How many YouTube views could creator integrations realistically bring us?
Built the classified video universe and the sizing model
2026
Scripted classification, video-platform exports, spreadsheet
The question
A creator-sponsorship programme on YouTube had grown by buying whatever fit. Before committing more budget, I needed three answers: how large the pool of high-intent videos really was, how much of it the programme already reached through creator integrations and paid placement, and what reaching more of it would cost.
The method
I classified every video in the universe by funnel stage and intent, using title rules plus a transcript check where one existed. I split the high-intent videos into a business-intent segment and a website-intent segment. For each segment I measured capture, the share of the segment's views that carried the programme's message. Creator integrations and paid placement are shown separately.
Views are not people, so I converted them in explicit steps. Each one was a stated assumption: the share of viewers still watching when the mention arrives, repeat views per person, relevance, geography and a recall rate. That gave exposed people, likely-aware people and a cost per aware person. I then built scenarios that scaled views up several times, with an assumed price rise at each step. They showed how cost per aware person behaves as the programme grows.
Last, I used the same funnel to ask whether the brand-awareness tracking survey could see the programme's effect at all. I worked out how many survey respondents would plausibly have been exposed in a single wave.
The evidence
| Finding | What the sizing showed | Reading |
|---|---|---|
| Universe size | Growing year on year. When I widened the classification it nearly doubled, so I re-ran the classifier and removed the new false positives | A sizing is only as good as its definition, so the definition has to be stated |
| Capture by segment | Business-intent capture rose more than tenfold from 2025 to 2026. Website-intent capture stayed flat and very small, in a pool several times larger | The programme had concentrated where it had creator relationships and barely touched the larger pool |
| Spend mix | Budget moved almost entirely from the website-intent segment to the business-intent one | Capture followed spend, not opportunity |
| Scaling | In the scenarios, raising views roughly two-and-a-half to seven times raised cost per aware person by about a twentieth to a third | Reach gets dearer as it grows. The question is where the curve bends |
| Measurement | Only a few dozen respondents per survey wave would have been exposed, and the largest possible lift in recall was well under one point against a margin of error of about five | The survey as sized could not be expected to show the effect. It would need a wave roughly eight times larger |
| Organic traffic link | Organic site traffic tracked YouTube views closely across six quarters, while search demand tracked them only weakly | Suggestive, not causal. There are six points, and both series moved with the season |
The result
The sizing changed the question from "buy more videos" to "where". It showed a high-intent pool that was largely untouched and a quantified cost curve for going further. It also showed that survey tracking alone could not prove the awareness effect, so funnel evidence had to carry the case.
I proposed a bottom-up follow-up as the next step, pricing what it would take to close the largest creators in the pool.