I went through a client channel last month with 214 published Shorts sitting behind a single wall of grey. Every cover was an auto-pulled frame. Half of them were a blurry mid-blink face. The channel had 3.1 million lifetime Shorts views and a Shorts tab nobody browsed, because there was nothing on that tab worth stopping for.
That is the whole problem in one sentence. Shorts get most of their views from the feed, where the cover barely matters. But they get their second life from the Shorts tab, from search, from your channel page, and from the related shelf, and in all four of those places the cover is the only thing doing any selling.
So when people ask me whether it is worth going back to refresh old Shorts covers, my answer is usually yes, but not for all of them, and not for the reason they think.
Why do old Shorts covers stop working?
They mostly never worked in the first place. A Short published without a chosen cover gets a frame lifted from the video, and that frame is picked for you, not for the viewer. You end up with a shot of your hand halfway to your face, or a caption sliced through the middle, or the exact moment your eyes were closed.
The second reason is context drift. A cover that read clearly in 2023 can read as dated now, because your face changed, your set changed, your caption style changed, or the trend the Short was riding is over. When your Shorts tab shows six different visual eras stacked on top of each other, a new viewer cannot tell what the channel is. That confusion costs you the subscribe far more than any single view.
The third reason is that the surfaces changed. Shorts are shown in more browse contexts now than when most of these videos were uploaded. Covers that never had to compete are suddenly competing.
A Short with a bad cover is not a failed video. It is a finished video with no front door.
What actually counts as a Shorts cover in 2026?
The cover is the vertical still that represents your Short everywhere outside the swipe feed: your channel's Shorts tab, search results, the Shorts shelf on the homepage, playlists, and any link someone shares. YouTube lets you set it after upload, and the current process is documented in the official YouTube Help documentation on video thumbnails.
Here is the part people get wrong. A Shorts cover is not a shrunk down horizontal thumbnail. The aspect ratio is 9:16, it is displayed at maybe 120 pixels wide in a grid, and it sits next to five other Shorts in a column. That means:
- Two or three words maximum, and they need to be huge
- One subject, cropped tight, no wide shots
- Contrast that survives a thumb sized render
- Text placed away from the bottom, where the view count and the duration chip land
The grid is the real test. Not the file open on your desktop at full size. I always mock up nine covers together in a fake Shorts tab before I sign off on any of them, because a cover that looks brilliant alone can vanish when the one above it is louder.
How do you pick which old Shorts to refresh first?
Start with the Shorts that already have views, not the ones that flopped. A Short with 80,000 views has proof that the content works. Improve the cover and you are adding a front door to a room people already like. A Short with 400 views has an unproven hook, and a new cover will not fix a weak first two seconds.
My ordering, in practice:
- Top 20 by lifetime views. These are the ones getting the most impressions in browse and search right now. Highest leverage, easiest win.
- Anything ranking in YouTube search. Check your search traffic report. If a Short is pulling steady search impressions with a low click rate, the cover is the bottleneck, full stop.
- Shorts that map to a current offer or series. If you are selling something or building a playlist, the covers in that cluster need to look like a set.
- The first row of your Shorts tab. Whatever sits at the top of that grid is your storefront window whether you planned it or not.
Everything else can wait or stay as it is. I have written before about what to do when older uploads quietly outperform your newest ones, and the same logic applies here: follow the demand that already exists rather than trying to manufacture new demand.
The four changes that do most of the work
When I refresh a batch of Shorts covers, roughly four decisions account for most of the lift.
Crop harder than feels comfortable. In a 120 pixel wide tile, a medium shot reads as a smudge. Go to the face. Chin to hairline. If the Short has no face, go to the single object that matters and let it break the frame edges.
Cut the text to two words. Not a sentence, not the title again. The title is already sitting under the cover in search and on the tab. The cover text should say the thing the title cannot, usually the stake or the number. "Lost $9,000" beats "How I lost nine thousand dollars on a bad trade."
Give the channel one repeating visual signature. One accent colour, one text treatment, one crop language. This is the difference between a Shorts tab that looks like a channel and one that looks like a folder. It is also the single fastest way to stop covers reading as amateur even when the video quality is fine.
Fix the bottom 15 percent. The duration chip, the view count and the play glyph all live down there depending on the surface. Keep that band clean or your best word gets a grey pill dropped on it.
One more thing that is not a design decision: rewrite the title while you are in there. Shorts titles are read in search and on the tab, and half the old ones say "Part 3" or "#shorts" and nothing else. Cover and title are one unit. Changing one without the other is doing half the job.
Does refreshing a cover reset the algorithm?
No, and this is where I want to be honest rather than sell you a fantasy. Swapping a cover does not reset a Short's performance history, does not push it back into the swipe feed, and does not trigger a re-recommendation event. The video keeps its watch data.
What changes is the click rate on the browse and search surfaces where the cover is visible. If a Short is already getting impressions on the Shorts tab or in search, a better cover converts more of those impressions. If it is getting none, a better cover converts more of nothing.
That is why the ordering above matters so much. Refreshing covers is a conversion improvement on existing distribution, not a distribution strategy. Anyone who tells you a new cover will revive a Short that never got shown is guessing. The YouTube creator resources are reasonably clear on how browse and suggested surfaces work if you want to read the primary source rather than take my word for it.
The compounding effect is real though. Better covers on your top 20 raise the browse click rate, which raises the number of people who land on the channel page, which raises the number who scroll a second row, and that second row is where subscribes happen.
How many should you do at once?
Batches of eight. That is not a marketing number, it is what I have found actually works, and it happens to be how we sell.
Eight is enough to establish a repeating visual language across a row and a half of the Shorts tab, and small enough that you can see the result before committing to the next batch. Doing three is not enough to look like a system. Doing forty at once means you lock in a style before you have any evidence it converts.
Our setup: $250 for 8 custom thumbnails, which works out to about $31 each, one per video, with unlimited revisions and a dedicated designer on the batch. Title consultation is included, which for Shorts refreshes is most of the value, because the titles on old Shorts are usually worse than the covers. Turnaround is 48 to 72 hours per thumbnail. If you want to test two directions on the same Short, an A/B variant is +$15. A title pack is +$25 and a rush is +$20.
Everything is drawn per video. No templates, no swapping your face into a layout we sold to eleven other channels. We test styles on our own channels before they go anywhere near a client batch. You can see the range of what that looks like in our work, and the full rate card sits on pricing and services if you want to compare it against what you are paying now. I have also broken down the wider market in a piece on what YouTube thumbnails actually cost.
What if you have a hundred or more?
Do not try to do all of them. I say this to almost everyone who arrives with a big archive, including channels who came to me after going back through videos from years ago and getting overwhelmed halfway.
With a large library, the right move is a triage pass first: sort by lifetime views, sort by search impressions, mark the clusters that share a topic, and only then start designing. Twenty five well chosen covers out of two hundred will outperform two hundred rushed ones, and you will still have a coherent channel at the end. I wrote out the full sequencing for that in a piece on repackaging a 100 video library, and the Shorts version of it is the same shape with a tighter crop.
When is it not worth refreshing?
Skip it when the Short is genuinely dated, when the content is wrong now, or when the video was a one off trend piece nobody will search for again. A beautiful cover on an irrelevant Short is 31 dollars set on fire.
Skip it also when your problem is retention rather than clicks. If your Shorts are getting swiped away in the first second, the cover is not the issue and no designer can fix it. Go back to the hook.
And skip it when you have fewer than about 15 Shorts total. At that size you are better off putting the effort into the next ten uploads and packaging them properly on day one.
If you do have an archive sitting there doing nothing, and you want a second pair of eyes on which eight to start with, send me the channel link at samy@pixelsamy.com or use contact and I will tell you honestly whether a refresh is worth your money.