.png)
What’s actually changing inside Google’s AI Overviews right now?
If you spend time around search teams, there’s a sentence you hear more and more often:
“I saw it in the AI Overview, not in the blue links.”
For a lot of queries, Google is no longer just showing a list of websites. It’s giving you a short answer at the top, generated by an AI model and backed by a handful of citations underneath. Those little links are effectively the “sources” Google is leaning on to build its answer.
At Omnia, we’ve been watching those sources quite closely.
Over the last few months, we looked at around 3.5 million citations inside Google’s AI Overviews. We simply wanted to see which domains kept popping up, and how that mix was changing over time.
One pattern stood out.
In July, less than 1.5% of all citations in AI Overviews pointed to YouTube. It was there, but fairly marginal. By mid-September, that number had climbed to just over 3%. Since then it has been sitting somewhere between 3% and 4.5%, rather than dropping back down.
In absolute terms, that’s still a small slice of the pie.
But in relative terms, it’s a big move in a short period of time: YouTube’s share of citations more than doubled in a few months.
That raises a few simple questions:
- Why is YouTube being cited more often?
- What kind of queries seem to trigger those citations?
- And, more practically, should content and SEO teams do anything about it?
The rest of this piece is an attempt to unpack those questions, using both our own data and what we see more broadly in how people search.
Why would Google’s AI Overviews start citing more YouTube content?
We obviously don’t know at all what’s happening inside Google’s systems. All we see is the output on the page.
But there are a few reasons why it’s not that surprising to see YouTube cited more often.
The first one is simple: a lot of the questions that trigger AI Overviews are practical.
How to do something.
How to compare two options.
How to fix a specific problem.
That’s exactly the type of query where YouTube has been strong for years. Tutorials, product demos, walkthroughs, reviews… if you’ve ever tried to assemble something or debug an error, chances are your first instinct was to search on YouTube.
The second reason is that YouTube is not just video. Each video comes with a title, a description, often chapters, and in many cases an automatic transcript. For an AI model, that’s a lot of structured text around a single piece of content. It makes those pages easy to understand and easy to reuse.
The third is about trust and signals. Google has an enormous amount of behavioural data around YouTube: watch time, engagement, likes, skips, and so on. Even if we never see those numbers, they are useful hints about which videos people actually find helpful. If you’re trying to pick a few sources to build an AI answer, it’s natural to lean on the parts of the web where you have the clearest quality signals.
Put together, it paints a simple picture:
for certain types of questions, YouTube is a very convenient place for an AI system to look.
That doesn’t mean YouTube will replace written content or that it’s suddenly the only thing that matters. But it does explain why, once AI Overviews became a standard part of the results page, YouTube started to move from “occasional citation” to “regular presence” in the sources behind those answers.
What does YouTube’s rise in AI Overviews mean for content and SEO teams?
If YouTube is being cited more often, the easy reaction is:
“Let’s start making videos for AI Overviews.”
That’s probably the wrong starting point.
A more useful question is: which parts of our content would make more sense if they also existed as video?
Not everything needs a video. But there are a few formats where it fits almost naturally:
- How-to and walkthroughs – showing a workflow or product in action.
- Comparisons and reviews – explaining trade-offs between options.
- Onboarding and “getting started” – where people want to see screens, not just read steps.
From a GEO perspective, the logic is simple: if YouTube is showing up more often among the sources AI Overviews use, and your best pieces only live as text, you’re competing in fewer places than you could.
It’s not a call to rebuild your whole strategy around video. It’s a reminder that some of your key pieces may have more impact if they also exist as a quick, clear video on YouTube.
Remember, this is new for everyone, so there’s no exact strategy to follow yet. We’re still learning from the data as we go.
Should you change your AI strategy just because YouTube is cited more often?
Probably not. At least, not if you think about YouTube in isolation.
When we look at the broader picture of AI Overviews, YouTube is only one of several sources that keep showing up. In the same charts where YouTube grows, you also see domains like LinkedIn, Reddit, documentation pages, review sites and big publishers gaining or keeping a strong presence.
In other words, AI Overviews are not “moving to video”.
They’re leaning more on a small group of trusted surfaces. And YouTube happens to be one of them.
What that means for you depends a lot on what you sell and how people make decisions in your category:
- If you sell a physical product, YouTube might be a natural fit: reviews, unboxings, side-by-side comparisons, “does it really work?” type videos.
- If you’re in B2B SaaS, LinkedIn posts, in-depth docs, comparison pages and trusted review sites may matter more than video for many prompts.
- If your users live in communities, Reddit threads or niche forums can have as much weight as your own blog.
From a GEO perspective, the question is less “Should we bet on YouTube?” and more:
For the prompts that really matter to us, which surfaces does AI already trust and how do we show up there in a way that makes sense for our brand?
For some teams, that will mean adding video on top of existing content.
For others, it will mean doubling down on LinkedIn, strengthening documentation, or working more closely with review and comparison sites.
YouTube is an option. It’s not the default answer.
How can I know where to create content for AI?
If you want to see this in your own space, you can sign up for free in Omnia here and see the top sources in your category.
You’ll know which pages AI is actually using when it recommends products like yours, and you can start optimising content with data, not assumptions.