The 71% "Decide in 3 Seconds" Hook Stat Started as "Stop and Watch" in a 2023 TikTok Ad Study. Here Is What to Cite Instead
# The 71% "Decide in 3 Seconds" Hook Stat Started as "Stop and Watch" in a 2023 TikTok Ad Study. Here Is What to Cite Instead
**The 3-second hook rule may be good advice, but the numbers guides attach to it are older, narrower and more different from each other than they look.**
If you have read a hook guide this year, you have seen some version of "71% of users decide whether to keep watching in the first 3 seconds" or "65% of people who watch 3 seconds will watch 10." This post follows both numbers back as far as the open web allows, shows where the wording changed, and gives you three things to copy: a five-question stat checker, safer sentences to use instead, and a one-week hook test for your own channel.
## The two numbers and where each one goes
Sources read on 2026-11-03. Page dates are what each page showed or what its metadata said; where a page showed none, that is stated.
### Number 1: "71% decide in the first 3 seconds"
- **Animoto (updated 2025-07-16)** writes: "71% of TikTok users decide whether to continue watching within the first three seconds." It links to a page on asia-research.net and says "according to recent research."
- **Asia Research Media (published 2023-12-18)** is the page that link leads to. Its wording is different: "71% of TikTok users stop and watch the first three seconds of the video." It attributes the figure to a two-wave global study by the research firm Eyesee, which the page says tested 60 ads from 20 brands and surveyed 7,280 respondents using TikTok simulations, standalone ads and surveys. The page does not say which of those methods produced the 71%, and it does not link to the study itself.
- **Dynamoi (updated 2026-04-26)** writes "research shows 71% of users decide whether to keep watching in the first 3 seconds" and names no study.
So the trail on the open web ends at an ad-testing study of 60 ads, reported in a trade write-up. The original says users "stop and watch" the first three seconds. The copies say users "decide." Those are different claims. Watching three seconds is something a viewer does. Deciding is something a viewer is supposed to have finished doing by then. Nothing the original page says supports the second version.
### Number 2: "65% of people who watch 3 seconds will watch 10"
- **Kapwing short-form statistics page (metadata dates 2026-07-08 published, 2026-07-13 modified; the page body says "June 2026")** writes "65% of viewers who watch the first 3 seconds will watch for at least 10 seconds" and "45% will watch for 30 seconds or more," credited to "Facebook IQ via Brandefy."
- **Brandefy (2025-07-10)** writes "nearly 65%" and credits a "study by Facebook," linked to a Facebook for Business page. It gives no year and no study title. That Facebook page blocked automated reading, so the original wording was not checked in this run.
- **Salesforce blog (URL dated 2016/02, page itself undated)** is the earliest copy found. It writes that "only 65%" of people who watch the first three seconds of a video will watch at least ten seconds, with 45% continuing to thirty, credited to "Facebook." It references Q4 2015 campaign data, gives no sample or method, and the surrounding article is about video advertising, not Shorts.
Three things changed along the way. "Facebook" became "Facebook IQ." "Only 65%," which reads as a warning, became a benchmark. And a stat from a 2016-era feed-video context became a rule about TikTok and Shorts.
## What does not carry a source at all
Several pages in the same search results state numbers with no source on the page:
- **Scenith (2025-09-10)** says "87% of viewers decide whether to watch or skip your video within the first 3 seconds," along with other figures including "340% higher engagement," and cites nothing for any of them. The same page uses 340% for engagement in one place and for watch time in another.
- **OpusClip, written by its Head of Creator Partnership (2025-11-11)** says videos with three-second retention above 65% "receive 4 to 7 times more impressions," and says the author analyzed accounts "ranging from 5,000 to 500,000 followers." No sample size, period or data is shown. It is a vendor's own claim, and it is the kind of number that is easy to repeat and impossible to check.
- **Dynamoi (2026-04-26)** states a "200-500 users" initial test audience and a 65 to 70% threshold, with no source.
An unsourced number is not necessarily wrong. It is just not evidence, and a post that repeats it is borrowing confidence it has not earned.
## What TikTok itself says
TikTok's own "creative best practices for top performing ads" page (no publication date shown) contains no "first 3 seconds" statistic. The nearest figure is "90% of ad recall impact is captured within the first six seconds," credited to the 2020 TikTok Creative Guide on brand equity, a "Value of View" analysis. It is about ads and brand recall, not about whether an organic Short is shown to more people, and the page gives no sample or method.
That does not make the 3-second advice wrong. It means the strongest primary source this run found points at six seconds, for ads, in 2020.
## Copy this: the five-question stat checker
Before you put a number in a script, a caption or a client deck, ask:
1. **Who was measured?** Ad viewers, feed viewers, or Shorts viewers?
2. **What was measured?** Watched, stopped, decided, or recalled?
3. **When?** If the page gives no year, treat it as undated.
4. **How many?** A sample size or a named study, or just "research shows"?
5. **Who is selling?** If the author's product benefits from the number being true, label it as vendor data.
If a number fails two or more, do not present it as a finding.
## Copy this: safer sentences
| Instead of | Write |
|---|---|
| "71% of viewers decide in 3 seconds." | "One 2023 TikTok ad study reported that 71% of TikTok users stopped and watched the first three seconds (Eyesee via Asia Research Media, 2023-12-18)." |
| "65% of people who watch 3 seconds will reach 10." | "A Facebook figure reported in 2016 put it at 65% for video on that platform. It has not been shown for Shorts." |
| "You have 3 seconds to hook people." | "Opening seconds matter, and most guides agree. No public platform source this run found gives the exact cutoff for Shorts." |
| "Retention above 65% at 3 seconds gets 4 to 7 times the impressions." | "One vendor says so, with no data shown." |
## Copy this: the one-week hook test
You do not need a borrowed statistic if you can run your own comparison. This is a test design, not a result.
1. Pick one finished Short and cut three versions that differ only in the first three seconds: a question, a result shown first, and a plain statement.
2. Post one version per day over three days at the same hour, then repeat once more with a different topic.
3. For each, record the share of viewers who stay past the opening (use your platform's retention graph or its viewed-versus-swiped-away figure if it shows one), plus the total views at 24 hours.
4. Keep the version that wins on both for each topic, and note by how much.
Two weeks of this is a small sample and the topics will differ, so treat the answer as a hint for your channel, not a law. It is still more reliable than a number with no source.
If you would rather measure hooks on real videos than quote a stat, ClipNovia's video analyzer reports the hook mechanic and how long it lasts on a Short you choose, so you can compare openings by structure. [Try it here](https://clipnovia.io/?utm_source=blog&utm_medium=post&utm_campaign=three-second-hook-statistic-71-percent-origin-test).
## What this post cannot tell you
- The Facebook page behind the 65% figure could not be read, so its exact wording and date are unverified.
- The Eyesee study itself is not public in anything this run found, so what the 71% was measured on is unknown beyond the trade write-up.
- No source found measures a 3-second cutoff for Shorts specifically.
## FAQ
### Is the 3-second rule wrong?
Not necessarily. The advice to open strong is common and sensible. What is weak is the precision: the percentages attached to it.
### Is a hook stat from 2016 useless?
It is a data point about a different platform and format. It can be cited as history, not as a current benchmark.
### What should I cite instead?
Cite the narrowest true sentence, with the source and date, as in the table above, or cite your own test.
### Where do I learn how to read a viral Short properly?
Start with [how to reverse engineer viral videos](/blog/how-to-reverse-engineer-viral-videos). For a related paper trail on captions, see [the 85% watch-on-mute stat](/blog/sound-off-statistic-85-percent-origin-captions-test). For opening structure on six real Shorts, see [the cold open sheet](/blog/short-form-cold-open-sheet-six-viral-shorts), and for wording templates see [viral hook formulas](/blog/viral-hook-formulas-for-short-form-video).
## Sources
- [Why the first 3 seconds of video matter more than the next 30, Animoto (updated 2025-07-16)](https://animoto.com/blog/video-marketing/why-first-3-seconds-matter)
- [Tap into hidden potential with new TikTok creative insights, Asia Research Media (2023-12-18)](https://www.asia-research.net/tap-into-hidden-potential-with-new-tiktok-creative-insights/)
- [What is the 3-second rule on TikTok, Dynamoi (updated 2026-04-26)](https://dynamoi.com/learn/tiktok-music-promotion/what-is-the-3-second-rule-on-tiktok)
- [Short-form video statistics, Kapwing (2026)](https://www.kapwing.com/resources/short-form-video-statistics-tiktok-reels-and-shorts-by-the-numbers-in-2026/)
- [Psychology of viral video openers, Brandefy (2025-07-10)](https://brandefy.com/psychology-of-viral-video-openers/)
- [Facebook video ads global, Salesforce blog (URL dated 2016/02)](https://salesforce.com/content/blogs/us/en/2016/02/facebook-video-ads-global.html)
- [The 3-second rule, Scenith (2025-09-10)](https://scenith.in/blogs/three-second-rule)
- [TikTok hook formulas that drive 3-second holds, OpusClip (2025-11-11)](https://www.opus.pro/blog/tiktok-hook-formulas)
- [Creative best practices for top performing ads, TikTok for Business (undated)](https://ads.tiktok.com/business/en-US/blog/creative-best-practices-top-performing-ads)
*Verified on 2026-11-03. Update by 2027-01-15, or sooner if a platform publishes a measured early-drop-off figure for Shorts.*