Video Retention Rate Calculator – Where the Audience Leaves
Views count the people who arrived. Retention counts the people who stayed, and it is the only one of the two that tells you anything about the video. Every platform measures it, and every platform hands it back in a different shape: YouTube Studio draws a curve, TikTok gives a percentage watched, Instagram Insights gives raw viewer counts, and Meta Business Suite gives an average view duration and leaves the division to you. This calculator turns whichever of those you actually have into the figures the others would have given you.
How to calculate video retention rate
Three divisions carry the whole page. Everything else — the curve, the half-life, the drop-off strip — is one of them repeated at every point in time.
retention % = viewers still watching ÷ views started × 100
drop-off % = 100 − retention %
average % viewed = average view duration ÷ video length × 100
total watch time = views × average view duration
watch hours / 1,000 = 1,000 × average view duration ÷ 3,600A 60-second clip with 12,000 plays and 7,320 viewers still there at the half-minute mark is holding 7,320 ÷ 12,000 × 100 = 61.0% — a drop-off of 39.0%, or 4,680 people gone. A long-form upload handed back the other way — a 3:42 average view duration on a 10:00 video — is a 37% average percentage viewed, and across 48,000 views that is 2,960 watch hours.
Reading a retention curve
A single percentage tells you that people left. A curve tells you where, which is the only version you can act on. Four features are worth naming, and this tool reads all four off whatever points you supply.
The steepest drop is the consecutive pair with the largest fall in percentage points. Because analytics exports rarely sample evenly, the tool also reports points lost per second, so a 5-point fall across two seconds is correctly ranked as worse than an 8-point fall across twenty. The retention half-life is the moment the curve first crosses 50%, interpolated between the two bracketing samples rather than rounded to the nearest one — it is the single cleanest way to compare videos of different lengths. The hook is retention at 3 seconds and at 30 seconds, the two checkpoints creators are judged on most. And the completion rate is the value at the end — which this tool will only call a completion rate when your last data point actually lands within 2% of the stated video length, and otherwise labels honestly as retention at the last data point.
Why the first three seconds always collapse
Every autoplay, every mistaken tap and every scroll-past registers as a play the instant it begins. The opening seconds of any curve are therefore measuring a decision, not a viewing — and a fall of twenty or thirty points through the first three seconds is ordinary on every platform. What matters is not the size of that first cliff but where the curve flattens, and how high. A video that steadies at 60% and slides gently has a different problem from one that steadies at 60% and falls off another cliff at forty seconds.
When percentage viewed goes over 100
Absolute versus relative retention
Absolute retention is the share of everyone who started the video that is still watching at a given moment. That is what this calculator computes, and what a pasted viewer-count column contains. Relative retention compares that same curve against other videos of a similar length on the same platform, so it answers a different question: not how many stayed, but whether more stayed than usual. If your analytics offers both, use the absolute curve to find the moment that is losing people and the relative curve to decide whether it is worth fixing.
What retention is worth
Watch time is views × average view duration, so retention is a multiplier on traffic you have already paid for. Lifting an average view from 30 seconds to 45 is worth exactly as much as adding half again as many views, and it is usually far cheaper. The watch hours per 1,000 views figure exists to make that trade visible in a single number you can carry between videos.
Nothing on this page leaves your browser. Type the numbers, paste a column out of a spreadsheet, or load the sample curve — the arithmetic runs locally and no analytics account is ever touched.