What is watch time: watch time explained, and how to read it without being fooled
Watch time is the total amount of time people spent watching a video, summed across every view and normally reported in minutes or hours. It is the only common video metric that treats attention as a quantity rather than as a yes or no event, which is why it behaves so differently from view counts.
It travels under several names. Total watch time is the sum. Average watch time, average view duration and average watched duration all describe the per-view figure. Watch hours is the same sum in a bigger unit.
What does watch time mean?
It means seconds actually spent, not people reached. A hundred people watching three seconds each and ten people watching thirty seconds each produce identical watch time, and represent completely different outcomes for your business.
Platforms define it precisely because it drives their own economics. LinkedIn's Campaign Manager documentation gives total watch time as "the total amount of time a video keeps playing on the screen, including any replays, while it remains visible to the viewer", and average watch time as "the average amount of time spent watching your video per view". YouTube Help defines average view duration as "average minutes watched among those who stayed to watch", calculated from engaged views. Note what each one quietly includes or excludes: replays, visibility, and which views qualify.
How is watch time calculated?
- Total watch time is the sum of every viewer's seconds watched, across all views in the period.
- Average watch time is total watch time divided by the number of views. Whichever view definition your platform uses lands directly in that denominator, so a loose view definition drags the average down.
- Average percentage viewed is average watch time divided by video length, times 100. This is the version you can compare across videos of different lengths. The raw seconds figure is not.
Watch time explained: a worked example
A service department sends 300 inspection videos in a month, each two minutes long. 210 are opened, and the platform records 168 minutes of total watch time, so average watch time is 48 seconds, or 40 percent of the runtime.
The following month the team cuts the videos to 70 seconds. Total watch time falls to 141 minutes, which looks like a decline, while average percentage viewed climbs to 63 percent and approvals go up. Total watch time rewards length. Percentage viewed rewards holding attention. Report the wrong one and you will optimize for the wrong behaviour.
Total or average: which one should you report?
Total watch time answers a reach question and scales with volume, so it is the right number for a library, a channel, or a quarter. Average watch time and average percentage viewed answer a quality question about a single video.
Wistia's State of Video reporting makes the trade explicit: shorter videos hold a higher percentage of their audience, while a longer video with a much weaker percentage can still accumulate more total watch time. Both statements are true at once, and quoting only the flattering one is the most common form of video reporting theatre.
Where watch time misleads
- Replays inflate it. Some platforms count replays toward total watch time, which is defensible, but it means the metric can exceed the runtime for a single viewer. Check the rule before you divide by anything.
- An open tab is not a viewer. Time counted while the player is technically playing in a background tab is time nobody spent watching. Visibility conditions matter, which is why LinkedIn writes one into its definition.
- It hides the shape. Two videos with identical average watch time can lose viewers in completely different places, and only the retention curve shows which. The average is a summary of the curve, not a substitute for it.
- Comparing raw seconds across lengths. Forty seconds on a 70-second video is strong. Forty seconds on a ten-minute video is a bounce. Convert to percentage before comparing anything.
- Averaging away the tail. A handful of long watches can carry the mean. The median, or the percentage still watching at 50 percent of the runtime, is more honest.
How watch time fits with the other numbers
Sequence them and each one earns its place. Play rate covers the invitation, view rate covers delivery and the counting threshold, watch time covers the attention you actually got, and engagement rate covers what people did with it.
One thing watch time cannot see is speed. A video watched in full three days late has the same watch time as one watched in full ten minutes after it was sent, and only one of those still has a customer on the other end. Platforms built for one-to-one sending, VentaVid among them, fire a notification the moment the customer opens the video for exactly that reason. Attention has a shelf life, and the totals do not record it.