A creator sent me a screenshot last month with one line: "this video did 400k in 2023 and it got 90 views yesterday." He wanted to know what he had done wrong. The honest answer is that he had done nothing wrong. His video did exactly what almost every video does. It burned bright, it cooled, and then it sat in a library where nobody had any reason to click it.
Old videos stop getting views for reasons that are mostly boring and mostly fixable. Not one of them is "the algorithm hates you." I have looked at hundreds of back catalogues at this point, and the pattern repeats with almost annoying consistency. So here is what actually happens to a video after its first two weeks, why the decay looks steeper than it is, and what you can change without filming anything new.
Why do old videos stop getting views at all?
Because a video only keeps circulating while it keeps earning its slot, and after the first push it has to earn that slot against a much harder crowd. In week one your video is shown mostly to your subscribers and to viewers who already watched something close to it. Those people are easy. They know your face, they trust the channel, they click at a rate you will never see again.
After that window, distribution shifts to suggested and to search. Now your thumbnail sits next to videos from channels with twenty times your budget, on topics that got covered again more recently, in a feed where the viewer has zero loyalty to you. The click rate that felt fine in week one gets crushed in month six, and once the click rate drops the impressions follow. That is not punishment. That is arithmetic.
There are really only four causes worth your time:
- The packaging aged. Your thumbnail style, your face, your title conventions and your on screen text all belong to a version of your channel that no longer exists.
- The promise expired. "2023" in the title, a tool that got renamed, a price that changed, a platform update that made half the advice wrong.
- The topic got recovered. Somebody made the same video better, or you did, and your own newer upload is eating the impressions.
- The video never had a searchable identity. It was made for your subscribers on the day it went up and it says nothing to a stranger scrolling.
Three of those four are packaging problems. That is the useful part.
Is it the algorithm or is it your thumbnail?
It is almost always the thumbnail and the title working as a pair, and there is a simple way to prove it to yourself. Open YouTube Studio, take a video that is a year or more old, and set the date range to the last 28 days. Look at impressions and click through rate side by side.
If impressions are still healthy in the thousands but CTR has fallen to 1 or 2 percent, you have a packaging problem. The system is still offering your video to people and those people are choosing something else. If impressions have collapsed to near zero, the video has already been shelved and you need packaging plus a nudge, which usually means a fresh title, a fresh thumbnail and a real reason for the video to resurface.
Impressions with a bad click rate is the best news you can get. It means the shelf space is still yours and you have only lost the argument at the shelf.
The reason people misdiagnose this is that decay feels like a verdict. It is not. YouTube's own creator documentation is fairly plain that reach follows viewer behaviour, and viewer behaviour at the impression level is a half second decision made entirely on the image and the words next to it. Nothing about the edit, the audio or the retention curve gets consulted before that click.
I wrote a longer breakdown of what actually happens to distribution when you swap an image, because the myth that changing a thumbnail resets the algorithm refuses to die and it makes people either too scared to touch old videos or far too optimistic about a single swap.
What does packaging mean beyond the thumbnail?
Packaging means the thumbnail, the title and the first frame of the video treated as one argument, not three separate assets made by three different moods. Thumbnail design alone is the old framing, and it is why so many redesigns fail. You get a prettier image that promises something the title contradicts, and the viewer resolves the conflict by scrolling.
Every piece of packaging answers one question for a stranger: what do I get, and why should I believe you have it. A working pair usually splits the labour. The thumbnail carries the emotional stake or the visual proof. The title carries the specific, checkable promise. If both try to do the same job you have wasted half your surface area, and if neither does it you have a mystery, which converts terribly outside your subscriber base.
That is the standard I hold at ClickTheory, and it is why title consultation sits inside the pricing and services rather than being sold as a separate thing. An image without the right words on it is decoration.
Which old videos are actually worth repackaging?
Rank your back catalogue by evergreen potential, not by past performance, and repackage the top eight. Past performance tells you what your audience of the time liked. Evergreen potential tells you what a stranger might search for next year.
Here is the sort I use, in order:
- Still impressions, low CTR. The shelf space is live. These move first and they move fastest.
- Evergreen topic, dead packaging. Anything where the advice still holds but the image screams 2022. Tutorials, frameworks, mistakes lists, comparisons.
- Good retention, few impressions. People who find these finish them. That is a strong signal wasted on a weak cover.
- High past views, now flat. Proof the topic works. Often just needs a broader promise for a non subscriber.
Skip anything tied to a dated event, anything where your on camera advice is now wrong, and anything you are embarrassed by. Repackaging a video you no longer stand behind is how you get traffic you do not want.
The economics here are the part people miss. A new video costs you a script, a shoot day, an edit and two weeks of hoping. A repackage costs you an image and a sentence. For most channels the back catalogue is the cheapest growth available, and I have argued the whole case for that in the guide to repackaging old YouTube videos if you want the longer version.
How long should you wait before judging a repackage?
Give it 14 to 28 days, and judge impressions and CTR, not views. Views are a lagging number that mixes traffic sources and misleads you in both directions.
Ordinary results look like this. Days one to three, nothing visible, and this is where most people panic and swap back. Days four to ten, impressions start to move if the new packaging is beating the old one. Days ten to twenty eight, you get a real read. A repackage that works usually shows a CTR improvement first and an impressions improvement second, in that order, because reach follows clicks and not the other way round.
Change one video at a time if you are learning. Change eight at once if you already know your style, because the sample size is what makes the lesson trustworthy. And write down what you changed. A back catalogue test with no notes is just redecorating.
Does a rebrand make this worse?
Yes, and it is the most common hidden cause of a catalogue that quietly died. If you changed your name, your niche, your face on camera or your colour palette, every video made before the change now looks like it belongs to someone else. A viewer who liked your new video and clicked through to your library sees a stranger's shelf.
That is a trust problem more than a design problem, and it needs a plan rather than a paint job. I covered the sequencing of it in the piece on repackaging after a rebrand, because the order you do it in matters more than the design itself.
Niche shapes the answer too. A finance channel around 80k subs and a fitness channel at the same size need completely different repair jobs, because their viewers are checking for different kinds of credibility. The same is true for teaching led channels, where the face and the claim carry almost everything, which is why thumbnail packaging for coaching creators is its own conversation.
What does it cost to fix a back catalogue?
At ClickTheory, $250 gets you 8 custom thumbnails, which works out at 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 a second image to test against the first, an A/B variant is $15. A title pack is $25 and rush is $20.
Eight is a deliberate number. It is enough to cover the strongest part of a back catalogue, enough to learn something real from the results, and small enough that you are not betting a quarter's budget on a hypothesis. Every thumbnail is custom for that video. We do not run templates, and the styles we use get tested on our own channels before they go anywhere near a client's library.
If you want to see how the pairs read together rather than as single images, our work is the honest version of that. And if you are going to brief anyone, us or a freelancer you already like, the walkthrough on how to brief a thumbnail designer will save you two revision rounds on its own.
Where does repackaging stop being the answer?
Sometimes the video is genuinely finished, and it is worth saying so. If the content is factually stale, if the audio is unlistenable, or if the topic has moved on in a way no title can bridge, then no image will save it. Repackage the ones that still deserve to be found and let the rest sit.
Also, do not repackage everything at once and call it a strategy. A library where all 60 covers changed on the same Tuesday looks like a channel that panicked. Work in batches of eight, learn, then do the next eight. If the underlying footage is the weak link rather than the cover, that is a different fix, and my other team edits video podcasts for exactly that reason.
We are based in Dubai and work async with creators worldwide, so time zones are not a constraint on any of this. If you have a back catalogue you think is dead, send me the channel link at samy@pixelsamy.com and I will tell you honestly which eight videos I would start with.