How to read YouTube analytics: the 7 numbers that actually matter

YouTube Studio shows you dozens of charts. Most creators glance at views, see they're low, and feel bad. That's not analytics — that's anxiety. Here are the seven numbers that actually matter, what each one tells you, and the threshold where you should care.

The 7 numbers

1. Impressions

How many times your thumbnail was on screen for at least 1 second with at least 50% visibility. Impressions are opportunity— the algorithm decided to show you to someone. Low impressions means the algorithm doesn't know who to show your video to. That's almost always a topic-clarity problem, not a quality problem.

Where it lives: Reach tab.

Healthy signal: Impressions trending up over a rolling 28 days, with each new video receiving more impressions than the previous one in the same niche.

2. Click-through rate (CTR)

Of viewers who saw the thumbnail, what percent clicked? CTR varies wildly across niches — see our dedicated CTR guide for benchmarks. Channel-average CTR is usually between 4% and 10%. Below 4% is generally weak; above 10% is strong.

Where it lives: Reach tab.

Healthy signal: CTR holding steady or trending up over time. Falling CTR on new uploads usually means your title and thumbnail patterns have gone stale.

3. Average percentage viewed (APV)

The percentage of the video the average viewer watches. APV is the closest single signal to whether viewers actually like your video. Length-normalized — a 50% APV on a 5-minute video is worth more than 25% APV on a 10-minute video, even though they have the same total watch time, because retention scales differently.

Where it lives: Engagement tab.

Healthy signal: APV above 35% for tutorials, 30% for long-form interviews, 50%+ for entertainment shorts.

4. Average view duration (AVD)

APV in absolute seconds. The combination of AVD and APV tells you more than either alone. A video with high APV but low AVD (because it's short) signals you should make longer videos. A video with low APV but high AVD (because it's long) signals the inverse — people are committed but the length is killing them.

Where it lives: Engagement tab.

5. Audience retention curve

The single most actionable chart in YouTube Studio. Open any video, click Analytics > Engagement > Key moments for audience retention. The curve shows what percent of viewers are still watching at every second of the video.

Read the curve like this:

• The first cliff (typically before 0:30).Where the steep drop starts. That's your hook problem. Trim everything before that point.
• Spikes upward. Re-watches. Whatever you did at that moment, viewers wanted to see again. Do more of it.
• Flat sections.Viewer attention is held steady. That's gold — replicate that pacing.
• Late-video cliffs. Viewers bailed before the end. Either the conclusion is weak or the topic resolved before the video ended. Cut earlier next time.

Where it lives:Per-video analytics > Engagement.

6. Returning viewers

Percent of your viewers in the last 28 days who've watched your channel before. This is the leading indicator of channel-level health. New viewers are easy; getting them to come back is the actual game.

Where it lives: Audience tab.

Healthy signal:Returning viewers as a percentage of total trending up. A growing channel with falling returning-viewer share is leaking — new viewers aren't sticking. Fix consistency before pushing for more reach.

7. Subscriber-driven views

What percent of your views come from existing subscribers vs non-subscribers. A healthy growing channel has roughly 20–40% of views from subscribers and 60–80% from non-subscribers. Below 20% from subscribers means you're not building a base. Above 50% means you've plateaued — your content isn't reaching new audiences anymore.

Where it lives:Audience tab > Watch time from subscribers.

How to actually use these numbers

Weekly review (15 minutes)

1. Open Studio. Set the date range to last 28 days.
2. Check impressions, CTR, APV trend lines. Are they up, flat, or down vs the previous 28?
3. Click into the top-performing video. Read its retention curve. Note one thing it did right.
4. Click into the worst-performing video. Read its retention curve. Note where it cliffed.
5. Write down one experiment for the next video based on those two notes. Just one.

Per-video review (after 14 days from upload)

1. Compare CTR to your channel median. Above? Title and thumbnail worked.
2. Compare APV to your channel median. Above? Hook and pacing worked.
3. Look at traffic sources. Where did views come from? If 70%+ are from Suggested, you piggybacked on an existing video. If 70%+ are from Search, you ranked for a query.
4. Note the one number that under-performed. That's the lever to pull next time.

Numbers to ignore

• Real-time analytics. Too noisy. The 60-minute chart is for confirming a video uploaded.
• Likes-to-dislikes ratio. Public dislikes are hidden anyway, and the signal is heavily lagged.
• Comments per video.A weak proxy for engagement. Highly variable; don't use as a primary metric.
• Subscribers gained per day. Too lumpy. Look at subscriber growth as a 28-day trend, not a daily number.
• Revenue.A lagging indicator that depends on RPM fluctuations you don't control. Useful at end of month, not weekly.

The single biggest analytics mistake

Comparing your numbers to global benchmarks. There is no global benchmark that matters. CTR varies 4× across niches, RPMs vary 30×, retention varies wildly with content type. The only meaningful comparison is your channel against itself, last week vs this week, last month vs this month. When you compare to other people's screenshots on Twitter, you're comparing against either a wildly different niche or a vanity stat from one of their best weeks.

Related reading

• How to improve YouTube CTR
• How the YouTube algorithm actually works in 2026
• The YouTube SEO guide for 2026