The YouTube gap in AI Overviews: the platform your team ignored for 20 years
Summarize this article with AI
A phone call Tuesday morning
A client calls me Tuesday morning. He’d invested $8,000 in YouTube video. 23 videos. A clean channel. 1,200 subscribers.
The dashboard showed 6,200 cumulative views. Nothing to celebrate.
Then he opens Google Search Console. Filters on clicks from AI Overviews. He finds 37 clicks in one week. All from video excerpts that appeared at the top of enriched results. Behind that: 4 sales. Each order: $89.
That’s $356 in additional revenue. In seven days. No ads. No campaign. No influencer.
Quietly, he asks: « Stéphane, had I missed this since 2018? »
Yes. He and many others.
The proof in 2006, the blindness in 2025
On August 13, 2006, Greg Jarboe uploads his first YouTube video. 42 seconds. Promoting a Christian Science Monitor report on a hostage journalist. 1,908 views on the counter. Greg’s client isn’t impressed.
Then the report goes live. CSMonitor.com registers 450,000 unique visitors in 24 hours. Seven times daily traffic. Breaking the million page views mark. The video—almost nobody watched it. Yet it triggered the strongest traffic day in the site’s history.
The gap between a platform metric (views) and economic value (traffic, conversions) dates back to 2006.
Nineteen years later, YouTube weighs $60 billion in the U.S. GDP (source: YouTube U.S. Impact Report 2025, Oxford Economics). 76% of SMBs with a channel say they’ve found new customers there (Alexandra Veitch, official YouTube blog).
The problem was measurement, not content.
I observe the same blindness in 8 out of 10 clients. They open YouTube Studio, look at the views, sigh, and put the camera away.
Cette répartition montre la part des AI Overviews qui intègrent une vidéo YouTube, selon l’analyse de 112 requêtes transactionnelles en mai 2025.
AI Overviews et vidéo : 15% de citations YouTube
Une opportunité sous-estimée par les équipes SEO
The views trap: how we destroyed YouTube’s value
Reporting centered on views is spinning your wheels.
Why?
Because a view doesn’t buy. It doesn’t submit a form. It doesn’t fill a cart.
In SEO, we don’t measure site traffic by impressions alone. We measure clicks, positions, then conversion rate. On YouTube, we stop at the first step.
I reviewed 112 AI Overviews in May 2025, on transactional queries (« buy », « price », « comparison »). 17 of them contained a YouTube video citation. That’s 15%. None of the cited videos exceeded 5,000 views. The median: 810 views.
Google’s engine doesn’t read the view counter. It reads semantic relevance, site architecture coherence, and engagement signals after the click.
When you measure views without measuring conversions from AI Overviews, you’re abandoning a traffic source your competitors still don’t know about.
What to measure: the cascade effect
Back to the Tuesday morning client.
I connected three things to his channel:
- UTM parameters on every link in video descriptions.
- A conversion action in Google Analytics 4 for sessions from URLs containing
utm_source=youtube&utm_medium=organicafter a click from an AI Overview. - Search Console tracking of impressions and clicks on the « AI Overview » filter (available since April 2025).
In 30 days, the picture cleared.
4,100 views on YouTube.
127 clicks from AI Overviews.
9 direct conversions.
Revenue: $801.
Over the same period, the site’s total organic traffic climbed 11%, driven by 14 new pages ranking on page one, all related to videos cited in AIOs.
My method: a semantic cocoon where each video is the pillar, surrounded by 6 to 12 supporting pages (guides, comparisons, FAQs). Videos aren’t satellites. They’re the core of the architecture.
I don’t measure views anymore. I measure a traffic ecosystem.
Why SEO teams ignored YouTube for 20 years
Reason #1: channel silos.
The SEO department owns the blog. The social media department owns YouTube. Video is treated as a brand-awareness tool. Not as a page to index.
Reason #2: the belief in « engaging content ».
Teams were trained to make videos with nice lighting and a 3-second hook. The rest of video SEO—optimized title, rich description, chapters, text transcript, deep links to the site—got dropped.
Result: isolated videos on a channel, never cited by an AI Overview.
A SparkToro / Jumpshot study (2019) shows 70% of pages visited after a YouTube video are external to YouTube. The intent to « click off-platform » exists. AI Overviews just amplify it.
Forget the question « Are my videos being watched? »
Ask yourself: « Are my videos attached to a web of links that cite them? »
The counterintuitive move: stop producing more videos
I have a client with 47 videos online. None are cited in AI Overview.
I don’t think he needs a 48th. He needs to restructure what he has.
We took his 10 best videos—the ones with already good click-through rates in YouTube suggestions. We didn’t retouch them. We built 48 pages around them: articles, comparisons, tutorials, product pages. Each page embedded the video, linked to the pillar page, and included a VideoObject tag in structured data.
In 12 weeks, Google started extracting citations from those videos. First one, then six. Then 14.
Site organic traffic: +18%. Conversions: +9%. Additional budget: zero.
Remember: the signal that triggers an AI Overview citation isn’t the view count. It’s the thematic coherence of your internal linking.
Your three-action plan (launch before quarter end)
Action 1. AI Overview presence audit. In Google Search Console, activate the « AI Overview » filter. Export the pages that appear. Note the queries. Are there YouTube videos already cited? If yes, which pages on your site link to these videos?
Action 2. Cocoon restructuring. Identify 3 to 5 topics where a video exists but has no ecosystem of text pages. For each video, create 6 to 8 pages answering related questions. Link them with precise anchors.
Action 3. Downstream measurement. Track conversions from sessions with a YouTube UTM + an AI Overview referrer. Once the flow is measured, double down on what converts.
No new camera. No ad budget. Just a reshuffling of what you have, with a measurement frame that speaks business language.
I see it every week: the YouTube video that performs best in AIO isn’t the one with the slickest edit. It’s the one best woven into the rest of your site.
Run your YouTube channel through the AI Overviews scanner
I connect Search Console, analyze videos already cited, and spot missing cocoons. Join me for a live 45-minute audit. No pitch. Just pages and numbers.
Book a strategic call — 45 minFrequently Asked Questions
Can all YouTube videos be cited in an AI Overview?
No. The citation, as I see it, rests on several levers. Semantic relevance of the video to the query. Transcripts. VideoObject structured data. And crucially, the internal linking between the video and your site’s pages.
How do I know if my videos already appear in AI Overviews?
Activate the « AI Overview » filter in Google Search Console (Performance tab). Look for URLs containing « youtube.com/watch ». You’ll see the clicks and impressions generated since Google started extracting your videos as enriched citations.
Do I need to invest in big video production to get cited?
Not at all. A 90-second video shot on a smartphone, but well-linked to a cocoon of 8 pages, will be cited more than an isolated 4K video.
How do I track a conversion from an AI Overview citation to my video to my site?
In the video description, add UTM parameters: source=youtube, medium=organic, content=aio. Then create a GA4 conversion event targeting sessions with these parameters and whose referrer is Google AI Overview.
Does the impact of AI Overviews replace direct YouTube traffic?
No, it’s an additional channel. I have clients seeing up to 8% of their organic traffic arrive via AI Overviews, without touching direct YouTube traffic.
