YouTube measures growth against your own channel
Two YouTube surfaces compare a video to your own recent uploads rather than to the platform. Build the baseline that makes those comparisons readable.
The most common analytics question a small channel asks is "is 3,000 views good?" It has no answer, and chasing one is where most channels lose a year. The number that would make it answerable is not published, does not exist as a single value, and would not apply to you if it did.
YouTube has said the useful part out loud in two places, and both point at the same reference frame: your own channel's recent results. Neither is a full description of how recommendation works, and anyone presenting them that way is overreaching. But together they are enough to fix how you read a dashboard.
The two places YouTube says it
The trending charts methodology. YouTube's page on Trending Charts lists the signals the charts combine. Among them, alongside raw view count: how quickly a video's view count (or its "temperature") is growing, where views are coming from including outside YouTube, the topic, the age of the video, and how the video performs compared to other recent uploads from the same channel. The page also notes that the video with the highest view count on a given day may not be number one.
That is a charts methodology, not the recommendation system. Say that clearly, because a lot of writing about YouTube launders this page into claims it does not make. What it does establish is that YouTube builds "compared to this channel's recent uploads" as a first-class signal in at least one production system, and that raw view count is not the deciding term even in the surface whose entire purpose is ranking by popularity.
YouTube Studio's own comparison baseline. Open the Content tab for a Short and the metrics arrive pre-framed. Views are reported compared to typical performance. Subscribers are reported compared to typical performance. Not compared to a category median, not compared to channels your size. Typical performance means yours.
You did not choose that comparison. YouTube built it, computed it for you, and put it on the default view. That is a strong statement about which comparison it considers meaningful.
Why this favours a small channel
If the useful signal were absolute view count, a channel with 800 subscribers would be competing directly against channels with two million for the same measurement. It would lose permanently and the loss would carry no information.
A channel-relative frame does not work that way. A video that beats your last ten by a wide margin is a strong result on a channel of any size, and the strength of the result does not shrink because someone else's channel is larger. The comparison is local by construction, which is why the practical advice for the first thousand subscribers is about finding a repeatable format rather than about hitting a view number.
This also explains an experience most creators have had and misread: a channel that grows steadily for months, then plateaus after one unusually large video. The plateau is often not a decline in quality. It is the baseline moving. Every subsequent upload is now being read against a higher recent average, including by you.
Building a baseline you can actually read
Studio computes its own, and the comparison it shows you is worth taking seriously. But it does not segment the way your production does, so it is worth maintaining a second one.
Four rules make a baseline usable.
1. Use the median, not the mean. One outlier video distorts a mean permanently and a median barely at all. Since the whole point of the exercise is detecting outliers, an average that has already absorbed them is the wrong instrument.
2. Roll it, and keep the window short. Last 10 to 20 uploads of the same type. Long enough to be stable, short enough that it tracks what your channel is currently doing rather than what it did last year. A baseline that includes a format you abandoned is measuring a channel that no longer exists.
3. Segment by format, not by date. Shorts and long-form belong in separate baselines. So do genuinely different Shorts formats. If you run three formats, run three baselines, because a mixed baseline means every video is being compared against a blend that describes none of them.
4. Set a minimum n and respect it. Below roughly ten samples a median is noise wearing a number's clothes. A new format has no baseline yet. That is a real state and it is better to say "no baseline" than to compute one from four videos and treat it as fact.
What each deviation tells you
Once the baseline exists, individual metrics become diagnostic rather than decorative. Read them one at a time against your own median.
| Metric | Reading against baseline | What it usually means |
|---|---|---|
| Shown in feed | Below median | Distribution never started; look at hook and eligibility, not content quality |
| Viewed vs swiped away | Below median | Packaging or opening frame failed at the gate |
| Views | Below median, feed impressions normal | The video was offered and declined |
| Views | Above median, retention normal | The topic or packaging landed; repeat the input, not the video |
| Retention curve shape | Deviates at a timestamp | Segment-level problem; regenerate that segment |
| Subscribers | Below median at normal views | The video worked and did not argue for the channel |
The middle rows matter most on Shorts, where distribution and selection are separately visible. Studio's How many chose to view metric — viewed versus swiped away — is the first-order number in that chain, and it is the one that tells you whether a weak video failed at the hook or failed later.
For long-form, the equivalent segment-level instrument is the retention curve, and reading its four shapes against your own median is the same discipline applied to a different chart.
What breaks a baseline
Four things, all of them common:
- A format change. The baseline is now describing a channel you no longer run. Start a new one; do not blend.
- One breakout video. It pulls the median less than a mean, but at small n it still moves it. Note it, keep it in the record, and expect a period of everything reading as "below typical".
- Seasonality and schedule gaps. A month off resets audience habit. The first video back is not a fair sample.
- Chasing the number instead of the input. A baseline is a measurement instrument, not a target. Optimising for "beats my median" produces safe videos that beat a median and never move it.
That last one is the failure mode worth watching for, because it looks like discipline. A baseline exists to tell you which experiments worked. If it starts telling you which experiments to avoid running, it has been inverted.
The one thing that is not channel-relative
Worth naming the exception, because conflating the two causes real confusion. Performance signals are read against your own channel. Revenue is not.
Shorts revenue is allocated as a share of a pooled amount, based on your proportion of total engaged views from monetizing creators within each country. That is explicitly relative to everyone else. So a month where your videos beat your own baseline and your revenue fell is not a contradiction: you improved against yourself and shrank as a fraction of the pool. How the Shorts pool splits has the mechanism.
Keeping those two frames apart is most of what separates a creator who can read a dashboard from one who cannot. Performance, local. Revenue, global. Different questions, different denominators, and a single "how am I doing" number that tries to answer both will always be wrong.
The practical upshot: the only benchmark worth maintaining is one you built from your own last twenty uploads, segmented by format, on a rolling window, using a median. Everything else is someone else's channel. Combined with a sane tracking list, it turns "is 3,000 views good" into a question with an answer, which is the whole point.
FAQ
Does "compared to typical performance" in Studio mean compared to my channel?
That is the comparison Studio presents by default on the Content tab, for Views and for Subscribers. It is YouTube's own computed reference, and its existence on the default view is a reasonable signal about which comparison YouTube considers meaningful. It will not segment by your internal format definitions, which is why a hand-maintained baseline alongside it is worth the effort.
Does channel-relative measurement mean small channels get pushed harder?
No, and that would be the wrong conclusion to draw. What the published material supports is narrower: performance relative to a channel's own recent uploads is one signal among several in the trending charts, and Studio frames your metrics against your own typical performance. That is a statement about measurement, not about promotion. It does mean absolute view count is a poor self-assessment tool at any size.
How many videos before a baseline is trustworthy?
Roughly ten of the same format, and twenty is better. Below ten, a median moves too much with each new sample to distinguish a real result from variance. A brand-new format genuinely has no baseline, and treating that as a temporary state produces better decisions than computing one from four videos.
Should I compare my numbers to other channels in my niche at all?
For deciding what to make, sometimes; for deciding whether a video worked, no. You cannot see their impressions, their audience geography, their traffic sources or their history, so their public view count is not a comparable measurement. Your own last twenty uploads are the only sample where you control for all of that.