Someone in a Discord I lurk in posted a thumbnail last month with a fat red arrow pointing at a laptop screen, and got told within four minutes that arrows were dead in 2019. The person who said it had a channel with 1,200 subscribers. The guy who posted the arrow had a channel doing consistent six figure views.
That is roughly the state of this argument. Arrows and circles are the most confidently dismissed technique in thumbnail design, and they are also still everywhere on the pages that actually get watched. Both of those things are true, and the reason they are both true is more interesting than the argument.
I have spent the last while testing this properly, on our own channels first and then inside client packs where we could run a variant. Here is what I actually think.
The short answer, then the long one
Arrows and circles work. They do not work the way people used them in 2018, and they stopped working as a decoration. They still work as a pointer.
That distinction is the whole post. A red circle drawn around a face because the template had a red circle in it does nothing. A red circle drawn around the one small detail in a photo that a viewer would otherwise never notice is doing real cognitive work. Same shape, same color, completely different job.
The people who say arrows are dead are usually looking at the first kind. The people still using them profitably are doing the second kind and often cannot articulate why it works, which is why they lose the argument in comment threads.
Why the technique got a bad name
Around the time the MrBeast style went fully mainstream, every thumbnail pack template on the internet shipped with an arrow layer and a circle layer. They were free, they were red, and they were dropped onto thumbnails that did not need them.
What happened next is predictable. Viewers saw thousands of red arrows pointing at nothing. The arrow stopped meaning "look here" and started meaning "this is a YouTube video by someone who bought a template." It became a genre signal instead of a directional one. That is a real cost, and it is why the backlash was fair.
The second problem was worse. Arrows and circles eat space. On a thumbnail that is already fighting for legibility at 320 pixels wide, a big graphic element that adds no information is the most expensive thing you can put in the frame. Every pixel it takes is a pixel not spent on the face, the subject, or the contrast that actually carries the click. I have written before about how contrast is the whole game at small sizes, and an arrow that does nothing is just a contrast thief with an attitude.
So the technique did not die. It got overused into meaninglessness, which looks identical from the outside.
What we tested
We run styles on our own channels before we put them in front of anyone paying us, which means I get to be wrong in private. Over a stretch of months we ran arrow and circle variants against clean versions on the same videos, same titles, same upload windows where we could manage it.
The pattern that came out of it was not "arrows lift CTR by X percent." That number does not exist and anyone quoting it to you is selling something. What came out was a condition.
The pointer earned its place when the thumbnail contained a small, specific, non obvious detail that the viewer needed to see for the title to make sense.
On a tech style video where the payoff was a tiny setting buried in a menu, the circled version outperformed. On a talking head thumbnail where the arrow pointed at the person's own face, it did nothing measurable and in one case looked slightly worse. On an educational explainer where we circled one number inside a chart, the lift was the clearest we saw in the whole batch, and it held across two reuploads.
One caveat I want to be honest about. These were our channels, mid sized, in niches with fairly literate audiences. That is not a universal law and I am not going to pretend it is. If you run a beauty channel with a completely different viewer profile, run your own test. I would rather you distrust my numbers and check them than repeat them.
An arrow is not a style choice. It is a sentence. If you cannot say out loud what the arrow means, delete it.
The rule I now use
Before an arrow or circle stays in a thumbnail, it has to pass one question: what would the viewer miss without it?
If the answer is "nothing, the image reads fine," it goes. If the answer is "they would not notice the crack in the wall, the number in the corner, the second monitor," it stays. That is the entire test and it kills about eighty percent of the arrows people want to add.
There is a second question underneath it. Is the thing you are pointing at actually interesting? A perfectly executed arrow pointing at something boring just tells the viewer faster that they can skip you. Pointers amplify, they do not create. If your subject is weak, the arrow makes the weakness legible sooner.
This is why the technique fails so often for people who add it late. They finish a thumbnail, feel like it lacks energy, and reach for an arrow to fix a composition problem. Arrows do not fix composition. They make bad composition louder. Fix the frame first, the way we approach thumbnail composition basics, then decide whether anything still needs pointing at.
Circles versus arrows, since they are not the same tool
People bundle these together and they behave differently.
A circle is an enclosure. It says "this region." It works when the thing you are highlighting has fuzzy edges or sits inside visual noise, like one face in a group, one window in a building, one line in a spreadsheet. Circles are forgiving because they do not have to be precise about direction.
An arrow is a vector. It says "from here to there." It works when there is a spatial relationship worth showing, or when the target sits near an edge and needs to be pulled into attention from across the frame. Arrows are less forgiving. A badly angled arrow points the eye out of the thumbnail entirely, which is the single most damaging mistake in this whole category.
My practical preferences after a lot of testing:
- Circles beat arrows for small details inside busy images.
- Arrows beat circles when you need to connect two elements, like a person reacting to a thing.
- Never use both in one thumbnail. It reads as panic.
- Keep the stroke weight heavy enough to survive at mobile size, which usually means thicker than looks right on your monitor.
- Red is not mandatory. On dark backgrounds a bright yellow or a saturated cyan often separates better, which matters a lot when you are designing for dark mode feeds.
That last one is underrated. The default red arrow was designed for a white interface era. Most viewing now happens against near black, and pure red against dark grey has less separation than people assume.
Where I see this land hardest
Educational and tutorial content is where pointers still earn the most, and it is not close. When the video promises to show you a specific thing, a thumbnail that shows the specific thing and marks it is doing the title's job visually instead of repeating it in text. That is the highest value move available in the format.
We do a lot of this in thumbnail design for education channels, where the subject matter is often small, technical, and easy to miss in a photo. A circled detail in that context is not decoration, it is captioning.
Entertainment and vlog content is where it lands worst. The face is already the subject, the emotion is already the hook, and adding a pointer just adds a layer between the viewer and the thing they came for.
Commentary sits in the middle and depends entirely on whether you are showing evidence. If the thumbnail contains a screenshot of the thing being commented on, mark it. If it is just you looking annoyed, do not.
How to test this without lying to yourself
Most thumbnail testing is bad because people change three things and credit the one they liked. If you want to know whether the arrow did anything, the arrow has to be the only difference. Same crop, same colors, same text, same title, same publish behavior.
Then give it enough impressions to mean something. Reading a swing after four hours is reading noise. If you are unsure what counts as signal, our notes on reading YouTube analytics for packaging cover how long to wait and what to ignore.
And size the decision honestly. A small change on a channel doing decent volume compounds into real watch time over a year, which is the argument for chasing a one percent CTR lift at all. But a small change on a channel doing 300 views a video is not measurable at all and you should spend that energy on the video instead.
If you want variants run properly without building the pipeline yourself, that is part of what we do. Our pack is $250 for 8 custom thumbnails, one per video, with unlimited revisions and a dedicated designer, and an A/B variant is a $15 add on when a video is worth the extra swing. You can see the range on our work or just reply to an email.
What I actually believe about this
Arrows and circles are not a trend that expired. They are a functional tool that got treated as an aesthetic, and aesthetics expire while functions do not.
The next time you see someone say the technique is dead, look at what they are actually objecting to. Nine times out of ten it is a red arrow pointing at a man's forehead, and they are right about that specific arrow. They are just wrong about the category.
Use the pointer when there is something to point at. Skip it when there is not. That is a boring rule and it has outperformed every stylistic opinion I have held about it.
If your packaging is stuck and you cannot tell whether the problem is the arrow or the frame underneath it, send me the thumbnail. I would rather tell you it is fine than sell you a fix you do not need. Same for anyone editing long form or podcast content who wants the video side handled too.