A field guide for marketers and growth teams. Twenty-one posting rules, tested against 23 million TikTok videos and 13.3 trillion views. Most of them are myths.
Contact: research@layers.com
Generic best practices, a guru's checklist, or whatever worked for somebody else's account are weak foundations for a growth plan. Layers builds and runs content-growth experiments for apps - and experiments are only as good as their starting assumptions.
This study exists to sharpen those assumptions. The evidence from 23 million videos is one input; it gets combined with your product and its functionality, your audience and use cases, your brand and growth surfaces, and what your own account's results have already shown.
Why 12.68M of 23.26M? Every claim here compares a creator's videos against that same creator's other videos - which requires repeated observations. We analyze the videos belonging to the 189,718 accounts with at least eight posts and complete view data; the rest are indexed but can't support within-creator comparison.
Scope: public TikTok videos, 2020–2025, across major geographies and languages. Deleted and unavailable videos are tracked as an outcome (see Part III), not silently dropped. Findings are TikTok-specific; treat any extension to Reels or Shorts as a hypothesis, not a conclusion. Full inclusion criteria: working paper.
13.3 trillion seconds is roughly 421,000 years - about 5,770 lives of 73 years, lived end to end. Each dot above is ten of those lives. One second per view is the most conservative math possible; real watch time is far larger. Our species, for scale, is about 300,000 years old.
Each post gets a rank among its own account's posts. Audience size, follower loyalty, and niche cancel out completely. If a posting tactic works, its posts must rank above the account's average. That's the whole test.
For the statisticians: every gap shown here clears a 95% confidence bound computed the conservative way - resampling creators, not videos. The intervals are in the full report and working paper.
Your median post = 1×. Your 90th-spot post ≈ 8-9× that for a typical account, and the spread is wide. Near the middle: +4 spots ≈ +14% views; from your middle to your best few ≈ 8× the views. Every number ahead is a spot, with its view-equivalent.
Take your 50th-best video out of 100. Post it at the perfect hour instead of the worst one, and it becomes… at best, your 49th. That's the entire prize. The perfect day of the week can't even do that.
That is the share of the variation in a video's rank - within its own account - that the hour and day of posting explain. If performance were a 100-meter race, posting time decides the first 7 millimeters.
Denominator: within-creator rank variance across the full analyzed corpus. Definition and confidence interval in the working paper.
Find each creator's golden hour in the first half of their history. Check the second half. Posts in the golden hour: 0.5924. Posts in every other hour: 0.5924. It "held up" for 51% of creators - a coin flip. A retrospective "best hour" is, on average, sampling noise.
Note the distinction: a dashboard reading the past overfits randomness. A controlled experiment - hold variables steady, repeat, update - doesn't. That difference returns at the end of this report.
Nobody's thumb hits second zero on the quarter hour - those timestamps are machines. Within the same 4,086 accounts, scheduled posts outrun organic ones - and the edge belongs to creators who schedule only a few posts: almost three spots, fewer flops, more hits. Accounts that schedule everything get nothing. The scheduler isn't boosting those posts. Creators put their best videos into the scheduler.
Tested outside the twenty-one numbered rules; it carries no scoreboard row.
Travel and food lean slightly weekend. Business and family content lean slightly weekday. Every one of these - the largest calendar effects in 23 million videos - is worth about one spot out of 100.
We searched 189,718 creators for anyone whose posting clock mattered at all. We found 616 - about 3 in every 1,000, all mega-accounts. And even for those few, timing covers the first meter and a half of the race.
Pick posting times that fit your team's workflow and keep them. Use scheduling software freely - it costs nothing and correlates with stronger work. Then reinvest every hour you used to spend optimizing the calendar into the next piece of content.
Part I spent four of the twenty-one rules on the calendar, and posting time explained 0.007% of the difference. The other seventeen are in the 32 slides we will email you: 13 that don't hold, 3 that are half-true, 1 that survives intact, then 4 slides on what to test next.
Free. We will email you the full deck, the report and the working paper. Layers will not sell, lease or transfer your email to any other company.