studies of product scarcity and purchase intentions. Their overall finding is that cues signalling current or potential unavailability generally enhance a product's value and desirability and increase purchase intentions.
Then the moderator, and it is not the one the cookie jar would lead you to expect. The meta-analysis does not crown demand-based scarcity, or any other kind.
What it reports is an interaction with the type of product. Demand-based scarcity is most effective for utilitarian products. Supply-based scarcity is most effective for experiences. Time-based scarcity is most effective for high-involvement products. There is no overall winner among the three in that sentence, and the authors' own framing is that managers need to consider these factors rather than pick a favourite cue.
That is a more useful finding than a simple ranking and a less quotable one. It also means the two literatures do not converge. Worchel found that scarcity caused by demand beat scarcity caused by an accident, which is a comparison between two reasons for the same shortage. The meta-analysis compares three kinds of cue across three kinds of product. They are different questions and only one of them has fifty years of distance on it.
The interaction above is stated in the paper's abstract. The pooled magnitudes for each cue type sit in the full text, so we do not characterise them here.
The shelf
Look at what retailers actually deploy and the three categories are all there, which is the point.
A low-stock message is supply-based. A live-viewers message and a recent-sales message are demand-based. [A countdown timer is time-based](https://thewhyweproject.com/articles/what-a-countdown-timer-adds-to-a-sale), and that is a category in the research with its own pooled estimate rather than a trick standing outside it. "Limited edition" is ambiguous and usually means supply.
What the evidence does not license is a ranking of those badges in the abstract, because on the meta-analysis the right cue depends on what is being sold. A supply cue on an experience and a demand cue on a utilitarian product are each being used where the research says they work best. The same cue on the wrong category is the weaker bet.
The strongest commercial versions combine all three: a small number remaining, attributed to other people wanting it, with a clock. That is three mechanisms in one badge, deployed without regard to which one suits the product, which is a reasonable description of most of what you see.
The part that should make you cautious
One caution from the literature, then two of our own.
The first is that purchase intention is the outcome most of these studies measure, and intention is not purchase. A meta-analysis of stated intentions tells you scarcity moves what people say they will do, which is real and is one step removed from the till.
The second applies to the old study rather than the new one, and it is the reverse of where scepticism usually gets pointed. Worchel is 1975 social psychology, a small sample, cookies and rated desirability, which puts it in the era and design class that has fared worst under re-examination. The meta-analysis is the more robust half of the pair and the founding demonstration is the fragile one.
The third is that we cannot tell you what the pooled effect looks like once corrected for publication bias. If the paper reports that figure, it sits in the full text, and we would rather leave the size open than guess at it. The pull of a crowd on a shopper is a different question, and one article asks [whether a long queue makes people want in](https://thewhyweproject.com/articles/does-a-long-queue-make-people-want-in).
What we would say about it
Our position is that the direction is well supported, that the size is not settled and we have no corrected estimate to offer, that which cue works depends on what is being sold rather than on a league table of cues, and that the everyday commercial version is unverifiable in any case.
The inventory number on a website is generated by a system you cannot audit. It may be true. It may be true of one warehouse. It may be a threshold below which the message triggers. The viewer count has the same property. You are being shown evidence about other people's behaviour by a party with an interest in the outcome, and the evidence cannot be checked.
None of which says retailers fabricate it. The claim is narrower: the cue's persuasive power comes from being a social signal, and a social signal from an interested party is worth less than one from a disinterested one, which is a distinction the badge is not designed to help you draw.
The question worth asking
When a scarcity cue works on you, it usually does so by converting a decision you had not made into a deadline you have to meet.
So the useful question is not whether the two remaining units are real. It is what you would have decided with no number on the screen at all.
If the answer is that you would have bought it anyway, the cue cost you nothing and possibly saved you a delay. If the answer is that you would have thought about it for a day, then the thing the badge removed was the day, and the day was the part you needed.
Sources
Worchel, S., Lee, J., & Adewole, A. (1975). 'Effects of supply and demand on ratings of object value.' Journal of Personality and Social Psychology, 32(5), 906-914.
Barton, B., Zlatevska, N., & Oppewal, H. (2022). 'Scarcity tactics in marketing: a meta-analysis of product scarcity effects on consumer purchase intentions.' Journal of Retailing, 98(4), 741-758.
A note on what this article flags: purchase intention, not purchase, is the outcome measured in most of these studies, and this is a marketing-adjacent literature of the kind this publication treats cautiously.