Instagram Reels Watch Time — the KPI the Reels algorithm actually ranks on, and the one most agency reports still bury.

How it works.

When Instagram serves a Reel to its first cohort of viewers, it is running a distribution test. It measures how long, on average, that cohort watched the Reel — both in absolute seconds and as a percentage of the Reel's duration. If the average watch time clears an internal threshold (which correlates in observed portfolio data to roughly 60% of duration for well-produced Reels), Instagram serves the Reel to a larger next cohort, and the process repeats. Compounding distribution over 24-72 hours is what the term "gone viral" actually describes. The reverse is also true — if average watch time comes in below 40% of duration, the test cohort has told Instagram the content is not worth showing to more people, and distribution collapses inside a day. The retention curve underneath average watch time is the diagnostic that tells you why: a curve that falls off in the first three seconds is a hook problem, a curve that plateaus and then drops mid-Reel is a pacing problem, a curve that dies in the final third is a payoff problem. Fixing average watch time without looking at retention curves is guesswork — the two together are the diagnostic pair.

1. Pull Reels insights per post. The Graph API returns total plays, unique accounts reached, and average watch time in seconds.
2. Compute watch time as a percentage of duration. A 22-second average on a 30-second Reel (73%) is very different from a 22-second average on a 90-second Reel (24%). Store both.
3. Pull the retention curve. Where did viewers drop off? First three seconds is hook; mid-Reel is pacing; last third is payoff.
4. Baseline against the account's own history. The useful comparison is against the account's trailing 30-day Reels median, not a public benchmark.
5. Act inside the first 24 hours. If a Reel is coming in below the 40% threshold, boosting it will not save the organic curve — use the diagnostic to fix the next production, not to rescue the last one.

Where Reels watch time is misread.

1. Reporting absolute seconds without the duration ratio. A 22-second average on a 90-second Reel is a distribution killer, not a win.
2. Ignoring the retention curve. Watch time tells you a Reel underperformed; the curve tells you why. Skipping the curve means the same failure pattern shows up again next month.
3. Boosting a Reel with weak organic watch time. Paid distribution will show reach on the report, but the algorithm's account-level signal has already been set — and the boost will not lift future organic distribution.
4. Publishing long Reels without a strong hook. The failure mode healthcare brands run into most often — a 75-second explainer that only holds 30% because the first three seconds looked like a clinic ad.
5. Treating a viral outlier as a template. One Reel that hit 70% watch time on a novelty format does not mean the format is repeatable — treat it as a data point, not a strategy.
6. Comparing watch time against public benchmarks instead of the account's own baseline. Public benchmarks average across categories that behave nothing like healthcare.

Frequently asked questions.