Scarcity and urgency: genuine and artificial

Note: This text is not legal advice. Have your specific implementation reviewed by a lawyer.

“Only 2 left”

Few patterns are so widespread and so rarely substantiated. Scarcity counts as a sure thing, and it is at the same time the area where shops most often cross the line into deception — frequently without experiencing it as a decision, because the feature is already built into the shop system.

Three things appear in practice as one: the quantity, the deadline and the reason. They work to very different degrees, and only one of them is properly supported. At the end comes the line at which information turns into deception.

Three forms that work differently

Quantity scarcity. “Only 2 left in stock.” The claim says something about availability — and incidentally something about popularity. That is why it can also be read as a form of social proof.

Time scarcity. “Offer ends today.” This is not about availability but about a moment of decision. The pressure comes from the deadline, not from the quantity.

Scarcity with a reason. “Sold out due to high demand.” Worchel, Lee and Adewole (1975) tested exactly this in their cookie study — and the ordering is revealing.

A second part of the same study did not work at all. Merely announcing that many further participants still wanted the cookie changed nothing — and on estimated price the finding reversed.

The reason for the scarcity works; the bare number of other takers does not. In shops the two regularly appear as one thing.

What the studies measure — and what they do not

Here lies the most important limitation of this field: a large share of the research measures valuation judgments, not purchases. Participants say how valuable, desirable or attractive something seems to them. That is a different outcome from a purchase decision with one’s own money.

Lynn (1991) pooled 41 studies with 51 tests. What was on trial is commodity theory — the assumption that anything useful seems more valuable the harder it is to obtain.

The result cuts both ways. Across the studies the effect is reliably present, and it is small: the mean effect size is r = 0.12, and 0.17 once five studies with reversed findings are removed. The r runs from 0 to 1; values around 0.1 count as small.

Lynn converts that into a more tangible form: instead of landing at 50–50, judgment lands at 44 against 56 when something is scarce rather than plentiful. At the same time he notes that hardly any of the studies used the scarcity cues that actually occur in marketing.

The direction is therefore well supported, the magnitude modest. The cookie study, incidentally, is one of the 41 works in that meta-analysis, so it is not a second, independent support.

Shi, Li and Chumnumpan (2020) systematically went through 66 papers published between 1970 and 2017. Their review is not a meta-analysis and yields no effect size; it sorts the field and names the gaps. Two points from it bear directly on pricing pages.

Scarcity works only where customers expect it. Where the claim does not fit the situation, what arises is not urgency but suspicion about the motive. That is exactly the case of the software download with three licences left.

Demand-based reasons count as especially liable to the suspicion of deception. Phrasings such as “in popular demand” or “over X sold” raise more doubt than a plain quantity, because customers do not credit sellers with being able to quantify demand reliably. The very pattern that worked most strongly in Worchel therefore carries its own risk in advertising. The advantage in effect is therefore bought at the price of a credibility risk.

The authors also note that research on the ethical and credible implementation of scarcity is practically absent.

When scarcity flips

Scarcity has a side effect that practice underrates. People who notice the pressure push back against it — that is reactance. Three triggers are typical:

• An implausible reason. A software download with “only 3 licences left” does not read as scarce but as dubious.
• A permanent state. If everything in a shop is always nearly sold out, customers learn that the display means nothing. The signal is then spent for the honest case as well.
• Too many levers at once. Countdown, stock display, visitor counter and pop-up on the same page do not add up; they devalue each other.

For shops this link is not verified. The meta-analysis by Rains (2013) pools 20 experiments — on flossing, alcohol, exercise, sun protection, organ donation. Exactly one concerned a commercial product. And the direct link between a threat to freedom and attitude was not significant there (r = 0.06). So reactance is a plausible objection here, not a measured finding.

Two findings from the Rosenberg and Siegel (2018) review are usable all the same. First, the counter-evidence from practice: across two experiments on anti-drinking messages, participants exposed to the strongly pressing version drank more than those who saw the restrained one. The reversal is therefore not a theoretical construct but something already measured — just not on pricing pages.

Second, a remedy, with a narrow condition attached: a closing line that explicitly leaves the decision with the reader — “the choice is yours” — damped reactance only for the strongly pressing version. With the restrained one it did nothing. Such a sentence repairs pressure you could have left out.

And one distinction belongs right here. Whether reactance disappears once the freedom can no longer be restored is open: Steindl and colleagues (2015) list the question explicitly as unresolved.

What does have support is a related distinction, and a closer one for shops. Laurin, Kay and Fitzsimons (2012) found that where a restriction is unavoidable, people rationalise it; where it is avoidable, they answer with reactance. Both are stronger where the matter concerns them personally. For a pricing page that means resistance arises not at the expired deadline but at the running one — where the pressure is visible and still avertible. None of this has been tested in a shop, and the finding is reported here at second hand, through Steindl and colleagues.

A reason changes not the whether but the how. Where a restriction came without a reason, people’s heart rate rose immediately; where it was justified, only after a delay. That fits Shi et al. (2020): scarcity with a comprehensible reason is processed differently from scarcity merely asserted.

Practical cases

What we deliberately do not recommend

• Resetting countdown timers. A timer that starts over on every visit asserts a deadline that does not exist.
• Fixed stock numbers. “Only 3 left” as a permanent state with no link to the warehouse.
• Scarcity for unlimited goods. For a digital product that can be delivered any number of times, a quantity claim is hard to justify honestly.
• Several urgency signals on one page. That is not amplification; it is a trust problem.

Ethics check: nudge or sludge?

Limits and criticism