One in Ten Shops Used Fake Countdown Timers. The Law Says No.

In 2022, consumer regulators in 23 EU states, Norway and Iceland checked 399 online shops. They found fake countdown timers on 42 of them [1]. On a UK shop page, a product was shown as "viewed 6,458 times in 24 hours". The shop sold seven units that day [5].

I started this post with a thesis: published sweep rates are high enough that you should treat any scarcity cue you cannot verify as no evidence. After reading the sources, I now hold a narrower thesis. The law is strict, the cues are sometimes fake, and the rates do not tell you how often a given cue lies. Here is the work.

The question

When a page says "Only 3 left" or "Offer ends in 09:58", should that change what you believe about stock or time? And what does the law say about it?

This is the sibling of my earlier post on fake "was" prices. I extend that post. Fake reference prices and fake scarcity are the same move: a number on the page that regulators test against a hidden record.

Data and where it came from

I used five kinds of source.

• The law. Annex I of the EU Unfair Commercial Practices Directive lists practices that are unfair in all cases. Point 7 bans "falsely stating that a product will only be available for a very limited time" to "elicit an immediate decision and deprive consumers of sufficient opportunity or time to make an informed choice" [3]. A blacklist item needs no case-by-case test of effect on buyers. The Dutch regulator applies the same logic to stock: a trader cannot say five items are left when that is not so [12].
• The 2022 EU sweep. 399 shops screened. 148 had at least one of three dark patterns. 42 used fake countdown timers [1]. The Commission published the results in January 2023 [2]. I did not find the sweep's own report with the full definition of "fake" or the count of shops with any timer. That gap matters below.
• A large crawl. Mathur and co-authors crawled about 11,000 shopping sites and about 53,000 product pages. They found 1,818 dark pattern instances and 183 sites with deceptive ones [7]. Their project page gives 393 countdown timer instances on 361 sites and 632 low-stock instances on 581 sites. The 183 deceptive sites and 234 deceptive instances are reported only as a total, not per pattern [6].
• Court and regulator cases. The UK CMA warned online sellers in April 2023 that "Only 5 left" misleads if there is no real shortage or stock refills fast [4]. The CMA case against Emma Sleep ended in a court-approved consent order on 22 May 2026, as reported by a law firm [5].
• Effect studies. I need these to ask whether real scarcity cues even work.

On the US side, I could not find an FTC case on countdown timers in this run. Law firm summaries say the FTC's September 2022 dark patterns staff report lists false urgency and scarcity among its pattern categories [11]. The enforcement examples in those summaries (Credit Karma, LendingClub, Amazon, Vizio) concern other deceptions [11]. So for the FTC I rely on a secondary source and I make no claim of a timer case.

Method

I did three things. I read the legal texts and case reports. I took the published counts and computed rates with a Wilson 95% interval. I computed bounds on what the counts can say about the share of cues that are fake. All arithmetic below is by hand, without the Lab. You can repeat it from the cited counts.

The EU sweep gives me my usual funnel of three numbers:

Step	Count	Share
Shops screened	399	100%
Shops with any of three dark patterns	148	37.1%
Shops with fake countdown timers	42	10.5%

Sample size? 399, and I do not know how the shops were picked. If the picks were not random, the interval below covers sampling noise only.

Result with numbers and uncertainty

The sweep rate. The share is 42 / 399 = 0.1053. The Wilson 95% interval is 7.9% to 13.9%. The Commissioner's "nearly 40%" headline is the 148 figure, which covers three patterns, not timers alone [1].

The crawl rates. Using about 11,000 sites as the denominator, which is approximate: 361 sites with timers is about 3.3%. 581 sites with low-stock messages is about 5.3%. All 183 deceptive sites together is about 1.7% [6][7]. The crawl is from 2019 data and the sweep is from 2022. They use different definitions, so I do not stack them.

The rate that matters is missing. A shopper does not meet a random shop. A shopper meets a cue. The question is: of the cues I see, what share is fake? Neither source gives that. The sweep counts shops with fake timers, not shops with timers. The crawl counts deceptive sites only in total.

I can bound it. In the crawl, at most 183 sites had deceptive patterns of any kind. So at most 183 of 361 timer sites had a deceptive timer. That is an upper bound of 50.7%. For low stock, 183 / 581 gives an upper bound of 31.5%. These are ceilings, not estimates. The true share is lower, because the 183 sites also include other deceptions such as fake testimonials [6]. I did not see the per-pattern split, so I cannot say how far below.

In the EU sweep, the lower bound is the other way. If every one of the 399 shops showed a timer, the fake share would be 10.5%. If fewer showed timers, the share is higher, up to 100%.

What a fake cue looks like. Emma Sleep is the clearest documented case. The CMA found timers that reset on page refresh or at expiry, and "high demand" claims with no stock shortage [5]. A law firm report says each countdown was presented as time-limited, but none were. It also says that when one sale ended, another replaced it, in some cases within 24 hours [5]. That shows a cue unrelated to the truth, not a cue that is sometimes wrong.

Do real scarcity cues work? The evidence is mixed. A field experiment by Li, Tsekouras and Cheng reports that scarcity and popularity cues raised purchase likelihood, and that cue-exposed customers produced 8.3% more revenue [9]. I read only the abstract, so I cannot give an interval. A before-and-after study of 199 items at one US retailer, Bon-Ton, found that a real "five or less left" notice cut the days between purchases by 14.8%. But it cut daily units sold by 17.6%, from 0.77 to 0.63, and net daily sales fell [8]. That design is not randomised, and it covers regular periods, not flash sales [8]. Real scarcity cues can backfire. Fake ones then carry a double cost to a shopper: no information, and a push to buy faster.

What the law adds

The law is stricter than most pages assume, in three ways.

1. No effect test. Blacklist point 7 bans the false claim itself [3]. A seller cannot defend a fake timer by saying it did not change sales.
2. The seller must prove it. A UK ASA ruling on a time-limited promotion went against an advertiser that could not show its claim was not exaggerated [Dutch and UK guidance in 12 describe the same stance on extensions: one extension is low risk, habitual ones are not] [12].
3. Real timers can be fine. The CMA says a checkout timer may be justified to stop buyers holding stock, but is harder to justify if used mainly to pressure them [4].

The Commission's 2022 behavioural study adds that countdown timers and time-limited messages are quite prevalent on e-commerce platforms. It also found 97% of the most popular sites and apps in its mystery shopping had at least one dark pattern [10]. That 97% is for a sample of popular sites, not all of the EU web [10].

Sensitivity: which assumption moves the result most

The result moves most with one assumption: the share of shops that show any timer at all.

• If 100% of the 399 shops showed timers, 10.5% of timer-showing shops were fake.
• If 25% showed timers (about 100 shops), the fake share is 42%.
• If 10.5% did, every timer was fake.

I picked these denominators to show the swing. They are assumptions, not data. Over that range the answer moves from 10.5% to 100%. No other input moves it that far. The interval on the 42 count (7.9% to 13.9% of all shops) moves the result by only about 3 points either way. The choice of sweep over crawl matters too, but I treat the two as separate estimates.

A second assumption is what counts as "fake." The sweep treats a timer as fake when it resets with the same offer still valid, or when the offer survives expiry [1]. A timer that is real but misleading in other ways falls outside that definition. So 42 may be too low. Equally, shops were screened by national authorities, and their selection may favour shops suspected of problems. That would push 42 too high. I cannot tell which effect wins.

Verdict

Three claims, three verdicts.

• "The law bans false scarcity and fake timers." Holds. Blacklist point 7 and the CMA consent order both support it [3][5].
• "Fake timers are common." Weak as stated. About one shop in ten in one 2022 sample is real, but "common among cues" is not shown. The data cannot separate 10% from 100% of cues [1].
• "Treat any unverifiable scarcity cue as no evidence." Weak. It is a sound rule of thumb when a page offers no way to check, because the cost to a seller of faking is small and the cost to you of believing is real. But the sweep rates do not prove it. A cue that is false in perhaps 10% to 50% of cases is still weak evidence, not zero.

My current view: I would not pay extra to beat a timer, and I would reload the page once. If the clock resets, I know. That is a check you can run in ten seconds, and it is better than any rate I can quote. My view would change if a sweep reported fake cues as a share of cues shown. A share under 5% would make me rank cues as modest evidence. A share above 50% would move me back to my first thesis.