Understanding analytics
Starting Your YouTube Channel
Course 1 · Chapter 7 · Understanding Analytics
YouTube Studio gives you access to more data than you will ever need. Most of it, in the early months of a channel, is noise. Acting on small-sample data produces bad decisions — pivoting your entire content strategy because three videos underperformed is how creators destroy momentum before they've given themselves a fair chance. This chapter is about what to look at, what to ignore, and when you actually have enough data to act on what you're seeing.
The Metrics That Actually Matter
There are dozens of numbers in YouTube Studio. For a channel under 1,000 subscribers, only a small handful are actionable. Everything else is context at best, distraction at worst.
Focus on these
Click-Through Rate (CTR)
Analytics > Reach > Impressions click-through rate
The percentage of people who see your thumbnail in the feed and click on it. This is a direct measure of how compelling your thumbnail and title combination is. Low CTR means people see your video and scroll past — the content is irrelevant until the packaging works.
Benchmark: 2–10% for most channels. Under 2% suggests thumbnail/title problems. Over 10% is exceptional and usually unsustainable at scale.
Average View Duration (AVD)
Analytics > Engagement > Average view duration
How long, on average, a viewer watches your video. This is the primary signal YouTube uses to judge whether viewers found what they came for. A short video with high AVD beats a long video with low AVD every time.
Benchmark: 40–50% of video length is solid. Under 30% suggests the hook or content isn't delivering on the title's promise.
Audience Retention Curve
Analytics > individual video > Engagement > Audience retention
Shows the percentage of viewers still watching at each moment in your video. Unlike AVD which gives an average, the curve shows where people leave — and that location tells you what caused it. This is your most precise editorial feedback tool.
What to look for: Cliffs (sudden drops) = a specific moment that lost viewers. Gradual slope = normal decay. Bumps upward = a segment people rewatched.
Subscriber Growth per Video
Analytics > individual video > Reach > Subscribers
How many subscribers each video gained (or lost). This tells you which content resonates with your target audience enough to make them commit. Viral traffic with no subscriber gain means you reached the wrong people.
Watch for: Videos that over- or under-perform on subscribers relative to views. Those outliers tell you what your audience actually wants from you.
Useful context — read carefully
Impressions
Analytics > Reach > Impressions
How many times your thumbnail was shown. Impressions alone mean nothing — a million impressions with 1% CTR is worse than 10,000 impressions with 8% CTR. Use this alongside CTR, never in isolation. New channels get low impressions by default; this improves as the algorithm gains confidence in your content.
Traffic Sources
Analytics > Reach > Traffic source types
Where your views are coming from — YouTube Search, Suggested Videos, Browse Features (homepage), External. The mix matters: search traffic is durable and grows over time. Browse/Suggested traffic spikes then fades. A channel built on search has more stable long-term growth.
Watch Time (hours)
Analytics > Overview
Total hours watched across all videos. YouTube historically used this for monetisation eligibility (4,000 hours). As a growth metric it matters less than AVD — 4,000 hours from the wrong audience watching the wrong content does not help the algorithm understand who to show your videos to.
Mostly ignore in year one
Likes, Comments, Shares
Analytics > Engagement
Engagement metrics are feel-good indicators but weak algorithmic signals. YouTube's algorithm is powered primarily by watch behaviour, not likes. A video with 5 likes and 45% retention is doing better than a video with 500 likes and 18% retention. Don't optimise for comments by asking viewers to comment.
Revenue & RPM
Analytics > Revenue
Irrelevant until you're in the YouTube Partner Programme (1,000 subscribers + 4,000 watch hours). Even then, RPM varies enormously by niche and season. Looking at £1.40 from your first three monetised videos will either mislead you about potential or demoralise you unnecessarily. Covered properly in Course 3.
CTR in Depth — Reading the Number Correctly
CTR is frequently misread. The most important context: CTR is highly dependent on where your video is being shown. Homepage impressions have lower CTR than search impressions. A new channel's CTR tends to be higher than an established channel's simply because YouTube shows new videos to a smaller, more targeted audience first. As your reach expands, CTR typically falls — that is normal, not failure.
CTR interpretation guide — varies by niche, placement, and channel age
<2%
Thumbnail or title is the problem — viewers aren't compelled to click
2–4%
Below average — room for improvement in packaging
4–7%
Healthy — solid packaging, algorithm will promote this
7–10%
Strong — thumbnail and title working well together
>10%
Exceptional — note the topic and format, replicate it
CTR without retention is a trap
A misleading title can produce a high CTR and terrible retention — viewers click, find the content doesn't match the promise, and leave immediately. YouTube penalises this. The combination of good CTR and good retention is what signals a genuinely successful video. If your CTR is high but retention is low, your title is overpromising what the content delivers.
Average View Duration — What the Percentages Mean
AVD percentage scenarios — same 12-minute video, different outcomes
AVD: 2:10 = 18%
Poor — hook is failing or content doesn't match title
AVD: 3:36 = 30%
Below average — hook works, but content loses them early
AVD: 4:48 = 40%
Good — solid retention, algorithm will promote this
AVD: 6:00 = 50%
Excellent — strong signal, expect increased distribution
Note that absolute watch time (minutes watched) matters too — a 50% retention on a 20-minute video is more valuable to YouTube than 50% on a 2-minute video. This is one reason mid-length videos (8–15 minutes) tend to perform well: they generate meaningful watch time while still being achievable retention targets.
Traffic Sources — Understanding Where Viewers Come From
Most valuable long-term
YouTube Search
Viewers actively looking for your topic. High intent, high retention, tends to grow steadily over time as your video ages and accumulates SEO history.
If low: titles aren't matching search queries. Research keywords more carefully (Chapter 1 of Course 3).
High upside, volatile
Suggested Videos
YouTube surfaces your video alongside a related video someone is already watching. Traffic spikes quickly and fades. Depends heavily on the algorithm finding your content relevant to established videos.
If dominant: you're benefiting from someone else's audience. Good short-term, but don't rely on it.
Algorithmic — earned
Browse Features (Homepage)
YouTube recommending your video on subscribers' home feeds. This source grows as your subscriber base grows. Requires consistent posting so the algorithm keeps your channel in rotation.
If low early on: normal — you need more subscribers before this traffic source is significant.
You control this
External
Traffic from links outside YouTube — social media, blogs, newsletters, Reddit. Tends to have lower retention than search traffic since the viewer wasn't actively looking for your content.
If promoting on social: don't judge video quality by this traffic. It skews AVD down due to casual visitors.
Using Analytics to Make Decisions
Analytics are only useful when they point to a specific action. The table below maps the most common diagnostic patterns to their likely causes and fixes.
| What you observe | Likely cause | What to do |
|---|---|---|
| Low CTR (<2%) on every video | Thumbnail design or title writing is the bottleneck | Study thumbnails of top videos in your niche. Redesign — brighter colours, clearer focal point, larger text. Test new title formats. |
| Good CTR, low AVD (<30%) | Title over-promises; hook doesn't land; opening is slow | Rewatch your first 60 seconds. Cut everything that isn't the hook. Rewrite the hook to match what the video actually delivers. |
| Steep cliff at a specific timestamp | A single segment that fails — too long, off-topic, or confusing | Watch the video at that timestamp. Identify what you were saying. That type of content needs to be cut or restructured in future videos. |
| High views, zero subscriber growth | Content attracted the wrong audience or no CTA | Check if the video topic is truly in your niche. Add a specific, contextual subscribe CTA at the end. Link to a second related video. |
| Traffic almost entirely from search | Evergreen content working well; algorithm not yet surfacing you | Good sign long-term. Keep producing search-optimised content. Browse traffic grows with subscriber count — be patient. |
| One video dramatically outperforms all others | Topic, format, or thumbnail/title combination struck a chord | Study what made it different. Make more videos with the same topic angle, similar thumbnail style, and comparable title structure. |
Where to Find Everything in YouTube Studio
YouTube Studio navigation reference — studio.youtube.com
Analytics > Overview
High-level view: views, watch time, subscribers, revenue — for any date range
Analytics > Reach
Impressions, CTR, unique viewers, traffic sources breakdown
Analytics > Engagement
Average view duration, top videos by watch time, likes, cards
Analytics > Audience
Returning vs new viewers, subscriber geography, age/gender, when they're online
Content > [video] > Analytics
Per-video metrics — same tabs as channel-level but scoped to one video
Content > [video] > Analytics > Engagement > Audience retention
The retention curve — the most useful single screen in YouTube Studio
Analytics > Reach > Traffic source: YouTube search
Exact search terms that brought viewers to your videos — free keyword data
Free keyword intelligence
The search terms report (Analytics > Reach > Traffic source: YouTube search) shows you the exact phrases people used to find your videos. This is some of the most valuable data available to you — and it's free. These are real queries from real viewers, not estimated keyword volumes. Use them to title future videos and find related topics to cover.
When You Have Enough Data to Act
This is the part most creators get wrong. They see one video underperform and rewrite their entire strategy. The problem: one video is not a sample size. A video needs at minimum 500–1,000 views before its metrics stabilise enough to draw conclusions. For a small channel, that can take weeks or months.
The rules of thumb:
- Under 5 videos: don't read anything into the numbers. There's no pattern to find. Just make the next video.
- 5–20 videos: look at averages across videos, not individual performance. Identify the two or three videos that clearly over- or underperformed and ask what was different.
- 20+ videos: you have enough data to identify patterns — which topics, formats, thumbnail styles, and title structures consistently outperform. Start optimising based on these patterns.
- Under 48 hours: never judge a video's performance in the first two days. YouTube is still distributing it to test audiences. A slow start often recovers to become a strong long-term performer, especially for search-driven content.
The only early metric worth watching weekly
In the first three months, check only one thing weekly: are your retention curves improving or staying flat? If the curve is gradually improving — even a few percentage points — your content is getting better. Everything else is noise until you have the sample size to act on it.