YouTube now counts a view from the first frame. Which of your video funnel ratios broke?

KEY TAKEAWAYS

1. YouTube's own ratios held. Any rate you compute with public views as the denominator fell on 24 August with no change in viewer behaviour.
2. Measured data puts public views at 1.67 times engaged views, almost all of the gap from feed autoplay.
3. Your inflation factor depends on playback location, so pull views and engaged views by location for each video.
4. Rebase link clicks, leads, subscribers and sponsor CPMs on engaged views, and mark 24 August as a break in every series.

Most writing on YouTube's new view count is about creators and sponsor deals. A brand that uses long-form YouTube as a funnel stage has a different problem. Every rate it builds by dividing by views, such as link clicks per 100 views or sign-ups per 1,000 views, fell on 24 August with no change in what viewers did. Q3 reports are the first to straddle that date. This is how to find which numbers moved and rebase them.

What changed on 24 August?

TeamYouTube announced on 17 August 2026 that public views would count from the first frame of playback in every format: long-form, Shorts, live and podcasts. It took effect on 24 August. Shorts had made the same move on 31 March 2025, per TechCrunch.

YouTube's 19 August blog post sets out the new terms. A view is any start, including a video that autoplays when someone hovers over it on Home. An engaged view is the old definition: a click or tap to watch, or playback past the opening seconds. A thumbnail shown with no play is still an impression, not a view.

Three details matter for reporting. Existing videos keep their totals, and only plays after 24 August use the new rule, per Forbes and vidIQ. The Analytics API changed its View metric on 27 August and kept engagedViews as a separate metric, per Google's revision history. And YouTube has never published the old watch threshold. The 30 seconds often quoted is a TrueView ad rule. YouTube only refers to the opening seconds.

Which numbers held, and which ones did not?

YouTube says most analytics stay anchored on engaged views, so impressions CTR, average view duration and retention are unchanged. Its help pages confirm that average view duration and unique viewers are computed from engaged views. In the API, card click rate divides card clicks by card impressions, so views play no part.

That leaves the ratios you build outside YouTube. They divide something you care about by the public view count, and the public count now includes plays nobody chose.

Early documents did not agree on this. In August, PPC Land and Search Engine Journal both read YouTube's glossary as saying engaged views equal views for long-form. The current help page now defines long-form engaged views as viewers who stayed past the opening seconds, not counting loops. The API changelog says the same. The newer documents settle it.

How big is the gap?

Estimates before the change were modest. Agentio modelled 35,800 long-form videos from June and July 2026, assumed a 30 second threshold, and predicted public views about 30% above engaged views. An executive quoted by TechTimes forecast 20% to 30%.

The measured figure is larger. Agentio's 3 September post covers more than 75,000 videos after the switch: 549 million public views and 330 million engaged views. That is a 60% engaged rate, or public views at 1.67 times engaged views. Tubefilter reported the same data as "over-report actual engagement by 40%". The 40% and the 67% are one gap expressed two ways, since 330 million is 40% below 549 million and 549 million is 67% above 330 million.

Why did the model miss? Agentio's own surface data points to the reason. Feed plays (Home, autoplay on Up next, recommended shelves) were 38% of public views, and only 20% of them became engaged. Watch page plays were 85% engaged. Feed autoplay produced 77% of the gap. Our reading is that a model built from retention curves could not see hover plays that were never recorded as views before. That would make the measured number the one to trust.

Two caveats apply. Agentio sells creator sponsorships and its sample is its own creator network, not brand channels. And CreatorDB, the other large dataset, published no before and after figures. Its Shorts versus long-form engagement comparison is observational and does not isolate the counting change.

Why your channel's factor is not 1.67

The gap varies by where plays happen. In Agentio's data the engaged rate ran from 54% for the smallest channels to 67% for the largest, and from 46% in technology to 63% in education. The surface split explains most of it. As a check, 38% of views at 20% engaged plus 61% at 85% engaged gives about 60%, which matches the total.

Take a video where 20% of public views are feed plays and 80% are watch page plays. Agentio's rates would imply about 72% engaged. That is an estimate for illustration. Do not rely on it. The Analytics API can report views and engagedViews by playback location (BROWSE, WATCH, SEARCH, EMBEDDED and others) and by traffic source, so you can measure your own rate per video.

Paid promotion adds a second distortion. YouTube's ads help page says public views from ads are now counted the moment an ad starts, and notes that previously only skippable formats added to the counter. Those public views are not billed. A promoted video's public count can therefore swell without a single extra billed view.

A worked example: two webinar replays

This example is hypothetical. It shows the arithmetic, not measured results.

A B2B software company posts a 40-minute webinar replay to YouTube each month. Its description link carries UTM tags, so GA4 records the sessions and trial sign-ups.

On public views, video B looks like 50% more reach and a one third fall in link performance. On engaged views, both videos did the same job. Viewers who chose to watch clicked and signed up at the same rate. A team reading the first set might rewrite a description that works.

The same arithmetic applies to sponsored videos. If a creator quotes a CPM on public views, the cost per thousand engaged views is the quoted CPM divided by the engaged rate. At Agentio's 60%, $30 becomes $50.

For what a replay viewer is worth further down the funnel, see webinar no-shows and replay viewers. The same denominator problem shows up in AI traffic, covered in AI referral conversion rates.

Does any of this matter if views were always a vanity metric?

Some commenters quoted by Influencer Marketing Hub said views never mattered, because pay and serious analysis run on watch behaviour. YouTube's position supports them on its own metrics. Agentio and Forbes argue that comparisons and pricing break, and EMARKETER calls the public count a weaker gauge of engagement. Both are right, about different numbers. YouTube's ratios are protected. A ratio in your own spreadsheet is not, unless you choose a protected denominator. Advice has not caught up. Directive's B2B YouTube guide, last modified on 2 October, still lists views and view-through rate as leading indicators with no note of the change.

A seven-step rebase

1. Mark 24 August in every report that charts views or a rate built on views. Note that older videos mix both counts after that date.
2. Add engaged views to YouTube Studio's Advanced Mode tables, and add the engagedViews metric to any API pull or connector.
3. List your view-based ratios. Include link clicks, leads, sign-ups, subscribers and any cost per view, in reports and in contracts.
4. Switch the denominator to engaged views for every ratio on that list. Restate Q3 and October on that basis.
5. Measure your own engaged rate per video by playback location and traffic source. Flag videos where feed plays dominate.
6. Split paid from organic before comparing videos, since ad starts now add to the public counter without being billed.
7. Rewrite sponsor terms to name engaged views, or convert any public view CPM by dividing by your measured engaged rate.

Long-form video is also one of the most cited source types in AI answers, which raises its own measurement questions. See which AI engines cite YouTube. If you want to track how your brand appears in AI answers, join the waitlist.

Questions

What is an engaged view on YouTube?

It is the old definition of a view. A viewer clicked or tapped to watch, or kept watching past the opening seconds. YouTube has not published how many seconds that is. Since 24 August 2026 it sits beside the public view count in YouTube Studio's Advanced Mode and in the Analytics API as engagedViews.

How much higher are public views than engaged views?

The only large measured sample is Agentio's. Across more than 75,000 creator videos after 24 August, 60% of public views were engaged, so public views ran 1.67 times engaged views. Its pre-launch model had predicted about 30%. Your own figure depends on where your videos are played, so measure it per video.

Did the change affect impressions click-through rate or watch time?

YouTube says no. Its August blog post says CTR, average view duration and retention stay anchored on engaged views, and its help pages say average view duration and unique viewers are calculated from engaged views. Card click rate in the API divides by card impressions, so it is also unaffected.

Do Google Ads video views follow the same rule?

Billing does not. Skippable in-stream ads still record a TrueView view at 30 seconds or the end. What changed is the public counter, which now ticks when an ad starts to play. Engaged-view conversions in Google Ads are a separate ad metric with their own 5 and 10 second rules.

Sources

1. YouTube Official Blog, What are engaged views on YouTube? blog.youtube
2. YouTube Help, Understand your YouTube content performance support.google.com
3. YouTube Help, How engagement metrics are counted support.google.com
4. YouTube Help, About YouTube ads and view metrics support.google.com
5. YouTube Help, Understand your YouTube video reach support.google.com
6. Google Ads Help, About engaged-view conversions support.google.com
7. Google for Developers, YouTube Analytics and Reporting APIs revision history developers.google.com
8. Google for Developers, YouTube Analytics API channel reports developers.google.com
9. Google for Developers, YouTube Analytics API metrics developers.google.com
10. Google for Developers, YouTube Analytics API dimensions developers.google.com
11. Agentio, New Agentio data. YouTube views are about to jump by about 30% agentio.com
12. Agentio, One week later. YouTube's new view count has a 60% engaged view rate agentio.com
13. Tubefilter, YouTube's new public viewcounts over-report actual engagement by 40% tubefilter.com
14. CreatorDB, The last honest engagement rate. YouTube redefines the view on August 24 creatordb.app
15. PPC Land, YouTube makes a view an impression across all formats from August 24 ppc.land
16. Search Engine Journal, YouTube changes how it counts views on long-form and live searchenginejournal.com
17. Forbes, YouTube changes how it counts views, handing marketers two numbers instead of one forbes.com
18. EMARKETER, YouTube's view count change makes engagement harder to gauge emarketer.com
19. Directive, YouTube for B2B guide directiveconsulting.com
20. TechTimes, YouTube splits its view count in two techtimes.com
21. TechCrunch, YouTube is changing how YouTube Shorts views are counted techcrunch.com
22. vidIQ, YouTube is changing how views are counted vidiq.com
23. Influencer Marketing Hub, YouTube's view count change will make views bigger influencermarketinghub.com

NEXT STEP

See where your brand stands.

Your Digital Authority Score, who AI recommends for your target questions, and the placements that close each gap. Your first report is free.

Get your free report →