Charlie says: “My additional insight for 2025 is: don't bury your head in the sand about AI search. AI search is on the rise, LLMs like ChatGPT are growing rapidly, and Google's released AI Mode.

The counter-argument that everyone is making here is that ChatGPT is getting around 5 billion visits a month, whereas Google eclipses that completely, with 105 billion visits a month. But then, Google released AI Mode in the USA this May (2025). It rolls it out in Labs in India just a couple of weeks after. If you are ignoring that and only optimizing for regular search, the future is going to be really bleak.

We only need to look back not that many years ago to what happened to Blockbuster, when the internet first rolled out in the late 90s/early noughties. If you're Gen Z and you're thinking, ‘Who's Blockbuster?’ Well, exactly. That's exactly the point. Blockbuster was a video rental service. They were overtaken online completely by streaming services like Netflix and subsequently went bankrupt after not really adapting to the advent of the internet.

AI search, now, is a change that is going to be as big as that was then, and ignoring it is going to lead to exactly the same outcomes.”

I was going to ask you what specifically is on the rise, because obviously you mentioned various engines there as well, and Google AI Mode. Google AI mode is not available in the UK now, as far as I'm aware.

“Not yet.”

Not yet?

“Not yet.”

When is going to be ‘yet’?

“It's a good question, and I wish I had an insider at Google to know the answer to that.

My prediction for it though is, if they've rolled it out in the US this May and they've rolled it out in Labs in India (which, of course, is a huge population), we can probably expect to see something happening by the end of the year.

That would be my prediction, but I don't have anything concrete to say that for sure.”

(NB: Since the time of recording, AI Mode has been introduced in the UK)

In relation to the States, then, what impact is it actively having on user behaviour at the moment?

“At the moment, we don't know, because Google hasn't separated out any specific data around AI Mode, and we think that's because they don't want people to know exactly what's happening.

What we probably can say is that AI overviews (which have overtaken featured snippets almost completely at the top of the search results. They're the AI-generated summaries that you get in the normal search) very much seem like a testing ground; an early version of what AI Mode, that they've just released in May, now actually is.

AI Mode feels like the next level – the more advanced AI overview. It seems very clear that Google's moving on a trajectory of trying to get searchers used to having AI in the search results and more comfortable and confident with AI giving them answers, because the initial reactions to something like AI giving us answers, that was quite a new thing a year ago.

A lot of people's reactions were, ‘Oh, I'm not sure that I like that. I don't know that I'm comfortable with AI giving me the answers.’ Google's trying to get us used to that kind of search behaviour, and for that to be our new normal, so that, actually, the whole of the search results could be AI.”

Is there a specific user type, vertical, industry sector, or consumer group that's likely to be more comfortable engaging with an AI overview type of result? Or are we seeing, across the board, anyone interacting with them?

“From the type of data we've seen, we know that Gen Z tend to be much happier and more comfortable with this kind of technology. I would say, what we don't really know is how many people even realise that they're getting an AI result, which sounds kind of mad to say.

For anyone who works in marketing, we're so super aware of what's AI and what’s not. But if I think about my mum, my cousin, my nephew, etc., they don't necessarily know what's AI or not AI. They're just used to searching on Google, so they're not necessarily even reading this as an AI overview, or that they’re looking in AI Mode right now.

So, the jury's out, I think. What Google definitely doesn't want, though, is people pushing back on AI, when that's the trajectory that they're trying to move towards.”

Are you seeing any move towards other search engines, like Perplexity, or are people by and large staying with Google at the moment?

“We're seeing a significant rise in searches happening in ChatGPT, Perplexity, and across LLMs, to be honest. We're also seeing it in other LLMs, particularly if people's workspaces use a specific LLM, then they will be using that one.

ChatGPT is the one that's grown the most rapidly – that we're seeing the most visits for and the most searches on. Google is still eclipsing that when we look at the total number of people making Google searches every month.

They're still the dominant player here, and their release of AI Mode (which is essentially an LLM but within the Google browser) feels like their step to try and get ahead of potentially being overtaken by ChatGPT and other LLMs. They've just integrated it into the search experience they already have because they don't want to lose those people who want to be searching in LLMs.

The other interesting thing about this is that we know that lots of people actually search in both. They'll do a ChatGPT search, potentially, and then they might move into Google and do a similar type of search.

So, there's a lot of simultaneous searching going on still, but we have also seen that some people have moved completely. They're just going straight into ChatGPT, for example, and not really using Google search anymore.”

Has this impacted the user journey as well? Are users taking more or fewer steps in order to actually reach their decision?

“Yeah, this has definitely impacted the user journey and, in some of my early experiments in AI Mode specifically, you can really see that Google's trying to compress the buying funnel. It's trying to answer the informational questions quickly in an AI-generated response, so that the user doesn't have to go to five, six, or seven different websites to read and gain that information. They're actually able to get that information quickly.

Then, what you tend to see in AI Mode is Google very quickly pushing you towards whatever the answer is that you need. For example, if you're looking up places, it will give you a Google Maps of the places that you're looking for. If you're actually looking to make a purchase for something, it starts recommending you products. It recommends products in a panel with buying links, straight in AI Mode.

I think the goal for Google is definitely to keep people in Google's ecosystem and then to move buyers towards a purchase as quickly as possible, and lessen the amount of research that they have to do in that journey.”

How different is it to optimize for AI search instead of traditional search?

“This is the golden question that gets asked all of the time. Part of the struggle, I think, is understanding the nuances between traditional SEO (Google's algorithm for the 10 blue links that we're used to) and how LLMs are actually processing information.

The huge, huge expanse of searches that ChatGPT or another LLM can run in the background to make just the answer for one search is so much more complex. When we're thinking about SEO and AI search, there are clear differences in how you optimize for them.

If you've not previously been doing SEO, and you don't have underlying things like great site health, good internal linking, good site speed – all of the things that make a user experience good, you're going to really struggle in AI search. All of those things are incredibly important.

But when it comes to AI search, I mentioned there are more nuances; it's not just, ‘Oh let's do SEO and hope for the best,’ because things like schema markup are so much more important for AI. Website structure is so much more important. Also, contextually relevant information on your site is so much more important. As is pruning your content and looking after the pages that you've got.

Perhaps the biggest difference is the PR side (what, in SEO, we traditionally think of as link building/trying to get links from other good authority websites). When it comes to AI search, they care about citations and mentions of your business, even if those are unlinked, in a much broader sense.

They care about the context that those links/mentions are in, how your business is spoken about, how it's positioned, third-party review sites, what people think, how you're trusted, and what football team your business might sponsor. All of those sorts of things come into that big, broader PR piece when it comes to AI search, which takes us way beyond just thinking about technical SEO.”

I'm wondering, is there an automated way, perhaps using AI, to ensure that you've got the right schema in place for AI?

“You could definitely use AI to help with your schema, but whether it will 100% give you the accurate answer, I'm not entirely sure.”

You also mentioned pruning content. What do you mean by that? How do you know what to get rid of and what to keep?

“Oh, this is a great question. I think you can do really well with your Google Analytics and Search Console data here, to decide which pages on your website actually get read – ideally, get conversions – and at least get some engagement as well, like clicks through to that content. If you have content that is really tangentially related, and almost not that relevant to what you actually do as a business/what your service actually is/ the products you actually sell, and it's not getting read, get rid of it.

If it's not having any impact, it's not ranking, people don't go to that page, and they don't convert, then it doesn't give useful context about what it is your business actually delivers. You've probably gone too broad – and this probably applies mostly to any website that has spent a lot of time content spinning (as we used to know it) or AI generating loads and loads of content across all sorts of topics.

That is generally not very useful, and you're much better off thinking about what your business does, what actual content you need to support that on the site, and focussing really hard on relevant content that is positioned towards the kind of customers that you actually want.

More and more with AI search, you can't be everything to everyone. You need to decide who your customers are and how your business is positioned around them.”

When we're getting rid of a piece of content, getting rid of a URL, do we just 404 it?

“Please, no. Please redirect it. Please redirect it to the homepage at least.”

Even if it doesn't have any obvious links to it, externally or internally?

“Yeah. Ideally, you don't want to have anything broken left on your website like that.

I would say, if there's no good place for it to be redirected to – no similar piece of content – the homepage will do the job. If you've actually published lots of similar pages or similar content pieces, you can redirect to the most relevant one.

For me, it just keeps the site nice and clean.”

You touched on brand as well, and a lot of users may experience a brand and get an answer without clicking through to a website from AI Search. So, how do we measure the success of AI Search?

“This is a great question, and we're seeing a lot more development recently in tools that can give us measures around AI search. AI Mode is going to be the next really tricky patch for those.

There are some brilliant tools out there at the moment. Two of my favourites are Semrush's AI Toolkit. This is a very nice, clean toolkit, which is particularly good for smaller/medium-sized businesses that need a one-shot of what's going on. That measures things like AI sentiment and how you're performing against your competitor's share of voice, and it automatically gives you AI-generated recommendations for updates to make to your site. That's just within Semrush's usual tool suite. It's an add-on. It's really amazing.

There is a much more advanced piece of tech called Profound, at TryProfound.com, and this is really good for enterprise-level businesses that need to take AI search to the next level already, because there are many businesses of this scale whose competitors have already started this, and are working hard and putting aggressive amounts of budget into AI search. This does a huge, huge comparison against as many competitors as you need.

You can put in your own search queries, but it also generates the kinds of queries that are coming up in AI searches (in ChatGPT searches, for example), compares them against your competitor, measures your brand sentiment, measures how you're actually referenced (the positives and the negatives), and gives you a huge suite of information. I couldn't even possibly explain all of the different tabs that Profound has, but I would say we're looking at a lot more now at those kinds of measurements.

For me, at the end of the day, what actually matters is the ROI for a business: the money that they’re getting from whoever is coming through to their website. We know already that AI Mode can’t yet be tracked. We hope that something’s going to change as that rolls out further and further.

With ChatGPT, Gemini, Perplexity, and other LLMs, you can set up your Google Analytics or, if you use a different type of reporting like Looker Studio, you can set that up just to see how much referral traffic is coming in specifically from those sources, which ones, and how many conversions (if you are a lead generation business and you are tracking e-commerce sales) have come from LLM traffic.

That's a great thing to do. It gives you a really concrete idea of which LLMs are being used to search topics relevant to your business. What's actually bringing traffic? What's actually bringing conversions for you?”

Charlie, what's the key takeaway from the tip you shared today?

“My key takeaway has to be to use the 80-20 rule here.

We can't abandon traditional Google search yet. 80% of your time still has to be spent thinking about that, because it's still the dominant mode of search. It's where most businesses' money is still coming from.

In 20% of your time, focus on AI search. We know it's expanding. We know it's growing, and you don't want to be the business that didn't spend time thinking about investing and doing well in AI search as this future unfolds.”

As the future unfolds, do you feel that the allocation of time and resources will have to flip the other way, with 80% to AI search?

“Oh, I think there will definitely be a shift.

Whether it gets to 80% in the short term, I think we'll see it ramping up slowly and slowly over the next couple of years.”

Slowly and slowly for AI is like...

“Not that slowly. Not as slowly as everyone is hoping, I think.”

Charlie Marchant is the CEO at Exposure Ninja, and you can find her over at ExposureNinja.com.