The eight
| Metric | Question it answers | Act when |
|---|---|---|
| Impressions | Is YouTube offering my video to anyone? | Flat or falling month over month |
| Click-through rate | Does the packaging work? | Below 4% with healthy impressions |
| Average view duration | Does the video hold people? | Below 40% of length on long-form |
| Average percentage viewed | Is length appropriate? | Long videos with low percentage |
| Traffic sources | Where do views come from? | Over-reliance on one source |
| Returning viewers | Is the channel building an audience? | Flat while views grow |
| Subscribers per video | Does the format convert? | Very low relative to views |
| RPM | What is a view worth? | Sudden unexplained drop |
Impressions and the impression funnel
An impression is your thumbnail being shown on YouTube — home page, suggested column, search results, subscriptions. It is not a view. Studio's impressions funnel shows how many impressions became clicks, and how many clicks became watch time. Read it top to bottom: the first stage that is weak is the only one worth working on.
Click-through rate, read correctly
CTR is not a score to maximise in isolation. A brand-new video shown mostly to subscribers might see 15% CTR on 900 impressions. The same video shown to non-subscribers on the home page might drop to 3% on 400,000 impressions — and that second scenario is vastly better for the channel.
- Always read CTR alongside impressions. High CTR on tiny impressions means a small closed audience.
- CTR declines naturally as a video reaches beyond your subscriber base. That is a good sign.
- Compare against your own channel's videos in the same format, not against generic benchmarks.
Work backwards from a target view count to the CTR you would need.
Free CTR calculatorAverage view duration vs average percentage viewed
AVD is time in minutes; APV is that time as a share of the video's length. Both matter but they answer different questions. A 20-minute video with 6 minutes AVD (30%) delivers more total watch time than a 5-minute video at 60% — but the 30% figure tells you the video is probably too long for what it delivers.
Traffic sources
| Source | What it means | Health signal |
|---|---|---|
| Browse features | Home page recommendations | The strongest growth source |
| Suggested videos | Sidebar and next-up | Means the system is pairing you with related content |
| YouTube search | Query-driven | Compounds over years; most stable |
| Channel pages | Existing audience browsing | Fine, but not growth |
| External | Off-platform links and embeds | Useful, but does not train the algorithm much |
| Notifications | Bell and subscriptions feed | Declining share is normal |
Returning viewers: the metric nobody watches
In the Audience tab, new versus returning viewers is the clearest measure of whether you are building a channel or just publishing videos. Views can grow while returning viewers stay flat — that means every video is finding strangers who never come back, which is an unstable base.
Reading date ranges without fooling yourself
- Use 28 or 90 days for trends; 7 days is mostly noise
- Use medians rather than averages when comparing videos — one outlier distorts everything
- Compare like formats. A Short and a 25-minute essay cannot be judged on the same metrics
- Ignore the first 48 hours of any video's data for decision-making