Glossary

What is audience retention?

Audience retention is the graph showing what share of viewers is still watching your video at each second. It is the metric platforms trust most, because attention can't be faked: people either kept watching or they left.

Reading the graph

Every retention graph tells the same three-part story.

  • The opening cliff: viewers who realise this isn't for them and leave in the first seconds. Some cliff is normal; a huge one means the packaging over-promised.
  • The middle slope: the slow leak as the video runs. Flatter is better.
  • Spikes and dips: spikes mark rewatched moments (gold, make more of those), dips mark the exact second people got bored.

The first fifteen seconds decide most of it

A video that holds its opening usually wins; one that loses the opening never gets to argue its middle. That's why hooks became a craft.

State the promise immediately, cut the throat-clearing, and get to the first payoff fast. Then check the graph: the drop-off points tell you exactly where to cut next time.

What retention decides

YouTube and TikTok promote on watch behaviour above nearly everything: retention is the proof that recommending your video was a good idea. It also fills the money meters directly, since watch hours are retention multiplied by audience.

One naming warning: delivery-service retention means followers surviving after a purchase. Same word, completely different metric. Context tells you which one is meant.

Common questions

What is a good audience retention percentage?

It varies hugely with video length: short videos hold higher percentages than long ones. Compare videos of similar length on your own channel, and study the shape of the graph more than the single number.

Where do I see my retention graph?

YouTube Studio shows it per video. TikTok and Instagram show completion and average watch time in their analytics.

Does retention matter more than views?

For ranking, usually yes. A video with fewer views that holds people beats a big-view video everyone abandons, because feeds optimise for time actually watched.

Every definition lives in the full glossary, and the services behind the vocabulary are on the services page.