In partnership with
If you make videos, you've probably spent a lot of time thinking about views. How many, how fast, and whether the algorithm liked it.
But something is changing quietly!
More people now skip the search bar and ask an AI assistant instead: "Which one should I buy?" or "Is this actually worth it?"
Those answers don't come from nowhere. The AI pulls from content that already exists, and a lot of it is video. So here's the question every creator should be asking: whose videos is it pulling from?
A new study went looking, and what it found is more encouraging than you might expect.
What the study looked at
The research comes from Sundogs, a creative intelligence firm. They analyzed 3,900 videos in US consumer goods and cars. Of those, 400 had been cited by AI assistants (provided by marketing firm Jellyfish from its tracking platform), and 3,500 had not. All came from the same creators.
That last detail matters. Since the creators were the same, the comparison is about the video itself, not who made it.
One thing to keep in mind as we go. This is a single study, and it covers only two categories.
So who is AI actually quoting?
Mostly creators. About 60% of the cited videos came from creators, around 30% from editorial sources like publications, and just 10% from brands.
A separate Jellyfish analysis, reported by Adweek, points the same way. Across 27 million responses from seven AI assistants, YouTube creator content appeared in over 25% of answers. Independent, niche creators outranked both brand-owned content and celebrity influencers.
Both studies come from marketing firms, so treat them as strong signals rather than final proof.
What didn't seem to matter
This is the part that might be a relief, especially if you've ever stared at a low view count:
Views. Some cited videos had under 1,000 views, while higher-traffic videos from the same creator went uncited.
Age. Many cited videos were one to two years old, or older.
Polish. AI voiceovers and stock footage showed up often in cited videos.
That last point isn't permission to cut corners. It just suggests these systems care more about what a video says than how it looks.
What cited videos did have in common
So if it isn't views, age or polish, what is it? Specifics. Among the cited videos:
77% were formal reviews or hands-on demos
76% named the problem the product solves
90% stated clear benefits
71% compared one brand with another
84% had a practical, confident tone
Two smaller details stood out too. Mentioning specific ingredients or science was linked to a 30 to 34 point lift in citations. A soft "check it out" ending added about 21 points.
If that sounds like the long, detailed "here's how these two actually differ" video you almost didn't post because it felt too niche, you're not imagining things. That's exactly the format the study describes.
And what about Shorts?
Length plays a part here. Cited videos were commonly eight minutes or longer, and Shorts were cited at a fraction of the rate of long-form.
That doesn't make Shorts pointless. They may just be doing a different job from the videos AI reads.
What you can try this week
You don't need to change your whole approach. A few small shifts are enough:
Open your next video with the problem the product solves.
State the benefit plainly, then compare it with an alternative.
Add the specifics: names, ingredients, numbers.
End with a soft "check it out" instead of a hard sell.
Ask an AI assistant a buying question in your niche and see which videos it cites.
One last thing to keep in mind
The traits above are linked to being cited. They don't guarantee it. Results for food, beverage and retail looked more mixed, and the study is US-based.
Sundogs' founder also argues creators should be paid when their videos are cited. That's his view, and there's no pricing standard for it yet.
Still, the takeaway is hard to ignore. The detailed video you made for a small audience may be doing more work than its view count suggests.
Why Most AI Tools Fail at B2B Customer Service
B2B customer issues rarely resolve in a single interaction. They move across teams, systems, and account history that most AI tools were never designed to track.
A new briefing paper from Harvard Business Review Analytic Services, sponsored by Front, looks at why AI built for high-volume consumer support creates new coordination gaps when it meets the multi-team reality of B2B service, and what leaders should ask before investing in their next AI tool.
Get the free briefing paper to see the three questions that separate tools built for your environment from tools that will create more work.
