Do YouTube Dislikes Hurt Your Video? The Evidence, the Studio Number, and the Button YouTube Keeps Moving

For about two weeks in October 2026, some creators opened a desktop browser and found the dislike button missing from under the player. Screenshots spread, 9to5Google wrote it up on 20 October, and within two days the story had become the one every creator forum repeats every couple of years: YouTube is finally killing dislikes.

It was not that. A YouTube spokesperson, Kimberly Taylor, told The Verge that the company is not testing hiding the dislike button for viewers, and that what people had photographed was a new layout for creators looking at their own uploads, where the dislike control sits behind the three-dot menu. Outlets quote the sentence slightly differently, but every version carries the same denial. Full dislike counts, YouTube said, remain in Studio alongside the rest of your analytics, and the creator-side layout would roll out over the following weeks.

The interesting question is the one nobody answered in the rush: does it matter? Four years after the public count disappeared, most creators still cannot say with any confidence whether a dislike costs them reach, whether a dislike spike is a crisis or noise, or what the number in Studio is actually good for. There is more real evidence on this than the forums suggest — YouTube's own documentation, one very large independent study of the feedback controls, and three published papers on what happened after the counts were hidden. Put together, they give a clear answer, and it is not the one either side of the argument assumes.

Where the dislike button and the dislike number live in late 2026

Before the question of effect, the question of fact, because the mechanics are now different on every surface and that is most of the confusion.

Surface	Dislike button	The number
Long-form video, viewer watching someone else's upload	Present, next to like, unchanged	Not shown publicly since November 2021
Long-form video, creator watching their own upload	Moving into the three-dot menu, rolling out from late October 2026	In Studio only
Shorts	Retired in mid-2026; the like became a heart	Historical Shorts dislike data retained in Studio
Live streams	Present	In Studio only
Public Data API	n/a	Stopped returning dislike counts in December 2021

The Shorts change is the one most creators have not registered, partly because YouTube did not blog it. Reporting in late June 2026 traced it to a post from TeamYouTube in the Help Center community and a Creator Insider video: the dislike button retired from Shorts, the thumbs-up replaced by a heart, alongside 2x hold-to-speed playback and a clear-screen mode. It does not extend to long-form or live, where the button stays. Viewers who want to suppress a Short still have "Not interested", "Don't recommend this channel" and "Report" in the overflow menu, and creators keep the dislike figures their Shorts accumulated before the change.

So the pattern across four years is consistent, and it is worth naming plainly: YouTube has been steadily reducing how visible negative feedback is, while keeping the signal itself. The button has not been abolished anywhere except Shorts. It has been moved, unlabelled and de-counted.

What YouTube actually says dislikes do

YouTube's public documentation on recommendations is more specific than its reputation suggests. The Help Center page on how recommendations work describes a system learning from an enormous number of signals, and lists among them watch and search history, subscriptions, likes, dislikes, "Not interested" and "Don't recommend channel" selections, and responses to satisfaction surveys — the prompts that ask a viewer to rate something they have just watched. YouTube's older explainer, On YouTube's recommendation system, puts clicks, watch time, survey responses, sharing, likes and dislikes in one list and says they are combined to estimate what a viewer finds satisfying, with the weight of each signal varying by person.

Read that carefully and the grammar of it matters. Every one of those signals is described as evidence about a viewer. The system is working out what to put in front of one person next. Nothing in the documentation describes a dislike as a mark against a video that travels with it, lowers a score and suppresses it for everyone else.

That is not a technicality. It is the difference between two completely different mental models, and creators act on the wrong one constantly. In the penalty model, a thousand dislikes is damage and the sensible response is to delete the video. In the signal model, a thousand dislikes is a thousand people telling the system to show them less of this kind of thing — which, for the overwhelming majority of them, it was already going to do, because they did not watch to the end either.

The one large study of what the button does

The best independent evidence on the dislike button is not about creators at all. In September 2022 the Mozilla Foundation published Does This Button Work?, a study of whether YouTube's feedback controls do what viewers expect. It drew on 22,722 people running Mozilla's RegretsReporter extension, covered 567,880,195 recommended videos, and had reviewers rate roughly 40,000 pairs of recommendations for similarity against a control group.

The finding was a ranking of the four controls by how much each one reduced unwanted recommendations:

Control	Unwanted recommendations prevented
Don't recommend channel	43%
Remove from watch history	29%
Dislike	12%
Not interested	11%

Mozilla's framing was critical — its point was that the controls viewers reach for most are the ones that work least. YouTube disputed the methodology, saying the report did not take account of how its systems actually work, and the caveat on the study is real: people who install a research extension are not a random sample of YouTube users.

For a creator, though, the number cuts the other way, and it is the most clarifying figure in this whole subject. If pressing dislike shifts only about an eighth of what the person pressing it sees next, the button is a weak instrument even for its intended job. A control that weak in the hands of the viewer who used it is very unlikely to be a strong lever on where your video gets shown to everybody else. The dislike is low-bandwidth feedback, and YouTube has been treating it that way for years.

Why the number in Studio cannot be compared over time

Here is the part that should change how you read your own analytics, and almost nobody has heard it.

When YouTube hid public dislike counts in November 2021, it ran a natural experiment on viewer behaviour, and researchers took it. A study presented at the ACM Web Science conference in 2022 by Jung, Salminen and Jansen examined a major international news organisation's YouTube channel, comparing one full month of videos before the change (478 of them) against one full month after (394). Both likes and dislikes dropped — and dislikes dropped substantially more. The authors repeated the analysis on four other news channels in Arabic, English, French and Spanish plus one non-news organisation, and got similar results in all but one case.

Their conclusion is the actionable bit: because the interface change altered how people use the buttons, comparing like and dislike metrics from before and after the change gives invalid results. The ratio did not move because audiences got nicer. It moved because the count stopped being a scoreboard, and a large share of dislikes had been people adding to a number they could see.

Practically, that means three things for anyone with a channel older than five years. Lifetime like-to-dislike ratios are a blend of two different measurement regimes and mean very little. The "my old videos had a better ratio" observation that creators make all the time is partly an artefact. And any benchmark you have read for a healthy ratio, if it was written before 2022, is measuring a different thing from your Studio dashboard today.

What the sentiment did instead: it went to the comments

Two further strands of research suggest the negative feedback did not evaporate when the count was hidden. It changed address.

A working paper by Kim, Lu, Ma and Tafti compared YouTube against the Chinese platform Bilibili in a difference-in-differences design and reported that likes and comments both declined after the policy, with a decline in creator posting too — the opposite of the effect YouTube had hoped for. A 2026 paper by Zhang and Ng in New Media & Society, using an experiment alongside an interrupted time-series analysis, found that hiding dislikes reduced how often people commented while increasing negative expression in the comments that remained. A separate 2026 study of health-related videos reported the same direction: more comments, longer comments, more negative sentiment.

These are observational findings on different samples and they do not agree on every point — the engagement effects in particular are contested. But the pattern they share has a straightforward implication. Since late 2021 your comment section has been doing a job the dislike counter used to do. If you want to know what an audience thinks of a video, the comments are now the higher-resolution instrument, and they are the one you can actually read. They are also the one that requires management, which is why comment moderation stopped being an optional chore around the same time.

The metric worth watching, and the one to ignore

Dislikes are a poor headline number and a decent diagnostic. The difference is in how you normalise them.

Ignore the raw count. A video with 400,000 views and 1,200 dislikes is not in worse shape than one with 9,000 views and 300. Ignore the lifetime ratio, for the measurement reason above. What is worth tracking is dislikes per thousand views, computed inside a fixed window — the first 7 or 28 days — so that older videos accumulating slowly do not distort the comparison. Do that for a dozen of your own uploads and you have something no benchmark can give you: your channel's normal rate, against which an abnormal one is visible immediately.

The number is in the video's analytics in the tab Studio now calls Insights, under the engagement reporting, where likes and dislikes are shown together. You will need to switch the metric explicitly; the default views put likes forward and leave dislikes a click away. That is a deliberate design choice on YouTube's part and it is of a piece with everything else in this article.

What a dislike spike usually means

Once you have a baseline, a spike is informative — but only in combination with two other things you already have. Dislikes on their own cannot distinguish between a video people hated and a video people were angry about for reasons that have nothing to do with you. Cross-reference against the retention curve and the comments and the ambiguity mostly resolves.

What the pattern looks like	Most likely cause	What to do
High CTR, sharp retention drop in the first 30 seconds, raised dislike rate	Packaging mismatch — the thumbnail and title set up a different video	Fix the packaging, not the video
Normal CTR, normal retention, raised dislike rate, busy comments	Disagreement with the content's position, not its quality	Nothing. This is a healthy opinion video working
Dislikes arriving far faster than views, early, from a standing start	Coordinated dislike activity	Moderate comments, do not delete the video
Retention collapsing mid-video at a consistent timestamp	A specific section — an ad read, a tangent, an audio problem	Trim or re-edit; check the exact timestamp
Raised dislike rate across every recent upload	A format, cadence or quality drift the audience is registering	Treat as a channel problem, not a video one

The retention graph is the instrument that makes this work, because it timestamps dissatisfaction. Dislikes tell you people were unhappy. Retention tells you when they became unhappy, which is almost always enough to tell you why.

The packaging-mismatch case, in detail

The first row of that table is the one worth dwelling on, because it is the most common and the most fixable, and because it is the only one where a dislike genuinely does predict a distribution problem.

The shape is specific: click-through rate above your channel's normal range, retention falling off a cliff inside the first half-minute, dislike rate up. Each of those alone is ambiguous. Together they describe one thing. The packaging won the click by making a promise, and the video's opening did not keep it. The viewer who leaves at 20 seconds and presses dislike on the way out is not telling you the video is bad. They are telling you it is not the video they came for.

This is the failure mode behind most of what gets called clickbait, and the reason the word is so unhelpful. A strong curiosity gap that the video closes in its first minute is good packaging. The same gap left open — or filled with something unrelated — converts clicks into dissatisfaction, and YouTube's satisfaction signals are designed precisely to catch that. We have written the mechanism up at length in the psychology of clickbait thumbnails, and the companion piece on what counts as a good CTR makes the same argument from the other direction: a click-through rate well above your baseline, on a video that does not hold people, is a warning rather than a win.

The fix is almost never to delete or re-upload. It is to change the thumbnail and title so that they describe the video you actually made, and to fix the first 30 seconds so that it answers the question the packaging asks. Both are available after publishing — swapping the thumbnail on a live video costs nothing and keeps the URL, the views and the comments intact, and the hook is the cheapest part of a video to re-cut. If you are not sure which half is at fault, check the thumbnail at the sizes it is actually seen first, with the thumbnail preview tool; a promise that is legible at 1280 pixels and incoherent at 168 is a promise viewers are reading differently from you.

Coordinated dislikes: what you can and cannot do

The third row needs its own treatment, because it is the case YouTube built the 2021 change for and the case creators handle worst.

YouTube's stated reason for hiding the counts was targeted dislike campaigns. In its announcement of the change, the company said its experiment found that when the count was not visible, viewers were less likely to target a video's dislike button to drive the number up, and that smaller and newer channels were being targeted at a higher proportion than larger ones. That second detail is the one worth holding on to. The protection was designed for channels with no capacity to absorb a pile-on, which is most channels.

What follows from this is a short list of things that do not help. Deleting the video destroys every real signal it accumulated and, if it had traffic, a page that was ranking; our piece on whether to delete old videos covers why that trade is almost always bad. Addressing it on camera extends the story's life and hands the activity an audience. Buying engagement to balance the ratio is a policy problem on top of a reputation problem, and it does not work anyway. The ratio is not public, so there is no scoreboard to correct.

What does help is unglamorous: moderate the comments, where the actual harassment lives and where you have real tools; hold the comment section to whatever standard you normally hold it; and keep publishing on schedule, because the signal that matters to your distribution is whether the next upload satisfies the people who choose to watch it. A dislike pile-on is a number in a dashboard. It is not an audience, and the system is not reading it as one.

Shorts: the negative signal you can no longer see

Everything above concerns long-form, where the button survives. On Shorts, since mid-2026, there is no dislike at all, and that changes what feedback means on the format.

The replacement for the positive signal is the heart, which YouTube has described as a way of marking something that genuinely connected rather than a flat approval. There is no replacement for the negative one. A viewer who does not want your Short swipes, and a swipe at two seconds is indistinguishable in your dashboard from a Short that simply did not reach anyone — both show up as weak performance with no diagnosis attached. The explicit suppression tools are still there in the overflow menu, but they take a deliberate tap in a format built around the swipe.

The practical consequence is that on Shorts, the first frame carries more of the burden of honest signalling than it does anywhere else, because it is doing the work of the thumbnail and the hook simultaneously, and because when it overpromises there is no visible metric to tell you so. Our Shorts thumbnail guide covers what the frame is doing on each surface, and why Shorts stall goes through the diagnostics that remain available once the dislike is gone.

Should you look at your own dislikes at all?

The October change is, read honestly, a wellbeing feature. Moving the dislike control behind a menu on your own uploads does nothing for viewers and nothing for the algorithm. It removes a thing creators press, or stare at, on their own videos.

Whether you take the hint is a judgement, and it depends on what you are actually going to do with the number. If you are going to compute a rate, window it, and read it next to retention, the number is useful and you should look at it monthly rather than hourly. If you are going to refresh it the evening a video goes out, there is no version of that behaviour that improves the video, and the first 24 hours have better things in them to watch. The counts are in Studio either way. YouTube has simply stopped putting them under your thumb.

The bottom line

Do dislikes hurt your video? Not in the way the question implies. There is no documented penalty, the one large independent study found the button is a weak instrument even for the viewer pressing it, and the count YouTube keeps for you is not comparable with the one it kept before 2021. What does hurt a video is dissatisfaction, and dislikes are one of its least precise symptoms — behind retention, behind comments, behind the survey responses you never see.

Where the number earns its place is as a cross-check. A dislike rate above your own baseline, on a video that over-performed on clicks and under-performed on retention, is the clearest evidence you will get that the packaging and the content are describing two different videos. That is a fixable problem, and it is fixable in minutes rather than in a re-edit.

If that mismatch is a recurring pattern on your channel rather than an occasional one, the cause is usually that the thumbnail gets made last, under time pressure, by someone trying to win a click rather than describe a video. Thumblore generates thumbnails in your channel's style from a description of the video itself, which makes it cheap to produce the honest version of the frame and compare it against the tempting one — and if you want to know which of the two your audience actually prefers, Test & Compare will tell you with watch time rather than with a number behind a menu.