Three of the numbers YouTube Analytics shows for the same video now count its audience three different ways: the public counter, YouTube’s Views (Co-viewed) and Unique reach. The public counter on the watch page still counts one playback per device. A new metric called Views (Co-viewed) estimates person-playbacks instead: three people watching one television twice count as six. The figure that decides a creator’s pay does not move either way.
What Is Views (Co-viewed) in YouTube Analytics?
Views (Co-viewed) is, in the YouTube Help Centre’s own words, an “estimated total number of views, including multiple people watching together,” covering “shared viewing on devices like YouTube TV and the main YouTube app on TV.” The page’s own worked example: if a family of three watches a video together twice, that counts as six Views (Co-viewed). The entry does not explain how the estimate tells one viewer from three; the page’s own word for the figure is “estimated”, and it goes no further.
Public Views, Co-viewed Estimates and Unique Reach Diverge on One Video
The same three people on one television produce three different numbers. A third figure sits next to Views (Co-viewed) in the metrics list, Unique reach, described in the same estimated, TV-scoped terms: “estimated total number of individual people who saw your video or channel,” also covering “YouTube TV and the main YouTube app on TV.”
| Metric | Affects pay? | What it counts | Worked example (3 people) |
|---|---|---|---|
| Public views | No | One count per playback, from the first frame | 1 view, regardless of viewers |
| Views (Co-viewed) | No | Estimated views including people watching together | 3 people, 2 sessions = 6 views (Help page’s example) |
| Unique reach | No | Estimated individual people who saw the video | 3 on one TV = 1 view, reach +3 |
None of the three touches the engaged-view count that YouTube Partner Program pay is calculated on, or the qualified views that decide Partner Program eligibility. Public views changed once already this quarter, when YouTube moved to counting from the first frame on 24 August 2026. That change and this one are separate counters moving on separate dates.
Does the Published Material Give an Error Margin for the Estimate?
Neither the Help Centre entry nor the Creator Insider video’s description publishes one, nor a sample size, nor a description of how the model separates one viewer from three. According to PPC Land’s account of the video, titled “New! Amazon Tags in YouTube Shopping, Views (Co-Viewed), Branded Content Updates” and dated 4 September 2026 on YouTube’s own page, the estimate rests on statistical modelling based on demographic patterns, video genre and viewing times.
PPC Land, which describes the video as hosted by Rene Ritchie, YouTube’s creator liaison, reports that Views (Co-viewed) sits alongside Unique reach in Analytics’ advanced mode starting 31 August 2026, and that the figure needs up to 48 hours to process. Any date range touching the last two days will therefore read low until it settles, according to that same account. None of those three details, the 31 August start, the advanced-mode location or the processing lag, comes from the Help Centre page itself.
Co-viewing Now Runs on Both Sides of YouTube
On 27 May 2026 the Google Ads Developer Blog said the Google Ads API would “transition reach metrics to a ‘Total Co-view’ definition” from 2 June, covering “all individuals who viewed the ad on connected TV devices, including cases where multiple people are watching YouTube together on the same TV screen.” On that advertiser side, Google says it measures co-viewing with “a combination of Google’s connect TV panel data and real-time, census-level surveys” across more than 100 countries, according to Google Ads Help.
Creators get a different description for what looks, from a media kit, like the same concept. Google’s advertiser documentation describes one process; the only description of the creator-side model comes from PPC Land’s account of the video, which ties Views (Co-viewed) to demographic and genre modelling rather than panel-and-survey research, and YouTube’s own creator Help entry describes none. Television ratings went through their own co-viewing revision days earlier: Nielsen’s Big Data + Panel currency changed how it folds data from wrist-worn wearables into co-viewing measurement, one of seven changes effective 31 August 2026.
What Should a Sponsorship Buyer Do With Views (Co-viewed)?
Not treat it as a bigger version of the number they already had. A creator can now hand a brand an estimated figure larger than the public view count without anything a sponsorship rate is actually priced on having changed. Dividing Views (Co-viewed) by public views on a given video would produce a per-video co-viewing ratio; that is our own way of framing what the two counters imply side by side, not a figure YouTube or anyone else publishes.
No audit of that ratio has been published, by YouTube or anyone else. A buyer comparing two media kits, each quoting Views (Co-viewed) instead of the public count, is comparing two unaudited estimates rather than two measurements of the same thing. YouTube frames the scale of the underlying behaviour with a figure PPC Land attributes to the company: “over one billion hours of YouTube on television screens every day”. What the Help Centre entry and the video description both leave out is how confident the per-video estimate is.