Jonas, 38, is a product manager at a software company specializing in warehouse logistics. Two years ago, he recorded a 14-minute video titled: How warehouses can reduce picking errors. The video has 312 views. It stopped appearing in the company’s social media reports a long time ago.
Then, a buyer at a furniture retailer searches Google for exactly that question. The AI Overview provides an answer and cites Jonas’s video.
40.83 percent of YouTube videos cited by AI have fewer than 1,000 views (source). In other words, a video can be unremarkable by traditional social media standards and still become relevant as a source for an AI-generated answer.
Other data also shows that YouTube is gaining importance overall. The platform’s share of citations in Google AI Overviews increased by 34 percent within six months (source). In Google AI Mode, YouTube’s share of citations more than quadrupled between January and April 2026 (source).
This gives YouTube a second function: In addition to being a reach channel, the platform is becoming a potential knowledge source from which AI systems assemble answers.
This is not only interesting from a technical perspective. Google’s AI Overviews reach more than 2.5 billion people per month (source). Which sources appear there therefore influences which information about brands, products and topics becomes visible at scale.
“YouTube is evolving in AI search from a pure reach channel into a knowledge source. For GEO, this means that visibility is not created only where content generates a lot of views, but also where AI systems select it as a relevant reference for their answers.”
Miriam Schwellnus, expert in PR in the GEO era at Mashup Communications
However, this development does not apply equally to every AI platform.
If the buyer from our example asks the same question in ChatGPT, the likelihood of seeing a YouTube source is significantly lower. In one study, only 0.99 percent of ChatGPT citations pointed to YouTube (source).
For companies, the key takeaway is that YouTube visibility is not a universal lever across all answer engines. Anyone discussing GEO therefore also needs to consider which platform their target audience actually uses for search.
Yet even within the same platform, the system alone does not determine whether video is relevant. The type of question matters just as much.
The data available so far suggests that YouTube appears particularly often as a source in AI search when an answer benefits from explanation, demonstration or concrete product experience – for example, tutorials, product demos, reviews, or pricing and comparison questions.
For more abstract topics such as strategy, careers or broader professional questions, other types of sources tend to play a larger role. In tests involving typical professional queries, for example, LinkedIn was the most frequently cited domain across six major AI platforms (source).
This distinction is especially important for B2B companies. The question is therefore not only: “Is YouTube relevant for GEO?” but rather: “For which questions asked by our target audience is video even a plausible source?”
That may be the case for Jonas’s tutorial on reducing picking errors. For an abstract forecast about the future of the logistics industry, however, his video is more likely to compete with trade publications, corporate websites, studies, LinkedIn posts and other sources.
The importance of video therefore does not necessarily depend on whether a brand operates in a B2B or B2C market. It depends on the information need behind the query.
And even if video is selected as a source for such a query, a second question remains: What does this visibility actually deliver for the company?
A citation in an AI-generated answer does not automatically mean that users will click through to the source.
When Google displays an AI Overview, users click on one of the cited sources in only around 1 percent of cases (source). This creates a measurement problem for companies: A piece of content can provide information that appears in an AI-generated answer without generating meaningful referral traffic.
Views, clicks and watch time remain important metrics for social video. But they no longer tell the whole story if videos also function as knowledge sources for AI systems.
This inevitably raises the question of how this second type of content value can be measured at all. Which metrics make sense for this purpose, and how brands can systematically measure AI visibility, is increasingly becoming a discipline of its own.
A recent example from the UK shows that companies are no longer simply observing this development but are beginning to align their content strategies with it.
At the beginning of September 2026, department store chain John Lewis launched its own video podcast, “Gift List”, hosted by TV presenter Angela Scanlon and featuring Louis Theroux as the first guest. One stated goal of the format is to make the brand and its products more visible to chatbots and AI search.
The background: According to John Lewis, one year ago 0.3 percent of customers used AI chatbots to search for products. That figure has since risen to 2.5 percent (source). The starting point is small. Nevertheless, the development illustrates how perspectives on content are changing.
“Gift List” consists of six longer interview episodes published every three weeks on YouTube, the company’s own website, Spotify and Apple. Whether the format actually leads to greater visibility in AI-generated answers has not yet been proven.
That is precisely what makes the example interesting: A successful entertainment format and a useful source for a specific search query are not automatically the same thing. Celebrity guests can generate reach. Whether individual pieces of content later serve as references also depends on whether they contain answers to questions that people actually ask search engines and AI systems.
John Lewis is therefore less proof that a GEO strategy works than an example of how major brands are beginning to think about content in terms of machine discoverability as well.
For a long time, the basic assumption in video marketing was that a video is successful when people watch it, share it or engage with it. Generative search adds a second potential type of value. A video can provide information that a machine uses as a source for an answer – even if it looks unspectacular by traditional reach metrics.
That does not mean views are suddenly unimportant. Nor does it mean that every organization should now produce as many videos as possible for AI systems. It means something more fundamental: Companies increasingly need to think of content not only as a destination, but also as a source.
This expands the central question. It is no longer only: Are we reaching enough people with this video? It is also: Does this video contain information that could serve as a reference for a relevant question?
For Jonas, the video with 312 views still looks like a marginal topic in the social media dashboard. From the perspective of a generative search engine, however, it could still be one of his employer’s more useful knowledge assets.
YouTube is a frequently used source particularly in Google AI Overviews and Perplexity. However, the extent to which the platform is taken into account varies significantly between individual AI systems.
Not necessarily. In one study, 40.83 percent of the cited videos had fewer than 1,000 views. However, this does not mean that reach is fundamentally irrelevant when a source is selected in the first place.
Yes, but in the datasets examined so far, much less frequently than Google AI Overviews, Google AI Mode or Perplexity. In one analysis, only 0.99 percent of ChatGPT citations pointed to YouTube.
No. The data available so far suggests that video plays a larger role especially for tutorials, demonstrations, reviews, and product or comparison queries. For more abstract professional or strategic topics, other sources may be more important.
No. Users can receive their answer directly within the AI interface without opening the source. The value of AI visibility therefore cannot be assessed solely through clicks and website traffic.
Outrage has long been more than just a risk for brands: it can become a…
Discover the key aspects of successful collaboration with PR agencies, the use of storytelling and…
Social Media Storytelling in Football: How Professional and Amateur Clubs Grow Their Audience and Build…
In pop culture, artificial intelligences are always portrayed as villains. By emphasizing the human-machine interface…
Semantic drift – a shift in reputation resulting from an AI’s tendency to interpret context…
As a PR and brand storytelling agency, we help companies find their role as mentors…