Most thumbnail projects do not go wrong in Photoshop. They go wrong in the twelve words someone typed into Slack at 1am: "video about my new camera, make it pop."
I have watched that single sentence turn into four revision rounds. Not because the designer was slow, but because nobody had decided what the video was actually promising. A brief is not paperwork. It is the place where you make the hard decision once, in writing, so it does not get relitigated three drafts later.
So here is the thumbnail brief template we use, field by field, with the reasoning behind each one. Steal it. Put it in a Notion database, a Google Form, a text file in your project folder, wherever your team already lives.
What goes in a thumbnail brief template?
A good thumbnail brief has nine fields, and seven of them take under a minute to fill. Here is the whole thing before I explain any of it.
- Working title (the one you would actually publish)
- The promise in one sentence (what the viewer gets)
- Who is scrolling past this (subscribed viewer, browse, suggested, search)
- The single visual idea (one noun, one verb)
- Assets available (raw frames, product shots, screen recordings, face shots)
- Text ceiling (max words on the image, usually 3 or 4)
- What it sits next to (2 to 3 competing thumbnails from the same niche)
- Hard nos (things you will reject on sight)
- Deadline and publish slot
That is it. Six minutes of typing. It replaces about ninety minutes of back and forth, which is the entire argument for doing it.
Why does the title belong in a thumbnail brief?
Because the thumbnail and the title are one unit, and designing either alone is guesswork. This is the whole reason the industry moved from calling it thumbnail design to calling it YouTube packaging. The image, the title and the first frame promise are a single argument aimed at one person deciding whether to spend twelve minutes with you.
If your title already says "I Spent $4,000 On This", the thumbnail must not also say "$4,000". You just spent your most valuable pixels repeating a fact the viewer already read. The image should show the thing that makes $4,000 feel insane, or the face that says it was a mistake.
So the brief carries both. When a client sends us a working title, the designer can build against it. When they do not, we are inventing a title in our heads and designing to that invented title, and then the real title arrives and the composition is wrong.
If you can swap the title on a thumbnail and nothing about the image needs to change, the image is decoration, not packaging.
This is also why we include title consultation in every pack rather than selling it as a favour. It is not generosity. It is the only way the image can be judged.
How specific should the visual idea be?
One noun and one verb. That is the target. "Camera falling" beats "something dynamic with the camera." "Bread collapsing" beats "show the fail."
The reason is compression. A thumbnail is read at roughly the size of a postage stamp on a phone, in under a second, usually while the viewer is half looking at something else. A single idea survives that. Two ideas turn into visual mush at 320 pixels wide.
When a brief comes in with three ideas, we do not average them. We pick one and build variants around it, which is what the A/B variant add on is for at $15. Averaging three ideas produces a thumbnail that tests worse than any of the three on its own.
If you genuinely cannot pick, write all three down and mark which one scares you most. That one is usually right. The safe option is safe because it looks like everything else in the feed, and looking like everything else in the feed is the actual failure mode.
The fields creators skip, and what it costs
Two fields get left blank almost every time, and they are the two that cause revisions.
What it sits next to. Nobody sends competitor thumbnails. But your thumbnail is never viewed alone. It is viewed in a column of eight, and the job is contrast. If the four videos above yours are all dark backgrounds with a white sans serif headline, the winning move might be a flat bright field with no text at all. You cannot know that without looking. Screenshot the browse feed for your topic and paste it in. It takes twenty seconds and changes the whole approach.
Hard nos. Everyone has them. Some creators will not do open mouth shock faces. Some will not put arrows on anything. Some have a brand colour that is non negotiable and one they hate. Writing three hard nos at the start prevents a full revision round that exists purely to discover them. Related: pinning down type is easier if you have already read our guide to thumbnail fonts and can name a family instead of saying "cleaner."
What should the assets section actually contain?
Every usable frame you have, at full resolution, in one folder, named plainly. Not a link to the whole edit timeline. Not a phone screenshot of your monitor.
Specifically, the assets that make thumbnails good:
- Three to five face frames with different expressions, shot cleanly, not grabbed from a compressed export
- The object the video is about, photographed separately if possible, on any plain background
- Any screen recording stills at native resolution, not scaled up
- Before and after pairs if the video has any transformation in it
The single biggest quality lever in a thumbnail is source resolution on the face. A face pulled from a 1080p export that was already compressed twice will look soft no matter what the designer does. Two minutes of shooting stills on set solves it permanently. If you build one shared folder of face shots and reuse it across a whole quarter, you have basically removed the problem, and it pairs neatly with keeping a proper version history of your thumbnails so you can see which expression actually performed.
How do you brief a batch instead of one video?
You write one system brief plus eight short per video briefs, and you do it in a single sitting.
This is how our $250 pack for 8 custom thumbnails is meant to run. The system brief covers the things that are true across all eight: colour rules, type family, how much of the frame the face occupies, whether the channel uses text at all. The per video brief is just fields 1, 2, 4 and 8 from the template above, three lines each.
Batching does two things. It makes the set look like a channel rather than eight unrelated images, which matters enormously on a channel page. And it lets the designer make decisions once instead of eight times, which is why a pack works out around $31 per thumbnail rather than agency one off rates.
It is also the fastest way to attack a back catalogue. Old videos are the cheapest growth most channels have. A video published eighteen months ago already has watch time data, already has a place in the suggested graph, and already survived the algorithm's first look. Repackaging eight of those often moves more total views than making a ninth new video, and the brief for a repackage is even shorter because you already know how the video performs. YouTube's own creator resources make the same point about impressions and click through rate being a diagnosable pair rather than a mystery.
Does a brief actually reduce revision rounds?
Yes, and the mechanism is boring: it moves disagreement to the cheapest possible moment. Disagreeing about the promise costs a sentence. Disagreeing about the promise after a full render costs a day.
We offer unlimited revisions on every pack, and I still want fewer of them, because revision rounds are almost never about craft. They are about a decision that was never made. The rounds that are about craft are fine. Those are quick. "Push the face 8 percent bigger" is a five minute fix.
The rounds that hurt are "actually the video is more about X." That is a brief problem wearing a design costume. Once the brief exists, feedback gets sharper too, which is a skill of its own and one I broke down in how to give thumbnail feedback. Point at the field, say which one changed, and the fix is obvious.
What does the brief look like in a niche?
It gets shorter and more specific, because niches have their own visual grammar.
A finance channel around 80k subs briefs almost entirely in numbers and faces. The promise field is usually a figure, the visual idea is a reaction, and the hard nos are about not looking like a scam.
A build or making channel briefs in objects and stages. The promise is the finished thing, the visual idea is the moment of transformation, and the text ceiling is often zero words because the object carries it, which is exactly the pattern I walked through in thumbnail packaging for DIY creators.
And when you launch a second channel, the brief has an extra field: how much it should look like the first one. Usually less than you think. I covered that tradeoff in packaging a second channel.
If you want the mechanical details of what YouTube accepts, resolution, file size, custom thumbnail eligibility, the official YouTube Help documentation is the source I check rather than trusting a blog post, mine included.
Use the template, or send us the video
Copy the nine fields, keep them in your video template, and fill them the moment you lock the edit while the video is still fresh in your head. Fill them badly and it still beats a Slack message.
If you would rather hand the whole thing over, that is what we do. Custom per video, never templates, styles tested on our own channels first, 48 to 72 hours per thumbnail, dedicated designer, title consultation included. Have a look at what we charge and what is in each pack, or send eight old videos to samy@pixelsamy.com and I will tell you honestly which ones are worth repackaging.