A channel owner sent me a screenshot last month with a red circle around 3.1 percent and the caption "is this bad". The video was eleven months old, sitting on 240,000 lifetime impressions, and about 80 percent of those impressions had come from browse and suggested feeds where nobody was looking for it. That 3.1 percent was not bad. It was also not the number he should have been staring at.
CTR benchmarks are the most quoted and least understood stat on YouTube. The averages get repeated until they harden into rules, and then people burn weekends redesigning videos that were never the problem. So let me lay out what I actually use when I look at a back catalogue and decide what is worth repackaging.
The benchmark everyone quotes is a blended average
YouTube's own guidance says roughly half of channels sit somewhere between 2 and 10 percent click through rate on impressions. That range is so wide it is almost useless as a target, and there is a reason it is that wide: it blends every traffic source together.
A video getting most of its impressions from your subscriber home feed will look great. A video getting most of its impressions from suggested placements next to somebody else's content will look grim by comparison, because it is being shown to people who have never heard of you. Same video, same packaging, wildly different numbers depending on where the impressions came from.
The first thing I do with any old video is split impressions by source before I judge a single percentage point. YouTube's own analytics documentation covers how those sources are counted, and it is worth ten minutes of reading before you decide your thumbnail is broken.
What is a good CTR for an old video?
For an old video, good means the number is holding steady or climbing while impressions grow. The absolute figure matters far less than the direction it moves as the audience widens.
Here is roughly how I read old video numbers, by dominant traffic source:
- Search dominant (people typed something and found you): 4 to 8 percent is healthy. Intent is high, so a weak title hurts more than a weak thumbnail here.
- Suggested dominant: 2 to 5 percent is normal and 6 percent plus is genuinely strong. You are competing in a sidebar against people with bigger channels.
- Browse and home dominant: 5 to 12 percent, because a good chunk of that audience already subscribed to you.
- External and shared: ignore CTR entirely. The number is noise.
A video at 3 percent that is 85 percent suggested traffic is doing fine. A video at 5 percent that is 90 percent subscriber home feed is quietly underperforming, because your own audience should click you more than that.
If you only remember one thing: CTR is a ratio, and both halves of it move. A falling percentage on rising impressions is usually success, not failure.
Why does an old video's CTR look worse than it is?
Because success itself pushes the number down. When YouTube decides an older video deserves a wider test, it shows it to colder and colder audiences, and colder audiences click less. The percentage sags while total views climb.
I have watched videos drop from 7 percent to 4 percent while their weekly views tripled. If you panic at the percentage and change the packaging in that window, you interrupt a distribution test that was going in your favour and you lose the read entirely.
The reverse trap exists too. A video with 900 lifetime impressions showing 11 percent CTR is not a hit, it is a rounding error. I do not treat any CTR figure as real under about 2,000 impressions in the period I am measuring, and I prefer 5,000 before I make a decision worth money.
There is also seasonality, competitor uploads on the same topic, and the plain fact that a video about a tool that changed its interface last spring is now offering something nobody wants. No thumbnail fixes obsolescence. Some old videos deserve to stay where they are, and pretending otherwise is how people waste a repackaging budget.
How do I know if a repackage actually worked?
You compare the same traffic source across a matched window, before and after, and you wait at least 14 days. Anything faster is guessing.
The method I use is deliberately boring:
- Note the 28 day impressions, CTR, and views before the change, split by source.
- Change one thing at a time where you can. Thumbnail, or title, or both if the video is dead anyway and you have nothing to lose.
- Wait 14 to 28 days without touching it again.
- Compare like for like. Suggested against suggested, search against search.
What counts as a win depends on where you started. Moving a video from 2.4 percent to 3.1 percent sounds tiny and is actually a 29 percent lift in clicks on the same impressions, which on a video with 300,000 annual impressions is thousands of extra views for one afternoon of work. Moving from 6 percent to 6.4 percent is probably just noise.
And watch retention alongside it. A thumbnail that pulls the wrong people in will raise CTR and crater average view duration, and YouTube will quietly stop serving the video. A repackage that lifts clicks and holds retention is the real signal. I go deeper on the sequencing in the economics of repackaging, because the maths of which videos to touch first is genuinely different from the maths of making new ones.
Which old videos deserve the work first?
The ones with high impressions and low CTR, not the ones with your favourite content. Impressions are the raw material. If YouTube is already showing a video to people and they are declining, you have a packaging problem with a free audience attached.
My priority order when I open a back catalogue:
- High impressions, low CTR. The obvious win. The algorithm is offering you traffic and the packaging is refusing it.
- High CTR, low impressions, strong retention. These are underfed. Often a title change to match how people actually search does more than a thumbnail here.
- Evergreen topics with dated packaging. Text styles, faces, and colour trends age faster than the advice inside the video.
- Anything where the thumbnail duplicates a neighbour. If three videos in your grid look identical, you are competing with yourself in the sidebar.
I keep a running list of what to check on each one in our old video packaging checklist, and the practical order of operations for a whole library sits in repackage your back catalogue. Between them you can run this without hiring anyone, which is a perfectly fine outcome.
What the benchmarks will never tell you
Benchmarks cannot tell you whether the promise on your thumbnail matches the promise in your title. That mismatch is the single most common reason an old video underperforms, and no percentage in your dashboard flags it.
When a thumbnail says one thing and the title says a slightly different thing, the viewer has to do reconciliation work in the half second they give you, and half seconds are not for homework. The fix is to design them as one unit. That is the whole reason I stopped calling this thumbnail design and started calling it packaging: the thumbnail, the title, and the first frame are one promise, made in three places.
Benchmarks also cannot tell you the shape of your competition. On a video about a niche software workflow, a plain screenshot with four words might beat an expressive face, because the sidebar is full of expressive faces and the screenshot is the odd one out. Being different in a crowded row often beats being objectively better.
Should you hire someone or fix it yourself?
Fix it yourself if you have design skill and time, and hire if the back catalogue is large enough that the arithmetic favours it. Both are honest answers, and I say so even though I sell the second one.
The reason I lean towards packaging specialists over general design help is that this job is judged by one number in public, which is unusual. Most designers are not trained to think about a sidebar full of competing rectangles at 200 pixels wide. I wrote about that difference in more detail when comparing freelance marketplaces against a dedicated studio, and the bigger channel version of the argument lives in our take on packaging at scale.
At ClickTheory the pricing is deliberately simple, because I hate quoting. It is $250 for 8 custom thumbnails, about $31 each, one per video, with unlimited revisions, a dedicated designer, and title consultation included. Turnaround is 48 to 72 hours per thumbnail. If you want an A/B variant to test two directions on the same video it is $15 more, a title pack is $25, and rush is $20. Every design is custom for the video. We do not run templates, and we test styles on our own channels before they go anywhere near a client library.
Eight thumbnails is usually the right first bite for a back catalogue, because it is enough to measure a pattern and small enough that you are not betting the year on it. You can see the range of styles in our work before deciding anything.
The number to write on your wall
Pick one video, note its 28 day impressions and CTR by source, change the packaging, and check again in three weeks. That single loop teaches you more about your audience than any benchmark table, mine included.
We are Dubai based and work async with creators worldwide, so timezone is rarely an issue. If you want a second opinion on which eight videos in your library are worth repackaging first, send me the channel at samy@pixelsamy.com and I will tell you honestly if I think it is worth doing.