Getting Found by Generative AI

Clicks are so last year

Did you know that, according to Lloyds Banking Group’s 2025 Consumer Digital Index, one in three UK adults uses AI every week to manage their money?

It’s happening now, goes the cliché. 2026 isn’t the year AI takes hold; it has already done that (goes another cliché).

‘One in three UK adults’ covers the repeat audience. The research goes further, reporting 56% of UK adults have used AI to manage money in some capacity. That’s 28 million people, or as I like to think of it, more people than I can comfortably imagine.

Half of users expect their use of AI tools to increase over the next year. The tools the respondents use are of interest as well. 32% use a bank’s own AI solution, 9% use dedicated finance apps, but interestingly/shockingly (depending on your perspective), 60% use ChatGPT.

That’s a third-party, non-specialist, prone to ‘hallucinations’ AI tool that real people, in the real world, are using to support their financial lives.

It’s here, then, that we should begin our journey into the nascent world of GEO (generative engine optimisation).

The Lloyds report tells us people are using AI, specifically in their financial lives. This means that, increasingly, if you want to be found, you need to be found via ChatGPT or one of its alternatives.

I also believe that ‘clicks’, as a performance benchmark, are dead.

What are you using AI for?

According to the Lloyds’ report, AI is most frequently used for the following finance tasks:

1. 53% for savings goal planning

2. 52% for budgeting advice and planning

3. 51% for general financial education

4. 45% for insurance comparison and advice

If there’s one word that I know makes financial services folk sweat, it’s ‘advice’. It’s always one of my favourite moments when a green executive mentions the word ‘advice’ in a boardroom; I await the shutdown with real anticipation (I can joke because I’ve worn that shame myself). Yet, look at this, people are accessing ‘budgeting advice’ and ‘insurance advice’ via AI.

Imagine the scenario. A person loads ChatGPT, sticks an insurance quote into the window and asks, ‘Does this cover my insurance needs?’

ChatGPT may well give a great answer, but it might not. The same can be said of an advisor, I suppose, but at least I can take legal action against an advisor if it goes wrong.

Then there’s the issue of market knowledge, which is of particular interest to me, as a marketer.

ChatGPT might have been trained really well on the major insurance providers of the world, but because the perfect specialist insurer doesn’t have the marketing resources of the mainstream players, ChatGPT doesn’t even consider them.

Our fictional user may have received some wonderful insurance advice, but it only references the top-five insurers by market share, not the niche specialists they need.

If you’re not ‘optimised’ for AI-driven search, you’re invisible. The same has been true of ‘traditional’ search forever, but the stakes this time are higher.

Why are the stakes in GEO higher than in SEO?

Because people don’t use search to find things anymore. They use it to get advice, answers, guidance, and decisions.

The top of the funnel used to happen on brand-owned websites. Now it happens in a chat interface.

In an effort to avoid jargon, what I mean when I say ‘the top of the funnel’ is the stage of a person’s decision journey where they carry out research.

It refers to the typical funnel-shaped journey we all go on.

Step one is I realise I have a problem that needs solving, so I carry out some research. Step two is where I assess the options I’ve found. Step three is where I make my decision.

In the (recent) past, this would have meant going to a search engine, searching for what you need, clicking on some results, carrying out your research and moving on from there.

A really easy, real-world example that demonstrates this is how we all get weather reports. In the past, it would require using Google to search for ‘weather in XYZ’, a click on the local weather reporting page, and there’s your answer. Then, Google introduced the weather widget on the search results page; now there’s no need to click to get your answer.

The same is effectively happening with more and more information now.

AI-driven tools visibly change the journey. The research happens within the chat interface. AI-driven search doesn’t just tell you which brands to review, it reviews them for you. It means there’s no need to click as often.

Why go to the brand’s website when AI has summarised it so well for me?

This is both a good thing and a bad thing for brands. The bad news is that top-of-funnel traffic drops (the data bears this out, more on that shortly), but the good news is that traffic that does eventually visit you is better informed and therefore, more likely to be further down the funnel and ready to dance.

The click is dead, long live the click

I took part in a webinar about GEO, AI and the future of the communications industry, and I said that tracking campaign success is tricky and only getting trickier, and that I think that’s an exciting thing.

Clicks have, in my experience, always been a proxy for success in modern marketing. One of the big early appeals of ‘digital’ in marketing was that it could be tracked. Many people still pretend it can be tracked perfectly, but that hasn’t been true for a very long time.

Don’t get me wrong, tracking continues to play an important role in marketing, but the days of being able to track absolute numbers for a campaign are gone.

There are several reasons: data protection regulations make standard tracking protocols opt-in (they used to be opt-out), technology providers are increasingly blocking tracking technology (except for those whose business model relies on advertising), and users are opting for solutions that protect their privacy (VPNs, blockers, privacy-first web browsers).

Thanks to the rise in AI usage, there’s now another good reason tracking is folly; users don’t click.

One of my closest friends is a genius. One of my favourite things to do when I was growing up was to sit in the pub with him and ask him about things I was curious about.

He’d almost always have a good answer, backed with solid references. I learned about new books, television shows, and resources to check out. I almost certainly spent money off the back of these discoveries, but at no point did the beneficiary track my ‘click’ back to my genius friend.

I believe we’re at the genius friend stage of generative AI tools. The click is no longer a button press; it’s a ‘click’ in a person’s mind. It’s an ‘ah-ha’ moment, it’s learning about something and retaining that knowledge.

Good luck with tracking that.

If a tree falls in a forest and no one is around to hear it, does it make a sound?

I don’t know, we’re not talking about trees.

But if a user finds out something via AI, but doesn’t click, does it have any value? Of course it does.

I believe we’re returning to the olden days of marketing at great speed. We’re getting back to a time when gaining ‘share of mind’ was the goal, and clicks weren’t a part of the discussion.

Returning to my point earlier in this article, I think that’s exciting. It reminds us that brave, memorable marketing gets rewarded. A campaign that catches your attention, even for a moment, is hard to track, but no less important.

So here’s my first conclusion about the new world of GEO; clicks will decline in importance.

So, how do we track GEO performance?

This really is the question right now. I think the answer starts with reminding ourselves of a long-standing fact; tracking search rankings (i.e., SEO) isn’t an exact science.

SEO, for many, is the game of chasing ranking position one. The trouble has for a long time been, though, that there’s really no such thing. There’s ‘position one’ for a single person at a single moment in time, but the next time a different person searches for the same thing, position one may (and often is) different.

The drive for personalisation is the culprit here. Personalised results get better engagement, which in turn drives better ad revenue.

So, like many companies, we track rankings, but we do it at scale. We do our best to use an ‘all else being equal model’, a large sample set (in time and scope), and we report on ‘share of search’ over absolute rankings.

It’s not perfect, but it gives us a good enough idea.

The same needs to be true of share of visibility in AI, but there are some new challenges:

1. AI is prompt-based, not keyword-based

Keyword research is easy. There are lots of tools that capture the data. Keywords are tidy little packages of information. Prompts are less simple. They tend to be longer than a simple keyword and by definition, more likely to be unique each time.

2. ChatGPT is the leader, but not the only show in town

At the time of writing (Feb ‘26), ChatGPT has 80% market share of the AI chatbot market (Statcounter). That’s a dominant position, but it’s not cemented into top spot by any means.

Between December 2025 and January 2026, Google’s Gemini rose from 4.7% to 7.2% share. Gemini, it is rumoured, is soon to become the brains behind Apple’s next version of its personal assistant, Siri. Given that it also powers Android’s equivalent, Gemini will effectively wrap up the mobile market all for itself.

The fact is that, as these tools evolve, user preferences is will shift quickly and it’s very possible that it will take some time before we have the ‘Google’ or AI (or maybe Google will be the Google of AI).

What I’m trying to say is, you need to track a good few tools.

3. Tracking mentions, citations, and clicks

Returning to the ‘classic’ search, a search result is essentially a ‘blue link’ that occupies one of a limited number of positions on a results page.

This means you can reasonably track:

• Ranking position

• Impressions

• Clicks/traffic

Generative AI presents information in a conversational way. The result could be a list of brands/resources, but it could equally as easily be a short essay, a diagram, a table, even a personalised podcast-style show just for you.

The challenge for marketers here is, what do you track?

Our first go at an answer

We get paid to answer this type of question, so we’ve put a lot of thought into it, and here is what we’ve settled on so far.

Prompt planning

There is (to my knowledge), no reliable source of ‘prompt research’ in existence right now. No generative AI platforms share prompt data, so it’s not possible to know for sure what people have to say via a prompt.

What we do have is:

1. Keyword research

2. User research

Keyword research still plays a part; it remains the best tool we have for ‘getting in the ballpark’ of what a set of people are interested in.

User research is more of a UX discipline, and I think SEO pros need to start getting more familiar with their colleagues in the UX team.

A good bit of persona research in combination with keyword research gives you the tools you need to build an organised set of ‘representative prompts’ for your audience.

Track those!

Scale

It’s not reasonable to manually track prompts. You need a system that can run the same prompts consistently, in the same way, at regular intervals across multiple tools.

That’s the closest you can get to the golden ‘all else being equal’ standard for reporting.

Introducing AI Share

It's the manual element we most wanted to overcome, so we built a new tool called AI Share.

The concept is simple; it's a platform that works in much the same way as a 'classic' keyword tracking tool. You load in prompts you would like to track, organise them by theme, and select the AI models you'd like to test.

The tool will then head off at regular intervals; daily, every three days, weekly, monthly, whatever you like. It will run the same prompts using the same methods every time and tell you when and where you're mentioned.

Is GEO tracking perfect?

Hell no. This is trend spotting, not absolute performance tracking.

But it's better than finger in the air, and that's better than sitting in a boardroom with no answer.

If you'd like to get early access to AI Share, visit our product page or get in touch with us to learn more.