The only official range YouTube ever published: 2% to 10%
YouTube's Help Center publishes exactly one benchmark for click-through rate, and it is deliberately wide: half of all channels and videos have an impressions click-through rate somewhere between 2% and 10%. That single sentence answers the question most people are actually asking. A 4% CTR is inside the normal half. It is not good and it is not bad. It is normal.
YouTube attaches three caveats to that range, and they matter more than the range itself. New videos and new channels, less than a week old or under 100 views, swing far wider in both directions. A video that collects a lot of impressions, for example because it got pushed to the Home feed, will naturally show a lower CTR. And impressions that come from your own channel page tend to show a higher rate, because the person was already looking at you.
Read those caveats together and you get the uncomfortable truth about benchmarks: 2% and 10% are both officially normal, and they are five times apart. A five times gap between two normal numbers is not a target. It is a shrug. Any blog post that tells you the good CTR is 6.4% invented the decimal to sound authoritative, and the number will be wrong for your traffic mix in either direction.
So the useful question is not what number is good. It is which of your numbers moved, against which of your own numbers, and what that specific movement tells you to change. That is a different discipline, and it is the one the five metrics that actually decide the next video are built for.
Why a benchmark from another channel is the wrong ruler: a 4% channel can beat a 6% one
Click-through rate is not a quality score. It is a ratio between an impression and a click, and you do not choose the impression. YouTube does. Change where the impressions come from and the same thumbnail produces a completely different percentage without a single pixel moving.
Take two channels in the same niche. Channel A shows a 4% CTR and gets 90% of its impressions from Home. Channel B shows 6% and gets 90% of its impressions from its own channel page and subscriber feed. Channel A is winning by a distance. It is being handed cold traffic at volume and converting it. Channel B is converting people who already subscribed, which is a much easier sale, and it is not reaching anybody new. The same trap sits one level earlier, when there is barely any impression to divide by at all, and why a video gets almost no impressions separates the counting rule from the surfaces that never register.
This is where a large share of channels break something that was never broken. They see a channel average below the benchmark they read somewhere, redesign a thumbnail that was performing well against cold traffic, and lose the one thing that was working. The fix is procedural, not creative: open the traffic source report, read CTR per source, and only compare a source against the same source.
It is the same discipline that separates a real drop from normal variance. When views fall, the shape of the fall and which number moved first decide the diagnosis, which is why the seven real causes of a sudden view drop are read in a fixed order instead of guessed at.
The pattern is consistent enough to plan around. Rough ranges by traffic source, and the reason each one behaves the way it does:
- Channel page and subscriber feed: the highest rates, often 8% to 20%. The audience is warm, already looking for you, and often already decided before the thumbnail loaded.
- Search: middle, roughly 4% to 10%. Intent already exists, so the thumbnail only has to confirm that you answer the query the viewer typed.
- Suggested video: middle to low, roughly 2% to 8%. You are competing against the video the person is already watching and against nine other sidebar options.
- Browse features and Home: the lowest, often 1% to 4%. YouTube is guessing at volume, showing you to people who never asked for you. A low rate here is the price of reach, not a verdict on your packaging.
- External and the Shorts feed: not comparable at all. The Shorts feed has no thumbnail decision in it, so a CTR from there does not measure the same thing.
Percentage viewed: the brackets that move with length, not with quality
For average percentage viewed, YouTube publishes no official benchmark at all, so everything in circulation is market observation. The published brackets do converge, and they converge on length rather than on niche: roughly 65% to 75% for videos under 5 minutes, 50% to 60% for 5 to 10 minutes, 40% to 50% for 10 to 15 minutes, and 35% to 45% above 15 minutes.
Then the contradictions start. One 2025 benchmark report puts the average YouTube video at 23.7% percentage viewed. Other 2026 roundups put the platform average at 35% to 45%. Both numbers can be defended, because one weights every video equally, including the millions that nobody watched, and the other weights by views or by established channels. The practical rule is to treat any single published platform average as noise and treat the length bracket as the ruler.
The mechanics behind the brackets are simple and they protect you from a bad decision. Percentage viewed falls as length rises, but total watch time usually rises with it. A 30 minute video at 35% delivers 10.5 minutes per view. A 5 minute video at 70% delivers 3.5. The video with the uglier percentage produced three times the watch time, and watch time is the currency that the Partner Program counts.
When a percentage viewed sits below your own bracket, it slips in one of two predictable places: the first 30 seconds, where a slow or dishonest hook loses 20% to 40% of everyone who clicked, and the 2 minute mark, where the video finishes delivering the promise in the title and gives the viewer no reason to stay. Inside FalconVid both of those are cheap repairs rather than a regeneration: rewriting the script costs 307 credits and renarrating costs 36 credits in economy mode or 288 in premium, against 1,008 to 26,760 credits to rebuild the whole video.
In 2027 the currency is the hour, so percentage viewed has a price tag
This is the year the abstract number becomes money. From 1 February 2027, channels entering the Partner Program need 1,000 subscribers plus 8,000 public qualified watch hours in 365 days, double the old 4,000, or 1,000 subscribers plus 20 million qualified Shorts views in 90 days, double the old 10 million. Channels already in the program are not held to the new bar, but they do have to accept the updated terms in YouTube Studio by 31 January 2027 or lose monetization on 1 February.
Now put percentage viewed into that equation. 8,000 hours is 480,000 minutes. A 12 minute video at 40% percentage viewed delivers 4.8 minutes per view, so the bar costs 100,000 views. At 50% it costs 80,000 views. At 30% it costs 133,333. Ten points of percentage viewed is worth tens of thousands of views, and that is the only honest reason to care about the metric at all.
Volume is the other lever and it is the one you control directly. At 1,000 views per video on a 12 minute video at 40%, each publication banks 80 watch hours, so 8,000 hours is 100 videos. A hundred videos in economy mode cost 100,800 credits, about US$ 316 at the Starter credit value, which is the Business plan at US$ 297 with 95,000 credits a month plus one pack, or the Agency plan at US$ 597 with 190,000 a month and room to spare.
Engagement and subscribers per view: the two numbers people read backwards
Engagement rate is likes plus comments divided by views. The market bands are stable: above 5% is excellent, 2.5% to 5% is good, 1% to 2.5% is average, and below 1% means people are watching passively without interacting. The trap is that the bands move with channel size. Channels under 1,000 subscribers often sit near 8%, channels between 1,000 and 10,000 near 5.2%, and channels between 100,000 and 500,000 near 2.1%.
That means a falling engagement rate as your channel grows is arithmetic, not decay. Your early audience was self selected and vocal. Growth dilutes it with people who found you from Home. Reporting that as a decline is one of the most common false alarms in the whole dashboard. Formats matter the same way: Shorts run 5% to 15% and long form runs 2% to 5%, so a channel doing both should never average the two into one number.
Subscribers per view is the number most often read backwards. Common conversion is 0.01% to 0.05% of viewers subscribing, and strong content reaches 0.1% to 0.5%. Expressed as a ratio, healthy channels sit somewhere around one subscriber for every 50 to 100 views per video. Above and below those bands both carry meaning, and the subscriber to view ratio read by bands is the cleanest way to tell a channel with a loyal core from a channel that was carried by one lucky video.
Build your own benchmark in 20 minutes: median of the last 10, never the average
Every number above is only there to stop you from panicking. The ruler that actually decides anything is yours, and building it takes one sitting. Use the median of your last 10 videos, never the average. A single outlier video moves an average by 30% or 40% and tells you nothing about what the next publication will do.
Write down three medians and nothing else. First, CTR per traffic source for the last 10, not the channel average. Second, percentage viewed for the last 10 inside the length band you actually publish. Third, views per day at day 7, because that is the number that separates a slow video from a dead one before the 28 day report exists.
Then apply one threshold: a number is a signal when it moves more than 20% away from your own median and stays there across three publications. Anything smaller than that is noise, and reacting to noise means changing thumbnail, title, length and cadence at the same time, which destroys your ability to learn anything from the result.
FalconVid keeps that ruler assembled for you across every channel you run, with analytics for all of them in one panel and an alert when a video breaks out of your own pattern instead of somebody else's benchmark. If you want the full sweep rather than three numbers, the 32 point channel audit checklist is the same exercise applied to packaging, retention, catalog, publishing and monetization.
The part nobody does: somebody has to read these numbers every 2 days
Knowing the benchmarks is the easy half. The hard half is that the reading has to happen on a schedule, forever, while you are also producing. This is why standalone channel audits exist as a paid service, and they are not cheap: a consultant charges US$ 300 to US$ 3,000 for a one off report, a freelancer charges US$ 25 to US$ 100 an hour, and an agency buries it inside an entry retainer of US$ 2,000 to US$ 4,000 a month.
Worse, the report ages the moment it is delivered. At four videos a week, three weeks after the audit there are 12 videos the report never saw, produced under advice written for a channel that no longer exists in that shape.
That is the gap the Senior AI Analyst fills. It is a fixed person per client, with a name, a face and a voice, who reads your account and writes to you every 2 days, in your language, with the specific move to make inside the product rather than a generic report. It comes on every plan from Pro upward, and Starter gets a 7 day trial of it, which is usually enough to see whether having somebody read the numbers changes what you publish. The AI analyst that reads your channel for you is the page that shows what those messages actually look like.
The other half of the gap is production, because diagnosis costs you nothing if the channel keeps moving while you diagnose, and costs you everything if it stops. FalconVid researches, writes, narrates, edits, captions and publishes to YouTube, Instagram, TikTok, Rumble and Facebook in up to 63 languages, from a calendar you approve once, with AI specialists working in parallel and a finished video in up to 30 minutes.
The decision table: what each combination of numbers tells you to change
Benchmarks are only useful if they end in an action. Here is the table that turns two or three medians into one change, with the price of each change so you never pay for a regeneration when a repair would do. Every price below comes from the same engine that runs the pipeline behind an approved calendar, which is why a repair and a rebuild have different bills instead of both meaning start over.
- CTR below your median, impressions flat: this is packaging. Title and thumbnail, nothing else. Redesigning a cover costs 479 credits in economy mode and 214 in premium, against 1,008 to 26,760 to regenerate the video, so you can test three or four covers for a fraction of one rebuild.
- CTR normal, percentage viewed below your median: this is the hook and the pacing. Rewriting the script costs 307 credits and renarrating costs 36 in economy or 288 in premium. Fix the first 30 seconds and the 2 minute handoff before you touch anything else.
- Both normal, views flat: this is demand, not craft. The video is fine and the topic has no audience at that size. Change the subject, not the video.
- Impressions collapsed while CTR and percentage viewed held: the algorithm stopped testing this one. Publish the next video, do not renovate this one. Nothing you edit brings back impressions that were never offered.
- Everything normal and revenue still low: it is RPM, country mix or format, not the algorithm. That is a monetization problem with its own levers, and no thumbnail will move it.
- Fewer than 10 published videos: you do not have a median yet, so you do not have a signal. Publish, keep the format stable, and read the numbers once the sample exists. This is the single most common reason a channel changes everything and learns nothing.