AI Answers the Question. Creators Are the Reason There Is an Answer.

Everyone is optimising for AI visibility. Far fewer are asking where the AI gets its material. If you are launching something new, the honest answer is nowhere.

Becky Owen, CMO at Billion Dollar Boy, was asked on the Contagious podcast this week whether AI will take over product recommendations from creators. Her answer was not a defence of creators. It was a description of how the machine actually works.

“Using AI for product recommendations, it’s like trying to get on the first page of Google,” she said. “It’s still a gamble, and who knows if you’re gonna be the one that they request if they’re looking for the latest comfy slipper. It’s really hard to control that. But what you can control is how many creators you hire, how much you boost that.”

Then she gave the example that should stop every brand team mid-launch. A new Korean skincare brand arrives in Boots this month. The AI has nothing to say about it. “It’s just launched, and it’s only in Boots or whatever. Claude won’t know about that yet.” So she wants a beauty buff to tell her.

AI has a blind spot, and it is called new

An AI assistant is a retrieval system before it is anything else. It reaches for what has already been published, discussed and corroborated by somebody other than you. That works well for “best running shoes for flat feet.” It is useless for the variant you are launching in October.

Nothing has been said about it yet, so nothing can be cited about it. Ask the assistant and it will recommend your competitor with total confidence, because your competitor has three years of reviews and you have a press release.

This is not a gap that better AEO closes. It is structural. Every launch starts at zero. Every new variant, every new market, every new SKU, every fashion drop. The brand is the only party who knows the product exists, and the brand is the one source these systems are trained to discount in favour of third-party corroboration.

Somebody has to go first. Somebody has to use the thing and say what happened. That is not a job AI can do, because AI is not in the room when the box is opened.

AI is not a person, and that is the second job

Even once the answer exists, people do not simply take it.

IAB’s research with Talk Shoppe, built on 450 screen-recorded AI shopping sessions, found that fewer than half of shoppers, 46%, fully trust the recommendations AI gives them. 89% double-check before they buy. Asked what would increase their trust, they named sourcing, where the information came from, at 88%, and verified customer reviews at 87%.

The most revealing number in that study is the journey itself. Before AI, the average shopper took 1.6 steps online before purchase. After an AI interaction, 3.8. AI did not shorten the path. It added verification stops.

So where do people go to verify? To someone like them. When IAB asked which voices AI should prioritise in its recommendations, everyday consumers sharing real experiences came first at 52%, ahead of professional reviewers at 43% and subject matter experts at 37%. Celebrities and public figures came last at 13%.

Becky put the same thing in plainer language. Strip the creator economy back to its core, she said, and it is “people that feel like me, maybe have a similar paycheck to me, have a similar life to me, saying if I’m gonna spend my money on something, it’s gonna be this. If you’ve got rosacea, I’ve got rosacea, it’s this.”

That is social proof, and it works precisely because a person is on the hook for it. An AI can summarise social proof. It cannot be social proof. It has no skin, no budget and no rosacea.

The machines are already reading creators, and size is not the filter

Here is the part most brands get wrong when they do start taking this seriously.

OtterlyAI analysed more than 100 million AI citations across six assistants. Popularity had essentially no relationship with being cited: the correlation between how often a video is cited and its view count was -0.03. Likes, -0.02. Channel subscribers, -0.03. More than 40% of cited videos had fewer than 1,000 views. 35% of cited channels had fewer than 10,000 subscribers.

Their conclusion: “The best answer wins, not the biggest channel.”

A creator with 900 views can be the source an AI quotes, if the video answers the question clearly and is structured so a machine can find the part that matters. That makes creator selection a findability question rather than a reach question, which is a very different brief and usually a cheaper one.

Both have roles. The brands ahead are not choosing.

Agencies have already started working this way. Trevant audits which creators and content types are driving citations in language models before it books anybody. Crispin prioritises creators already being cited for a client’s category. Zoom hired a creator-journalist specifically to build up the authority its AI answers were missing.

The sceptic’s objection is worth keeping close. James Chandler of IAB UK: “Showing up in an AI answer isn’t the same as causing a recommendation. Without that distinction, AI influence could quickly become another vanity metric.” He is right, which is why presence has to be measured alongside demand rather than instead of it.

And while everyone argues about it, only 16% of brands systematically track their AI visibility at all, according to the framework IAB published in August. The tools that do it disagree with each other. You cannot yet manage the output. You can absolutely manage the input.

Three things to do about it

Budget the creators before you budget the AI visibility. For a launch there is nothing to optimise until something exists. Content first, optimisation second, in that order.

Brief the question, not the product. Ask what your buyer types or says out loud before they buy, then have a credible person answer exactly that, on camera, in their own words, including the parts that are not perfect. The systems that cite creator content are selecting references, not recommendations.

Measure presence, not just reach. Views tell you who saw it. Presence tells you whether you come back as the answer when your buyer asks.

Creators and AI are not competing for the same job. AI decides which answer to give. Creators decide whether there is an answer worth giving, and whether anyone believes it.

If you want to know where you stand today, we built Answerz to measure exactly that: the share of your buyers’ questions where you come back as the answer, across AI, search, social, retail and video. The first check is free.

Sources: Becky Owen, CMO Billion Dollar Boy, on This is Contagious, 23 September 2026. IAB with Talk Shoppe, “When AI Guides the Shopping Journey,” October 2025, 450+ recorded AI shopping sessions and 600 US consumers. IAB Insights Engine / Attest, “The Creator Signal,” September 2026, 2,200 AI-assisted shoppers across five markets. OtterlyAI YouTube Citation Study 2026, 100m+ AI citations across six platforms (vendor research). IAB, “Measuring Visibility in the AI Era,” August 2026. Digiday, “Creators are moving into brands’ AI search playbooks,” August 2026.