← All posts

Data · August 9, 2026 · 3 min read

Dwell Time: The Metric Platforms Trust More Than Likes

Platforms quietly weight how long readers linger on a post more than public reactions. Why dwell time is hard to fake and how it changes writing.

There's a metric shaping your reach that you've never optimized directly, because you can't see it: dwell time — how long a reader actually lingers on your post before scrolling on. Platforms have leaned on it increasingly, and once you understand why, a lot of otherwise-confusing distribution behavior starts making sense.

Why platforms trust it

Every explicit signal has a corruption problem. Likes can be traded, comments can be farmed by pods, shares can be begged for — the moment creators learn a signal is rewarded, some of them manufacture it, and the signal degrades. Dwell time resists this, for a simple reason: it's the one signal readers emit without deciding to. Nobody pauses on a post as a favor. A reader either found the text worth their next fifteen seconds or they didn't, and their thumb reports the verdict honestly whether or not they'd ever admit to reading you in public.

That last part matters more than it seems. Plenty of valuable reading is deliberately silent — people research vendors, evaluate future employers, and follow competitors without ever leaving a public trace. Reaction counts miss this entire population. Dwell captures them, which means it measures something closer to actual attention than any engagement number does.

What it changes about writing

Once you know time-on-post is being measured, some tactics look worse and some look better:

The caveats

Dwell is honest but not precise. It can't distinguish a rapt reader from someone who wandered off with the app open, and it structurally favors longer content — a perfect four-sentence post will never post big dwell numbers, however hard it lands. Platforms know this and normalize for length and format, which is also why padding a short idea into a long post doesn't work: slow reading of thin content produces the leave-early pattern the system is specifically tuned to notice. And you can't audit any of this — dwell rarely appears on creator dashboards, so it disciplines your writing without ever showing you a scoreboard.

How to use a metric you can't see

You optimize dwell the same way you'd write if metrics didn't exist: by holding attention honestly. Run the thought experiment on your own drafts — read one at scroll speed and mark the sentence where a busy stranger's thumb would move. That sentence is your leak. Fix it, and you've improved the one number the platform trusts most, using the one instrument that measures it as well as the platform does: an actual reader, getting bored. The quiet good news in all of this is that the machinery has drifted toward rewarding exactly what writing teachers always wanted — being worth someone's uninterrupted time.