YouTube's Made for Kids Setting: What It Switches Off, and Why the Thumbnail Ends Up Doing All the Work

Every upload asks you one question that has nothing to do with YouTube. It sits below the description box, it is required, and it is phrased as if it were a category: is this video made for kids, or not made for kids. Most creators answer it in half a second and never think about it again.

In September 2025 the Federal Trade Commission announced a settlement with Disney over that half-second. The allegation was that child-directed videos had been designated "not made for kids", that this caused personal information to be collected from children watching them without parental consent, and that those children were shown features never intended for them — autoplay into general-audience videos, and unrestricted public comments. The order carried a ten million dollar civil penalty. The FTC wrote up the lessons for everyone else who uploads to the platform.

So the box is worth more than half a second. But compliance is only half of why it matters, and the half creators already worry about. The other half is mechanical: answering yes removes roughly a dozen features from the video, including nearly every route a viewer has to your next upload, and concentrates the whole remaining job of getting clicked into one rectangle. This is what the setting does, how to work out which answer is true for your channel, and what packaging looks like when it is the only distribution you have left.

Why the question exists at all

The Children's Online Privacy Protection Act governs the collection of personal information from children under 13 in the United States. It applies to operators of child-directed online services, and it treats persistent identifiers — the cookies and device IDs that make behavioural advertising work — as personal information. In September 2019 the FTC and the New York Attorney General announced a settlement with Google and YouTube over alleged COPPA violations: the company had collected persistent identifiers from viewers of child-directed channels and served them behavioural advertising without verifiable parental consent. The penalty was 170 million dollars — 136 million to the FTC, 34 million to New York — the largest COPPA settlement in the law's history at that point. The FTC's own write-up of the case is still the clearest short statement of what the law expects.

YouTube's response was structural. Rather than classify every video itself, it pushed the determination down to the people making the content and built a platform-wide switch behind the answer. That is the box on your upload page. It is not YouTube asking your opinion about your audience; it is YouTube collecting a declaration it can rely on, and passing you a share of the responsibility for it.

The regulatory floor has moved since. The FTC published substantial amendments to the COPPA Rule in April 2025, effective that June, with a compliance deadline of 22 April 2026: a broader definition of personal information, separate parental consent for disclosure to third parties, tighter security and retention obligations. None of that changes the box you tick, but it explains why the enforcement temperature around it has risen rather than fallen since 2019.

What "made for kids" actually means

The test is not "would a child enjoy this". It is a multi-factor test borrowed from the FTC's definition of child-directed content, restated in YouTube's own documentation. The factors:

• The subject matter of the video.
• Whether children are the intended audience, and whether they are the actual audience.
• Whether the video features child actors or models.
• Whether it features characters, celebrities or toys that appeal to children, including animated characters.
• Whether the language is pitched for children to understand.
• Whether it shows activities that appeal to children — play-acting, simple songs, games, early education.
• Whether it includes songs, stories or poems written for children.
• Any other objective evidence about who is watching, including whether the content is advertised to children.

The qualifier comes straight after the list: the presence of some of these factors does not automatically make a video made for kids. A Lego build with a forty-year-old presenter and a technical voiceover is not child-directed because Lego appears in it. The test asks who the video is for, and treats the factors as evidence towards an answer rather than as triggers.

The three cases creators get wrong

Family vlogs. Children appearing in a video is one factor, and on its own a weak one. A family channel whose real audience is parents — adult humour, adult pacing, the experience of raising children rather than of being one — is generally not child-directed. A family channel whose videos are children playing, filmed for children to watch, is. The question is who is meant to be laughing.

Gaming. The largest grey area on the platform, and it splits by game and by treatment. Commentary-led play of a game with a broad audience, aimed at people who play it themselves, is general-audience. Wordless, brightly packaged play of a game whose player base is overwhelmingly young, with cartoon characters front and centre and nothing a teenager would find interesting, hits nearly every factor in the list at once.

Animation. Animation is not a genre, and the factor list says "animated characters that appeal to children", not "animation". Adult animation is plainly general-audience. Simple, brightly coloured animation with nursery-rhyme pacing is not, whatever the creator's intent was.

Channel level, video level, and the machine that overrides both

There are two places to answer the question. The channel-level setting in Studio's settings panel sets a default for everything you upload, and can declare that the entire channel is directed at children. The video-level setting on the upload page overrides that default per video, unless you have set the whole channel as made for kids, in which case every video inherits it.

Setting the whole channel carries a wider set of removals than setting a single video — YouTube's documentation lists the channel-level Posts tab among the features that go. If your channel makes nothing but children's content, that is the honest declaration and it saves you from the one upload where you forget. If your output is mixed, leave the default at general audience and answer per video: the failure mode of a wrong per-video answer is one video, and the failure mode of a wrong channel default is your entire catalogue.

Behind both settings sits a machine learning system that reads uploads for content clearly directed at young audiences, and it can override your answer. When it does, the setting in Studio reads "made for kids — set by YouTube" and you cannot change it back; the route is the feedback button, not the dropdown. YouTube is explicit that you should not rely on that system to make the determination for you, because it may not catch content the FTC would consider child-directed. In other words: the classifier is a backstop against error and abuse, not a second opinion you can wait for.

What it reads matters for the rest of this piece: the upload as published, which includes the title, the description and the thumbnail. Packaging is not decoration to a system like this. It is evidence.

Everything the box switches off

This is the part most creators discover after publishing rather than before. Marking a video made for kids does not adjust it — it removes features, permanently, with nothing anywhere that restores them.

What goes	What that costs you
Notification bell	No push to your subscribers when you publish. Subscribing to the channel stops meaning what it means everywhere else.
End screens and cards	The engineered second click is gone. Nothing at the end of the video points at another one.
Channel watermark	No in-player subscribe affordance.
Playlists and Watch Later	Viewers cannot queue you. Parents cannot build a shelf of your videos to come back to.
Autoplay on home	The preview that sells a video in the feed does not play.
Comments	No community, no engagement signal, and no read on what landed. This one is not optional even if you set comments to allow all.
Personalised advertising	Contextual inventory only. The highest-bidding half of the auction is not in the room.
Channel memberships, Super Chat, Super Stickers, Super Thanks, merch and ticketing, the donate button	Every non-advertising revenue line the platform offers, in one go.
Live chat and live chat donations	Live streams for children are broadcast-only.

Read that as a whole rather than as ten separate inconveniences. Notifications, end screens, cards, playlists, autoplay and comments are not features; they are the retention layer, the mechanisms that turn one view into a second and one viewer into a returning one. Kids content operates without any of it.

What is left is a grid of thumbnails

Strip out everything in that table and count the routes a viewer still has to a made-for-kids video. There are four: home and browse, search, suggested videos, and the YouTube Kids app. Every one presents your video the same way — a rectangle of artwork with a line of text under it, in a grid of competitors doing the same.

This is the structural fact that gets missed. On a general-audience channel the thumbnail is the most important single asset but it is not the only one: a subscriber can be notified, an end screen can catch someone at the end of a video, a playlist can carry a viewer through six uploads without a decision, and the comments give the recommendation system an engagement signal to work with. Remove all four and the thumbnail is not merely the most important asset. It is very close to the only one. The traffic source mix on a kids channel collapses towards the artwork-driven surfaces because the others have been switched off at the platform level, and the second click you would normally engineer has to be won the same way the first one was — by being the best-looking tile in a grid.

Compounding it: revenue per view is lower, so you need more views for the same money, so the click-through rate you need is higher rather than lower. Both pressures point the same way. Packaging is where a kids channel is won — and it is the part most kids channels treat as an afterthought, competing against other channels that also treat it as an afterthought.

The revenue arithmetic, honestly

You will find confident RPM tables for kids content online. Ignore them: they disagree with each other by factors of three or more, none names a methodology, and revenue data never leaves the Studio of the channel that earned it — the same problem that makes every niche pay-per-thousand-views chart guesswork.

What can be said is mechanical rather than numerical. YouTube sells advertising through an auction. On general-audience content, advertisers can bid using behavioural data about the individual viewer, and that data is what they are paying a premium for. On made-for-kids content that bidding is not available: ads are served contextually, matched to the video rather than the person. The 2019 case is the clearest evidence of what the difference is worth, because behavioural advertising against child-directed channels was the entire practice the FTC went after. Removing it removes the premium.

On a general-audience channel, the standard answer to a thin ad RPM is to earn elsewhere: memberships, Super Thanks, merchandise, sponsorship. On kids content the platform has switched off every one of those except sponsorship, and the YouTube Kids content policies restrict overtly commercial material, which narrows sponsorship too. The realistic model is advertising revenue at a discount, multiplied by volume, plus whatever can be built off-platform — a harder business than most creator advice assumes, and a reason to spend more on the packaging rather than less.

The quality principles invert the clickbait playbook

YouTube maintains a set of quality principles for children's and family content, developed with child development specialists. High-quality content, in their framing, is age-appropriate, enriching, engaging and inspiring; the published themes include being a good person, learning and curiosity, creativity and imagination, engagement with the real world, and diversity, equity and inclusion.

For a long time those principles governed recommendations and inclusion in the YouTube Kids app but not money. That changed in late 2021, when YouTube told creators it would begin applying them to monetisation: an individual video judged low-quality could receive limited or no ads, and a channel found to be predominantly low-quality could be suspended from the Partner Programme. The two named categories were heavily commercial or promotional content, and content encouraging negative behaviours or attitudes.

Now put that next to the YouTube Kids content policies, which name the packaging directly. Misleading titles and thumbnails are prohibited. Sensational titles and thumbnails are prohibited. Keyword stuffing in titles is prohibited. There is a specific named category — deceptively educational content — for packaging that promises to teach colours or numbers and then does not.

Read those together and the standard high-CTR playbook is largely prohibited in this category. The shocked face, the impossible promise, the number that overstates, the curiosity gap the video never closes: every one of those extracts a click from an adult who can evaluate the claim afterwards and decide they were had. Applied to a five-year-old, the same technique is what the policy is written to stop, and the enforcement runs through your ad revenue and your eligibility for the Kids app rather than through a strike you would notice. What that leaves is the honest half of the craft — contrast, composition, a recognisable subject, a promise the video keeps — which was doing most of the work anyway.

Designing for an audience that cannot read

The most useful thing to know about packaging for young children is that much of the intended audience cannot read the title, cannot read the text on your thumbnail, and often is not the person holding the device. That reorders every priority.

Text is decoration, not information. On a general-audience thumbnail, two or three words add a promise the image cannot make. On a preschool thumbnail, words are read by the parent scanning the grid and ignored entirely by the child. Keep them for the adult — a series name, an episode number — and never let them carry the click alone. The size the letters need to be is unchanged; the amount of work they do is not.

The subject has to be recognisable at tile size. A child re-finds a video by its picture, which means the picture has to survive the shrink to a phone-sized tile and still be identifiable as the thing they want. One large, centred, unambiguous subject beats a composed scene every time in this category, and the margin is wider than it is for adults.

Series consistency is worth more here than anywhere else. Elsewhere a house style is a branding decision with a real cost, because a wall of identical tiles on your channel page is a problem. For a young audience it is a navigation system: if every episode shares a colour, a frame and a character position, a child who cannot read your titles can still find the sixth one. Vary the subject inside the frame, keep the frame.

Colour still has to be the odd one out. Kids content converges on saturated primaries, which means saturated primaries are the camouflage in this feed rather than the differentiator. The principle in the colour guide holds and matters more: what gets noticed is the tile that does not match its neighbours. A consistent, slightly unusual palette held across a series will outperform another set of red-and-yellow rectangles, and it doubles as the recognition cue in the previous point.

Check it in the grid, not on the canvas. All of that is invisible at editing size and obvious at feed size. Drop the candidate into a thumbnail preview beside the videos it will actually sit next to before deciding it works.

The reverse problem: looking child-directed when you are not

The classifier reads packaging, and packaging is easy to get wrong in the other direction. A general-audience channel can end up with a video marked "made for kids — set by YouTube", losing its comments, its end screens and its personalised ad revenue overnight, because the artwork and title read as child-directed even though the video is not.

The signatures are predictable: cartoon characters filling the frame, toys as the main subject, primary-colour backgrounds with rounded bubble lettering, a child's face as the focal point, a title in the cadence of children's television. Any one alone is ordinary. Several together, on a video whose subject is also on the factor list, is a strong signal.

If you have been reclassified and believe it is wrong, the route is the feedback option in Studio rather than the audience dropdown, which will be locked. Alongside that, look hard at the packaging, because it is the cheapest thing to change and often the thing that caused it. A presenter's face in the frame, a title in adult vocabulary and a palette that is not nursery-bright all pull the reading back towards general audience, and none of them costs you anything if your audience really is adults.

The temptation runs the other way, of course — declare general audience, keep the comments and the ad rates — and that is the direction the FTC has now shown it will act on. But over-declaring is a real error too: marking a general-audience video "made for kids" to be safe hands away your notifications, your end screens, your memberships and most of your ad revenue for no legal benefit at all, because the declaration was never true. The setting is a question of fact, and the only safe answer is the accurate one.

What changed around this box in 2025 and 2026

The checkbox has not moved, but almost everything around it has, and the direction is consistent.

• April 2025 to April 2026 — the amended COPPA Rule. Published in April 2025, effective in June 2025, with full compliance required by 22 April 2026. It widens the definition of personal information and tightens consent, retention and security obligations.
• August 2025 — age estimation in the United States. YouTube began rolling out a machine learning system that estimates whether a signed-in viewer is under 18 from behavioural signals rather than the birthday on the account, applying teen protections including the removal of personalised ads when it concludes a viewer is a minor. Adults misidentified by it can verify with ID, a credit card or a selfie check.
• September 2025 — the Disney settlement. Ten million dollars, for designating child-directed videos as not made for kids. The first loud demonstration that the label is enforceable against the uploader rather than only the platform.
• December 2025 — Australia's under-16 rules. From 10 December 2025 Australian users under 16 were signed out of YouTube accounts. They can still watch; they lose everything that requires an account, including subscriptions, playlists and likes.

The pattern is worth naming, because it changes how much weight your declaration carries. Age used to be a single self-reported field, and the made-for-kids box was a single self-reported answer sitting on top of it. Both are now being cross-checked — the viewer's age by inference from behaviour, your audience declaration by a classifier reading your upload and, if it comes to it, by a regulator reading your catalogue. Answering the question carelessly is a worse bet each year.

There is a consequence here even for creators nowhere near children's content. Australia's rules mean a population of viewers now browses signed out, with no watch history and no subscriptions feed, and what a signed-out viewer sees is closer to a generic grid — which is another way of saying the subscriber relationship carries a little less of the load, and the thumbnail a little more.

A procedure for answering the question

1. Decide at the channel level first. If everything you make is for children, set the channel and stop thinking about it. Otherwise set it to general audience and answer per video. Mixed channels should never carry a made-for-kids default.
2. Run the factor list on the video, not the topic. Subject, intended audience, actual audience, child actors, child-appealing characters or toys, language, activities, songs and stories. Count how many apply, and be honest about the second one.
3. Read your own packaging as the classifier will. Look at the thumbnail and title alone, with the video hidden. Would a stranger seeing only those two things say this was made for children? If yes and the video is not, the packaging is the problem.
4. If the answer is yes, plan for the removals before you publish. No bell, no end screens, no cards, no playlists, no comments, no memberships. Build the series structure and the artwork system that replaces them.
5. If you are genuinely unsure, get advice rather than guessing. This is a legal question with a regulator attached, not a preference, and neither YouTube's documentation nor this article is legal advice.

What this adds up to

The made-for-kids setting is the only control on the upload page whose consequences are at once legal, financial and architectural — and when the honest answer is yes, you are running a channel with the retention layer removed, on ad inventory that pays a discount.

Which is why the conclusion for kids creators is the opposite of the one they usually reach. The response to losing the retention layer is not to accept a smaller ceiling; it is to put the effort that would have gone into end screens and community posts into the one surface still working. Four artwork-driven feeds and a Kids app whose automated filters read your thumbnail before anything else — that is the whole distribution system, and it is decided by a rectangle. The craft that wins it is not the clickbait craft, which the quality principles exist to stop. It is a recognisable subject at tile size, a series frame a child can navigate by, a palette that is the odd one out rather than another set of primaries, and a promise the video keeps.

Thumblore exists for that part of it — generating consistent variants fast enough that a series frame stays a system rather than becoming a chore by episode ten. If you are working out where your packaging currently sits, the honest starting points are what your click-through rate actually means and the design rules that survive at feed size, both of which apply here with the sensationalism turned off.