There's a question I keep seeing from marketers right now, it goes roughly like this: "We need to be visible in AI. How do we get our brand into ChatGPT's answers?"

Fair question. Wrong framing.

This week, Shopify launched storefronts inside ChatGPT and Copilot. Meta rolled out a conversational shopping assistant across Facebook and Instagram. Over half of marketers now have dedicated budgets for generative engine optimisation.

The surface-level read is straightforward. New channel. Optimise for it. Move fast.

That read isn't wrong. But it misses the bit that actually matters.

The scramble to "get visible in AI" is repeating the exact mistake most brands made with SEO fifteen years ago. They optimised for the system instead of earning the authority the system was designed to surface.

This time, the system is catching up before most teams have even started.

‍

The Volume Play Is Already Broken

Here's the stat I keep coming back to.

A University of Florida study, published in the Journal of Marketing Research, found that mid-quality AI-generated content, the kind most teams are producing right now, actively congests recommendation algorithms. It doesn't just fail to help. It makes it statistically harder for your best work to surface.

Producing average content at volume is degrading the environment your good content competes in.

This isn't theoretical. The researchers modelled outcomes across a quality spectrum. The middle tier, which is exactly where most brands sit with their current AI content workflows, produces the worst outcomes for everyone. Consumers. Creators. The platforms themselves.

So the first instinct most teams have when they hear "you need to be visible in AI", which is to produce more, faster, across more surfaces, is precisely the strategy that makes it worse.

For them. And for everyone else.
‍

What AI Discovery Systems Actually Select For

This is the bit most teams haven't internalised yet.

AI discovery isn't search. It works differently at a structural level.

In traditional search, you optimise a page. You target a keyword. You build links. The system ranks pages. The user clicks through and evaluates. Your brand gets a chance to make its case on its own turf.

In AI-mediated discovery, the system doesn't send the user anywhere. It reads everything, synthesises, and delivers an answer. Your brand either makes it into that answer or it doesn't.

According to AirOps, only 15% of pages ChatGPT retrieves actually end up in the final response.

That's a brutal filter. And it's not filtering for volume or keyword density. It's filtering for the kind of content an AI system treats as authoritative enough to cite.

Authority. Not visibility.

When Shopify builds checkout into ChatGPT and Meta routes product discovery through Llama, they're not creating new billboards. They're creating systems that need to trust your data, your content, your product information enough to recommend it in a conversation.

The quality of your catalogue metadata. The depth of your product descriptions. The consistency of your reviews.

These aren't admin tasks anymore. They're the raw inputs that determine whether the AI includes you or skips you entirely.

‍

The Trust Corridor Is Narrower Than You Think

Kantar's consumer data makes this uncomfortably concrete.

Fifteen percent of consumers say that if an AI doesn't suggest a brand, they assume it isn't right for them. AI omission isn't neutral. It's a negative signal.

But the other side is just as sharp. 41% believe brands recommended by AI paid to be there.

Absence disqualifies you. Presence invites scepticism.

The only way through that corridor is genuine authority. The kind that makes a recommendation feel earned rather than purchased.

This is where the volume play falls apart completely. You can't flood your way into trust. If your content reads like it was generated to game the system, and the consumer already suspects recommendations are paid, you've confirmed their bias before they've even clicked.

The Honest Objection

I can hear the counter-argument, because I've made it myself.

"This sounds great in principle, Andy, but authority takes years to build. The window is now. Teams need to move fast, and moving fast means producing at volume."

There's truth in that. The 55% of marketers with dedicated GEO budgets aren't wrong to be investing. PMG recommending clients pilot at 1.5 to 2 times their existing search budget isn't reckless. The window for building structural advantage in AI citation is genuinely open, and it won't stay open forever.

But speed and volume are not the same thing.

You can move fast by auditing your product catalogue data this week. By rewriting your twenty most important product descriptions with the kind of specificity an AI system can actually use. By ensuring your Meta Shops feed is treated as a strategic asset rather than something an intern updates quarterly.

That's fast. And it builds something durable.

Spinning up fifty blog posts a month that say roughly the same thing in slightly different ways, hoping one gets cited? That approach worked passably for SEO in 2015. It's already failing in AI discovery in 2026.

The Real Advantage Is Structural, Not Tactical

I've watched this film before.

When I started in marketing, print was king and digital was the side project nobody took seriously. I watched that shift play out over a decade.

The teams that won weren't the ones who moved first. They were the ones who understood the structural difference between the old channel and the new one.

Digital wasn't just "print, but online." It rewarded different things. Measurability. Speed. Iteration. The marketers who treated it as a new medium with new rules built advantages that compounded for years. The ones who moved their print ads onto banner placements wondered why nothing worked.

AI discovery is the same kind of shift. It's not "SEO, but for chatbots." It rewards depth over breadth. Specificity over volume. Genuine expertise over keyword coverage.

The tactical response is to optimise for AI.

The structural response is to become the kind of brand that AI systems select for naturally, because your content is genuinely the most useful, specific, and authoritative answer to the question being asked.

That's a harder brief. It's also the one that compounds.

What Changes Tomorrow Morning
‍

Stop thinking about AI visibility as a distribution problem. Start thinking about it as an authority problem.

The question isn't "how do we get into more AI answers?" It's "when an AI evaluates our content against every competitor's, do we win on substance?"

Pick your ten most important product pages or service descriptions. Read them as if you were an AI trying to decide whether to recommend this brand. Are they specific enough? Do they contain the kind of detail that makes a recommendation feel confident? Or are they generic copy that could belong to any competitor in your category?

Then look at your content pipeline. If your current AI content strategy is primarily about volume, pause it. Redirect that resource toward making your best existing content sharper, more specific, and more genuinely useful. The research is clear. More average content makes things worse, not better.

And if you're on Shopify or Meta Shops, treat your catalogue data like a first-class strategic asset starting this week. Not next quarter. The AI systems reading that data are already making recommendations. The quality of what they find determines whether you make the cut.

The discovery layer has moved. The brands that thrive in it won't be the loudest.

They'll be the ones that gave the AI the best reason to trust them.