Changed 24 August 2026
Your views went up. Nobody watched more.
YouTube now counts a view the moment a video starts playing. Paste a link and see what it would have counted last month.
If this worries you
Five things worth doing this week.
Draw the line at 24 August
Mark it on every chart you keep. A chart that crosses that date is mixing two different things. In six months nobody will remember why the line jumped.
Make engaged views your headline
Studio, Analytics, Advanced mode. It kept the old definition, so it is the only view number that runs clean through the change.
Re-baseline before judging a new video
Do not hold a September upload against an August one on raw views. Convert it, or compare engaged to engaged.
Fix the media kit first
If your deck quotes lifetime or 90-day views, it now mixes two definitions. Footnote it before a buyer asks.
Watch retention instead
How long people watch is built on engaged views, so it did not move. If that stays flat while your views climb, nothing real changed.
The change
One word moved. Every number followed.
A view used to mean somebody started watching and stayed. Now it means the player started.
YouTube never said what the old rule was. People repeat “30 seconds,” but that number is nowhere in YouTube’s own help pages. Here is what we can show. Before the change, the public count and the engaged viewscount were the same number, give or take half a percent, across 26 videos.
Why
They gave up an argument they won in 2015.
In 2015 Facebook counted a view from the first frame, autoplay on. Scrolling past counted. Creators called the numbers fake: Hank Green wrote the essay, and the BBC quoted Tom Ridgewell calling them “not true statistics.”
YouTube’s answer was that its own views meant something. Somebody clicked and stayed. It sold watch time on that basis for a decade.
Ten years on, TikTok and Instagram count from frame one, and the same clip kept looking smaller here. Shorts moved in 2025. This finishes the job.
YouTube gave its reason in the announcement itself. The last line is the one that matters.
We’ve heard that creators want to eliminate this metric confusion and accurately understand their true exposure … We hope that having metric parity across all formats gives creators a clearer picture of their overall exposure and makes it easier to confidently express their true scale and value to brand partners and sponsors.
So: stop the confusion, show creators their real reach, and help them sell. YouTube’s creator liaison told Business Insider the same thing. The change should help with brand deals, because it gives creators a better sense of how many people they are reaching.
One language
A YouTube view now means what a TikTok view means. Nothing left to translate in a media kit.
Bigger reach to sell
YouTube says so itself. It is easier to show sponsors how big you are. It also makes every channel look bigger to buyers. And it costs YouTube nothing.
Tidying up
Shorts already worked this way. Long-form and live were the odd ones out inside the same app.
First thing everyone asks
Is it counting people?
No, and it never did. A view is a playback, not a person. The change widened the gap between the two.
Rewatches count?
Yes. One person watching three times is three views. If you want people, the metric is Unique viewers, in Advanced mode.
Can I farm it by refreshing?
No. YouTube drops plays it thinks are fake. It names both tricks: the same video open on lots of devices, or in lots of tabs.
Do Shorts loops count?
Not as engaged views. Google excludes loops there. The raw counter is the looser one, which is where the gap comes from.
A room around one TV?
One view, but unique reach goes up by three. Undercounts there, overcounts elsewhere. It was never a headcount.
Money
What it does to a brand deal.
Your payout did not change. What you can ask a sponsor for did.
Creator
A flat fee priced off last quarter just got easier to hit. Do not plan around the bigger number, though: creators negotiating through the change say sponsors switched to engaged views within the week. Lead with engaged views and watch time.
Brand
Every YouTube deck is about to show a jump that is not growth. Check whether the date range crosses 24 August, then ask for engaged views, watch duration and comments.
Partner Program earnings and eligibility did not move. Both still run on engaged views.
“But my RPM dropped.”
RPM means the money you make per thousand views. Your money did not change. Your engaged views did not change. But the view number on the bottom of that sum went up. So the RPM YouTube prints goes down while the same amount lands in your bank. Your pay did not break. The shortcut you used to guess it did. If 20,000 views used to mean about $100, use engaged views instead and the old math works again.
Our numbers
Google gave the date. Nobody gave the size.
So we measured it two ways that do not depend on each other. Engaged views kept its old meaning and arrives in the same API call as raw views, so the ratio between them sat flat until the day the rule changed. That only works on a channel you own. For everyone else, a mature video’s daily gain decays smoothly, so a step up in it is the change.
| YouTube format | How we measured | Videos | Result |
|---|---|---|---|
| Long-form, our channel | engaged ÷ raw | 26 | 0.996 → 0.472 2.11× |
| Long-form, other channels | daily step | 293 | median 1.68 1.83× |
| Live | daily step | 61 | median 1.34 1.46× |
| Shorts | daily step | 235 | median 0.92 no change |
The Shorts row is the control. Same platform, same window, same query, and no movement. If something had simply sent more traffic to YouTube that week, Shorts would have moved too. Only the formats YouTube named changed.
We ran the same test on 13 TikToks and 7 Instagram posts. Both sat flat, at 0.85 and 0.56. They rule out a bad week for video in general, nothing more.
Google’s own documentation held up too. Minutes watched per raw view halved. Minutes per engaged view did not move. Retention did not move. Revenue per engaged view stayed flat while revenue per raw view halved.
About the date. Google announced 24 August; our data breaks on 27 August, on two separate measurements. YouTube posted again in the same thread to say the rollout had finished, and that post is dated 27 August. The announced date started a staged rollout. If your own numbers moved on a different day, that is why.
How this compares to the only other published data
One other outfit has published numbers. Agentio modelled 35,800 long-form videos and estimated views would land about 30% above engaged views. We measured 2.11×. That is a wide gap.
The gap is about method, not who is right. From outside a channel you cannot see engaged views at all. So they guessed, using watch-time curves and the old 30-second rule. That is the same 30 seconds YouTube has never written down. Their own report calls the number a rough guide. It also says their videos come from big professional channels, which hold viewers better than most, so their guess is probably low. The one number they did measure straight, on Shorts, came out at+65%.
We did not assume a threshold. Engaged views arrives from the API next to raw views, so on a channel we own the ratio is read rather than modelled. Our own weakness is plain: 26 videos on one channel is a narrow base. Treat both figures as a range. Your exact number is in your Studio.
Where this could be wrong, and why we run on it anyway
Every number on this page is our calculation, not a figure YouTube published. Google gave the date and the direction and has never given the size. So here is the case against us and the case for us, side by side, and you can decide.
Where it could be wrong
- The strong number rests on 26 videos on one channel. Ours. A different audience that quits faster or stays longer would sit somewhere else.
- The other 293 videos were measured a different way. From outside a channel nobody can see engaged views, so that figure is inferred from a step in daily gain, not read. It landed at 1.83, ours at 2.11, and we do not pretend those are the same kind of evidence.
- Anything posted before 24 August needs a model. Only part of its count ran under the new rule and the public counter never shows the daily curve, so how much is estimated. That is the roughest number we print.
- Nothing external confirms us. The only other published estimate says about 30%. We say about 110%. One of us is wrong, or the truth moves by audience.
Why we run on it
- It is read, not guessed. Engaged views comes back in the same API call as raw views, so on our own channel the ratio is a measurement. We never had to assume the old 30 second rule that YouTube has never written down.
- Every video moved, and none held. All 25 with data on both sides fell. Of the 12 carrying 50 or more views each side, 0.99 to 1.00 became 0.41 to 0.54.
- The control stayed flat. Shorts, same platform and same week, did not move, because YouTube moved Shorts to this rule back in 2025. A good week for video would have moved them too.
- Google’s own documentation predicted four things and all four hold. Minutes per raw view halved, minutes per engaged view did not move, retention did not move, and revenue per engaged view held flat while revenue per raw view halved.
- We publish the middle, not our best number. Two independent methods landed 1.83 and 2.11, so the page says 2 and shows the spread rather than quoting the one that flatters us.
Which is the long way of saying: this is a good estimate for a video you do not own, and a poor substitute for the exact number if you do. If it is yours, go read Engaged views.
Resources
Ten links, and why each one.
Most coverage restates the announcement. These do not. The first three are the whole story if you only have five minutes.
Straight from YouTube
- The announcement itselfFour hundred words, and every other article is downstream of them. The follow-up further down the thread is what confirms the rollout finished on 27 August.
- Help Center: views and engaged viewsThe shortest correct statement of what changed, and the line saying earnings and Partner Program eligibility did not.
- YouTube’s blog: what engaged views areThe only place YouTube states that CTR, average view duration and retention still run on engaged views, and that recommendations did not change.
- Google Ads: how ad views feed the public counterWritten for buyers, not creators, which is why it is candid. It also quietly notes that bumper and non-skippable ads now increment the public count when only skippables used to.
If you want the mechanics
- PPC Land: explaining YouTube viewsThe most complete piece written on this. Walks every threshold YouTube now runs at once, and kills the 30-second myth properly.
- vidIQ: the practical creator guideHas the exact click path for putting engaged views into your Studio dashboard, and is honest about what is still unknown.
- Agentio: the only published datasetThe +30% estimate we compare ourselves against above. Read the methodology note, not just the headline.
If you want the argument
- Business Insider: same platform, bigger metricsPeter Kafka, four minutes, and the only on-record quote from a YouTube exec about brand deals.
- Tubefilter: the sponsor-read problemMakes the point nobody else did: a sponsor read lands minutes in, so somebody who bounces on frame one was never reached by the ad you sold.
- The Verge: views are liesFrom 2025, before any of this. The best context you can read, because it shows a view was never a stable unit on any platform.
Two things to watch for while you read. People keep saying that just hovering over a thumbnail now counts as a view. Nobody has shown that it does. And people argue about whether this pulls your average watch time down. YouTube says no, because watch time is still built on engaged views. Some experts think it has to drop anyway. Nobody has proved it either way, so watch your own numbers.
We make video for a living, so when our own numbers jumped we went looking for why before anyone explained it. This page is what we found.
If you are making content and it is not landing, or you have a drive full of footage and no time to cut it, that is the work we do. Tell us what you are trying to make.
Work with us →And the show it came from
Long interviews with people who built something real. Including the parts they usually leave out.
The video in this page’s own example is one of ours. It went up four days after the change, which is why it makes such a clean before and after.
Watch the show →