Thumbnails: what to change first when nobody is clicking
Thumbnail advice is usually a list of tricks. Here is the diagnostic order instead: what to check, in what sequence, when click-through is low.
Short answer
Fix thumbnails in this order: readability at small size, one clear subject, contrast against the surrounding feed, then curiosity. Most underperforming thumbnails fail the first test, because they were designed on a large screen and are viewed at roughly the size of a postage stamp.
A thumbnail is looked at for well under a second, at a size somewhere between a stamp and a business card, surrounded by competitors.
Almost every thumbnail problem comes from designing it in the wrong conditions: large, alone, on a good screen, by someone who already knows what the video is about.
Here is the order to fix things in.
One: is it readable at 120 pixels wide?
Shrink it. Actually shrink it, do not squint at it full size.
At that size:
- Can you tell what the subject is?
- If there is text, can you read it without effort?
- Is the main element large enough to register?
Most failing thumbnails fail here. Three elements competing for attention, text at a size that works on a monitor and disappears on a phone, a subject that occupies a fifth of the frame.
The fix is subtraction. One subject, large. Text if it earns its place, in three to five words, at a size that looks absurd on your screen and correct at thumbnail size.
Two: is there one clear subject?
The eye needs somewhere to land immediately.
One face. One object. One clear before-and-after. Anything requiring the viewer to work out what they are looking at has already lost, because they were not going to give it that second.
Composites with four elements, arrows pointing at three things, and split screens with a person on each side all fail this test for the same reason.
Three: does it stand out from what is around it?
This is the test people skip, and it is contextual rather than absolute.
Open the feed your video will appear in. Your own channel page, or a search result for your topic. Put your thumbnail among them.
If everything in that feed is dark and moody, a bright thumbnail wins. If everything is high-saturation yellow and red with shocked faces, the calm one wins. There is no universally correct palette, only contrast against the neighbours.
Four: is there a reason to click?
Only once the first three pass. Curiosity on an unreadable thumbnail is wasted.
What creates a reason:
- A visible outcome or result
- A contrast the viewer wants explained
- An expression implying a story
- Something specific enough to feel like a real thing rather than a category
What does not: a logo, a neutral portrait, text repeating the title, and anything that could sit on any video in your niche.
The title is half of it
Thumbnail and title are one unit. They are shown together everywhere and they should not say the same thing.
The pattern that works: the thumbnail shows a situation, the title supplies the tension. Or the title states the claim and the thumbnail shows the evidence. Two halves of one message rather than the same message twice.
If your thumbnail text is your title in fewer words, you have used the most valuable space on the page to repeat yourself.
A checklist before publishing
- Shrink to 120px. Still readable?
- One subject, clearly dominant?
- Placed beside ten competing videos, does it hold up?
- Is there a reason to click that the title does not already give?
- Does it look like it belongs to your channel?
That last one matters over time. Consistency in treatment, palette and framing makes a returning viewer recognise you before they read anything. It is the slowest-paying of the five and the one that compounds.
What is not worth your time
Chasing whatever is trending in thumbnail style. By the time a style is recognisable it is saturated, and contrast is the actual goal.
Perfecting a thumbnail for a video with a weak title. They fail together.
Redesigning old thumbnails at random. If you are going to go back, start with the videos that get impressions and few clicks. Those are the ones where the packaging is the whole problem, and they are the cheapest wins available.
Frequently asked questions
How much text should a thumbnail have?
Three to five words at most, and often zero. The title sits directly beside the thumbnail on every surface, so text that repeats the title wastes the space. Text earns its place when it adds a second idea the title does not carry.
Do faces always perform better?
Faces with a clear, readable expression usually do, because people process expressions faster than anything else in a frame. A face that is small, neutral or poorly lit performs no better than no face at all.
Should I use a frame from the video?
Rarely. Video frames are composed for motion and context, not for a static image at thumbnail size. A shot taken deliberately, or a still composed for the purpose, almost always reads better.
How do I test thumbnails without built-in A/B tools?
Shrink it to about 120 pixels wide and look at it beside your last ten videos and a few from channels your audience watches. Most problems become obvious immediately, and it costs nothing.
Want us to run the page, not just edit the videos?
We edit short form, long form and thumbnails, then title, caption, schedule and post them. Send one video and we will show you what we would do with it.
Book a free call