Creator analytics: metrics that matter
Views, watch time, likes and daily charts each answer one question. How to read them in the Creator Studio, how to compare videos fairly, and a fifteen-minute weekly review.
Four numbers carry most of the signal. Views measure the pitch. Watch time measures the video. Likes measure the ending. The per-day chart tells you whether a change did anything. Read each for its own question, adjust for video age and length, and review weekly rather than daily.
Analytics dashboards are where good creators go to make bad decisions. The numbers are accurate. The trouble is that each one answers a specific question, and it is easy to ask a number a question it cannot answer. Views cannot tell you whether a video was good. Watch time cannot tell you whether the thumbnail worked. Reading them together, each for its own question, is the skill.
VideoBB's Creator Studio reports views, watch time, likes and per-day charts for each video and for the channel. The analytics help page shows where to find them, including a views-by-hour chart. This guide explains what each number measures and how to compare videos of different ages and lengths fairly. It ends with a weekly review short enough that you will actually do it.
One metric, one question
The most useful habit is to name the question before looking at the number.
| Metric | What it measures | The question it answers | What it cannot tell you |
|---|---|---|---|
| Views | How often the video was watched | Did the title and thumbnail earn the click? | Whether anyone stayed |
| Watch time | Total minutes watched | Did the video hold people once they arrived? | Why they left, or where exactly |
| Likes | Explicit approval after watching | Did the ending leave people satisfied? | Anything about people who left early |
| Per-day chart | Views or watch time by calendar day | Did a change on a date change the trend? | Cause; traffic sources shift on their own |
Views: the pitch metric
The choice to click is made before seeing any of the video. That is why views are the metric for titles, thumbnails and distribution, and for nothing else. A video with high views and low watch time did not succeed. Its pitch succeeded and its content did not, which is a different problem with a different fix. The thumbnail guide covers the pitch side.
Views also depend on age. A video published yesterday has had one day; a video published in June has had a summer. Comparing raw totals across videos of different ages tells you mostly which is older. Compare views in the first seven days, or views per day since publishing, and the comparison becomes about the videos.
Watch time: the video metric
Watch time is the metric to protect. It reflects the choice viewers make every few seconds to keep going. That is the closest thing analytics offer to "was this good". A long video collects more minutes than a short one of the same quality. So divide watch time by views to get average minutes per view, then compare that to the video's length. A ten-minute video averaging six minutes per view is holding people. A four-minute video averaging one minute is not, even if both show similar totals.
When a video's average is low, the fix is almost always in the opening or in a mismatch between the pitch and the content. The subscribers guide describes the opening structure that holds attention, and the honesty rule for thumbnails is there for exactly this reason.
Likes: the ending metric
People like a video at the end, or not at all. So likes per view measures how the video finished for the people who stayed. It is a rough measure, since most satisfied viewers do nothing. Use it to compare endings between your own videos, not as an absolute score. A drop in likes per view on one video usually points to a rushed ending, a broken promise or a closing ask that irritated people.
The per-day chart: the experiment metric
Daily charts are where you find out whether a change did anything. Swap a thumbnail on a given date, and the views-per-day line before and after that date is your evidence. Publish on a new day of the week, and the shape of the first week tells you whether the audience followed. The chart is also where you see the long tail. A video whose daily views stay steady months after publishing is getting search or embed traffic. That is the kind of video worth making more of. If that traffic comes from other sites, the embedding guide shows how to make those embeds work well.
The caution is that traffic sources move on their own. A link from a popular page, a search ranking change or a slow week in your subject will shift the line without you doing anything. Treat a single before-and-after as suggestive, and look for the same effect across two or three videos before calling it real.
The fifteen-minute weekly review
Once a week, same day, with the studio open. Anything more frequent invites reacting to noise.
Look at the channel's per-day chart for the last four weeks
Is the trend flat, rising or falling? Note the direction, not the number. If a spike appears, find which video and which date caused it.
Rank the last ten videos by minutes per view relative to length
Write down what the top three have in common. That is the template for the next upload.
Rank the same ten by views in their first seven days
Write down what the top three share in title and thumbnail. If the two rankings disagree, the pitch and the content are pulling apart and the fix is to align them.
Check likes per view for anything unusual
A video that is normal on watch time and low on likes has an ending problem. Watch the last minute and see why.
Decide one change
One. Write it and the date in a note, so next week's per-day chart can be read against it.
Metrics that mislead when read alone
None of these is useless, but each is routinely asked the wrong question.
- Lifetime views. Mostly a function of age. Old mediocre videos beat new good ones every time.
- Subscriber count. A lagging number that follows watch time. Chasing it directly leads to asks in the wrong place, as the subscribers guide explains.
- Total watch time across the channel. Rises with more uploads regardless of quality. Use the per-video average instead.
- A single day's spike. Usually an external link. Interesting, but not a trend until it repeats.
- Comparisons with other channels. Different subjects, ages and audiences make the numbers incomparable; compare your videos with your videos.
Reading the numbers in the right order
If you take one habit from this guide, make it the order of questions. First, did people click, which is views normalised by age. Second, did they stay, which is minutes per view against length. Third, did they leave satisfied, which is likes per view. Fourth, did the thing I changed change anything, which is the per-day chart around a date. Asked in that order, the four numbers describe the whole journey from thumbnail to ending. Each one points at a specific part of the next video to improve. Asked out of order, they describe nothing. They then push you toward the worst choice a creator can make, which is to change everything at once. The same habits matter if you later join the Creator Program, where watch time from Premium members shapes each channel's share of the monthly pool.
Frequently asked questions
Which single metric should I care about most?
Minutes per view relative to video length. It is the closest thing to a measure of whether the video was good, and the other outcomes follow it. A misleading thumbnail can win views, and likes only come from people who stayed. Average watch time reflects the choice viewers make every few seconds to keep going.
How long should I wait before judging a new video?
Give it at least a week, so weekday and weekend patterns both appear. Compare its first seven days with the first seven days of earlier videos, not with their lifetime numbers. Videos that gain search or embed traffic can keep growing for months. So look at the per-day chart again after a month before deciding a video underperformed.
Why do my daily views change when I have not published anything?
Traffic sources move on their own. A link from another site, a search ranking change, a seasonal shift in interest or an embed on a page that got popular can all move the line. That is why the per-day chart should be read against dated changes you made, and why one spike or dip is not a trend until it repeats.
Do unlisted videos count towards my channel analytics?
They are reported in the studio like any other video. But their views come only from people who have the link, since unlisted videos are not on the channel page, are noindex and are absent from sitemaps. Leave them out when ranking videos against each other. Otherwise a preview shared with two people will look like a failure.
Sources
- Google Search Central: video best practices — Why some videos keep receiving search traffic long after publishing.
- Google Search Central: creating helpful, reliable, people-first content — The audience-first test that watch time approximates.
- MDN: HTMLMediaElement — The player events that analytics such as playback start and progress are built on.
Related guides
How to grow subscribers
Subscribers are what happens after a video delivers. The levers that produce them, in the order they pay off, and why distribution comes last.
Video thumbnails that get clicks
The thumbnail is the pitch. How to compose one that reads at a glance, when text helps and when it hurts, and how to use VideoBB's generated candidates and custom uploads.
Video monetization models
Five ways a video channel earns, how each scales with audience size and trust, what each demands of you, and how creators combine them.