5 things every brand needs to know about AI search

We recently said that writing about a technological revolution whilst you’re in the midst of it can feel a bit futile; by the time you’ve hit ‘Publish’, the insatiable creature which the myths call ‘AI search’ has shapeshifted beyond recognition for the fifth time that week.

But instant obsoletion of media is just par for the course nowadays, so we’re taking a crack at tossing our tuppence into the AIO/AEO/GEO/SEO discourse anyway.

To help make our advice a little less ships-passing-in-the-night, we hosted an event in June 2026 featuring a panel of experts on this very topic, including:

• Laura Iancu, an SEO consultant with a developer background who spent six months running live tests across three continents on Octopus Energy’s web presence

• Luci Cunningham from 125 Data & Insight, whose State of the Nation report collected nearly 30,000 consumer responses across 300,000 questions

“So, where do I start?” was one of the big questions of the evening. Laura reassured us all, saying that ‘traditional’ SEO is still relevant, and if you’re already doing SEO right, you’re already way ahead of the starting block.

The practices that have always made a brand credible online (consistent information across platforms, trustworthy expertise, third-party mentions and citations) are mostly the same things AI search is built to surface.

What has changed is where AI search draws its answers from, and how far beyond your own website it looks. When ChatGPT, Claude, or Gemini recommends your business, it’s looking at review platforms, community forums, independent publishers, industry directories: any credible third-party source that mentions the brand adds to the picture AI is constructing.

The practices Laura talked about all come back to the same thing: earning your place in trusted spaces that aren’t your own.

It’s what we do a lot of at Trapeze Media: helping businesses take a degree of ownership over third-party media, signposting prospective customers towards a sense of brand loyalty, flipping user-generated content into digital land-grab, and so on. Nail down those factors, and you’re on solid ground.

Laura drew a distinction between ‘citations’ and ‘mentions’ that’s going to become more discussed as AI search matures and the question of “what actually feeds these models?” gets answered more concretely.

Basically, a citation is when your brand appears in an AI platform with a link. A mention is when it appears without one: in a Reddit thread, a community forum, a Quora answer, anywhere that real people gather on the internet to chin-wag about things.

The mention still counts, she argued: it still builds authority, it’s still a signal to AI models that your brand exists in a real context being discussed by real people.

The practical implication is that link-building isn’t the only measure of third-party credibility worth caring about, and someone recommending you on a subreddit without tagging your URL is still doing something useful.

Harder to track, maybe. But without getting too in-the-weeds, there are certainly tools to track those sorts of ‘messier’, off-the-beaten-path metrics. If you want to talk about how, get in touch.

Y’know how adding ‘Reddit’ to the end of any ‘traditional’ search is often the only surefire way to get something decent from a Google these days? Not a shocker, then, that AI search engines basically do the same thing.

Our favourite story from Laura during the evening involved greenwashing claims being made against Octopus Energy on Reddit. Rather than seeking a takedown or issuing a corporate rebuttal, they worked on finding the actual scientists who’d conducted the studies being referenced, brought them into the relevant subreddits, and had them engage. No sales tactics needed. Trust and authority did all the heavy lifting.

It worked, but that doesn’t mean you should just go rushing to Reddit willy-nilly. The moderation is aggressive, and communities can smell marketing from miles away… well, mostly. There’s certainly an art to Trojan-horsing promos onto the platform.

If you want Reddit to function as an authority signal (and it’s a significant one, since most AI crawlers draw heavily from forum-based, first-person content), you have to be legit useful in those spaces, which usually means having someone who actually knows what they’re talking about do the talking. A top-level example is AMAs (Ask Me Anything sessions), which is literally just celebs dropping by to answer questions. It’s all marketing, but it’s done in an interesting, sometimes helpful way.

In simple terms: don’t be boring.

Laura stirred the pot with this one! But she’s walked the walk with these tools and earned the right to go against the grain.

The pitch for premium AI visibility reporting is tempting: know exactly how often your brand appears in AI-generated answers, track your share of voice against competitors, get to the nitty-gritty of the numbers.

The problem is that these platforms literally can’t give you real data. GDPR prevents ChatGPT from handing over keyword-level search information, so what you’re actually paying for is a mathematical approximation of what real queries might look like, modelled with synthetic results.

Laura’s advice: go to the AI platforms directly instead. Ask them about your brand, ask them who they’d recommend in your sector, and see what comes back: which sources they cite, how they describe you, who they put you alongside. You’ll learn more from ten minutes of that than from a month of synthetic dashboards.

She also flagged that France briefly banned AI Overviews in Google’s search results after they launched. The regulatory picture could shift quickly, and expensive subscriptions to tools that track a feature which might not exist in your market in six months probably ain’t worth the risk.

If you absolutely have to hop on one of these tools, keep the subscription monthly (no annual commitments). If we’re fretting about our blog being out of date by the time we publish it, it’s probably a good signifier that you don’t want to marry your eggs off to a single basket.

Luci’s State of the Nation data highlighted that 75% of consumers said their decisions about where to eat, drink, or go out came down to either previous experience or peer recommendation.

This ain’t a eulogy for digital marketing, of course. It’s more a case for why trust is the thing worth obsessing over within your organic and paid ad campaigns. Discounts have slipped as a decision driver. Social media’s influence on final decisions (AKA, conversions), as opposed to awareness, has softened too. What’s filled the gap is credibility: the sense that this place or product is worth your time, confirmed by someone or something you already trust.

AI search fits perfectly here. Luci mentioned she uses AI for trip planning, with a hefty asterisk: she uses it to confirm a place is decent, once it’s already been recommended to her. A friend suggests a restaurant, she Googles it, maybe pins it on Google Maps, looks at the Instagram, and if it also appears in an AI search with good reviews, that’s the cherry on top which seals the deal.

The flipside to this concerns the other way AI is being used en-masse right now: generative images. The more AI-generated content that floods every channel, the more customers retreat to things that can’t be overtly manufactured. We’ve seen it in the campaigns we make for clients: creatives embracing rougher-around-the-edges aesthetics, scanned-scrapbook mixed-media imagery, anything that reads as visibly human amidst the deluge of glossy-to-a-fault visuals. The desire for something ‘real’ (a subjective concept in and of itself) is increasingly prevalent.

Again: don’t be boring.

The thread connecting all of this: trust. Laura’s focus on third-party citations, verified expertise, and Reddit’s community credibility. Luci’s data on peer recommendation and previous experience as the top two decision drivers. AI search as one of the checkpoints where trust either gets confirmed or falls apart.

The mechanics behind AI search aren’t as different from what we’ve always known as the acronyms suggest. The brands that show up well in ChatGPT and Gemini are, broadly, the brands people were already talking about. Which makes the brief pretty simple, even if the execution is anything but.

Did we mention ‘don’t be boring’ yet?