The audience you didn’t know you had.

Did you know there are two versions of your business’ brand out in the world right now?

One of them you deliberately created. It’s on your website, it’s in your sales decks and your brand guidelines. It’s your messaging that you agreed on potentially years ago, and somebody spent months making it sound like you. 

The other one is an AI summary that nobody wrote; an amalgamation made by AI tools that piece together brands from websites, social channels, company profiles and whatever else it can find online.

It’s a second (AI-generated) brand that you never signed off, and it may be the version a potential customer encounters before they ever visit your website.

Two audiences that want different things

Humans and AI tools don’t necessarily respond to the same things. 

When we use words like “friendliness” or “trust”, we interpret them through a lifetime of experience and emotion. AI doesn’t. It has no lived frame of reference, no nuance built in from context. 

A person can be persuaded by a story, a point of view or an idea that makes them stop and think. A machine feels nothing and only wants details: who you are, who you’re for and what you’re not. 

Until recently, this was a theory with little evidence behind it…

…Then the 73rd Cannes Lions International Festival of Creativity took place from the 22nd June – 26th June 2026. 

At the festival, INSEAD, Jellyfish and WARC looked at 480 campaigns entered into the Creative Effectiveness Lions, a category that rewards creative work based on demonstrated results. They ran the campaigns through AI and compared its assessment with the scores given by the original human judges. 

The AI assessments and human judges’ scores differed substantially. The comparison highlighted a distinction between what people respond to and what systems can interpret.

So, in short, your brand needs to work for two very different audiences. All the classic brand best practices still matter, but there’s now the extra challenge of providing enough clear, specific information for AI systems to interpret what your business does and why it may be relevant.

“Let’s make everything machine-readable”

Great in principle – restructure all your content so AI models can parse it – better structure, clearer answers, and all the most up-to-date info. Job done. 

All important changes to make, but it doesn’t answer the more fundamental question: what is the machine going to understand about you? 

If your website says you’re a leading provider of innovative solutions for businesses in your sector, making that statement easier for AI to parse doesn’t suddenly make it more useful. It’s still hollow; still says nothing. 

AI might get to you quicker if you make things machine-readable, but it still has no reason to name your brand to its users.

It’s the same reason GEO alone won’t rescue you. Good optimisation surfaces whatever’s already there. If you’ve not got an established brand; if nobody has taken the time to decide what your business is and who it’s for, there’s nothing to differentiate you from everyone else. 

If four companies describe themselves in almost exactly the same language, there’s very little for an AI model to distinguish between them. They become part of the same blurry whole rather than an individual answer within it.

Let’s get to the bottom of this: run this prompt on your own business

There’s a simple test you can run on your own business.

Step 1

Sign out of your account and start a brand-new chat for each step to reduce the chance that previous conversations or personalisation influence the answer. This won’t make every result identical, but it gives you a cleaner starting point.

Ask ChatGPT or Claude about your company by name. You’ll probably get a reasonably accurate answer. You might even get one that’s a little more flattering than you’d expect.

It’s tempting to presume this means you’re all good, but the cracks start to show when you try…

Step 2

Now name your industry and the services you provide. See if you’re one of the businesses mentioned. 

Finally:

Step 3

The real challenge. Describe the biggest problems your target audience want to solve.

If you come up here, that’s a good sign. Chances are you might not, though.

Chances are you don’t, though. 

We conducted this test with a new client, before we got started reworking their brand. 

This client is an industrial business in the UK (we’ve kept things vague here for obvious reasons). 

Step 1

No problems – when we used named brand terms, they showed up in all of the runs we conducted.

Step 2

When we use industry/service terms, we see a bit of a drop-off. 

These prompts were along the lines of “I need an [x product], suggest 10 local suppliers who meet my needs.”

The client appeared in 52.5% of the runs, a significant drop from searches that named the business directly.

Step 3

This is the problem, because this is how buyers are searching in AI tools:

“I’m struggling with [x problem]. My business is [describe business], and I need help finding a supplier who can help me.”

The client did not appear in any of the pain-point-driven runs we conducted.

These figures reflect one client, the prompts we used and the runs we conducted. Results can vary by tool, wording and timing, so this isn’t a universal measure of AI visibility. It did, however, show a clear gap between how the business described itself and the situations in which buyers might need it.

The lesson we need to take from this:

If you’re not talking in the right language or answering the right questions, AI systems have less to work with when deciding whether your business is relevant.

We call this “Artificial Irrelevance”. You aren’t being misdescribed; you’re missing from a conversation you didn’t even know was happening.

Nobody searches for a name they’ve never heard

If you already know the name of a company, you can Google it.

If you don’t know who could solve your problem, you’re more likely to describe the problem to ChatGPT.

A buyer might explain what’s going wrong, what they’ve tried already or what they’re struggling to achieve. The AI then gives them an answer, sometimes including a handful of businesses that appear relevant.

And nobody buys off one answer. They ask follow-up questions, they read, they go away and think about it, they mention it to a colleague. Weeks later, with a budget and a deadline, they open a browser and type a name.

Which name they search for may depend on what AI recommended weeks before.

G2 surveyed 1,076 B2B decision makers in March 2026. Sixty-nine per cent chose a different vendor than they’d planned to, based on what an AI chatbot told them. A third bought from a company they’d never heard of before that conversation.

Your branded search is the receipt. The sale may have started somewhere you weren’t watching. Being one of the answers to an original AI question can lead to a later branded search. If your business never enters that early consideration, you may never make the shortlist.

The final consideration: what is AI doing to your brand internally?

There’s a very good chance your brand was agreed on before ChatGPT peeked its head above the parapet. 

You’ve likely got a brand guidelines document sitting somewhere in your shared drive. When was the last time anyone paid them any attention? 

AI tools are useful, but when teams use them without clear guidance, the brand can slowly get eaten away.

After 18 months of your team improvising, how long until you come across something that reads like a different company?

That’s exactly why, for our clients, we build a SignalIndex™. 

It’s a deliberately dry piece of content – a dedicated AI governance document within the wider brand strategy.  

It includes:

It’s a very different piece of content to your customer-facing brand. That needs personality and flavour; it can tolerate ambiguity and ideas that make humans want to spend time with you. 

The SignalIndex™ has a different job – to be clear enough that there’s very little left for AI to interpret. 

Essentially, you end up with one brand that can operate in two modes.

The customer-facing brand can retain the personality, ambiguity and ideas that make humans want to spend time with you. The SignalIndex™ gives AI and your teams a clearer, more structured version of the same underlying truth.

Two audiences. One strategic foundation.

Go and run the second prompt

Sign out of your AI tool and describe the problem you solve the way a buyer would describe their problem. Don’t name your company or tagline and see who comes back. Run it a few times because the answer can vary.

Treat the results as a starting point, not a verdict. A single prompt cannot tell you how visible your brand is across AI tools, but repeated results can help reveal whether your business is being associated with the problems it is equipped to solve.

This is one of the many tests we conduct in our Brand Audit. We’ll look at what AI says when nobody names you specifically, what you say about yourself, and how you’re perceived by the market. 

It gives you a clearer view of what a potential customer might encounter, rather than relying on what you think you are saying.