Everyone has an opinion about AI search – or ‘discovery’ as it is more accurately referred to.
Everyone is an expert in their own minds. But what is the situation in reality? And what does it mean for brand owners, marketers and their agencies?
What are we seeing in the AI market?
User statistics: the numbers are booming! Well up on last year, the proportions of the main players have not changed significantly:
- ChatGPT leads the market with 810m users per week (they don’t publish monthly figures). This share equates to 68% of the market and is now dropping, down from 87% in 2024.
- Google Gemini is now close behind them with 750m users per month (note: not per week, like ChatGPT). Now up to 18% from 5% last year. They really missed a trick by letting ChatGPT launch before they put their own version on the market! Though I should point out that Google Search has nearly 2bn regular users, so they are still a long way ahead of ChatGPT in the market as a whole.
- MS Copilot trails behind at 275m users per month.
- Perplexity and other AI-only platforms are much further behind – though growing fast – with 30m users per month. That’s nearly double last year’s figure! They tend to offer more niche benefits, so time will tell if they get to challenge the big three above.
User behaviour is also changing.
60% of queries now end without a click. Although it’s hard to extract the figure for B2B, it may well be the same. Essentially, it is saying that a majority of internet users are just looking for information. They’re not looking for suppliers or companies to engage with – but rather answers to questions.
This may have always been the case, that searchers are just looking for information to back up their research. In which case a click-through before AI may have had less value to the website owner than one now. Since it looks like a significant portion of users are just looking for the right information.
However, there is also no way of knowing how many of these ‘time wasters’ might have engaged with the company and eventually become customers. Not to mention, in the world of B2B where the website owner is selling to a DMU (Decision Making Unit) someone just initially searching for answers may still turn out to be a valuable influencer within the DMU.
All of which goes to say that – this group should be treated with as much importance as click-throughs from more traditional, organic searchers. This is reinforced by the fact that conversions from AI-referred visitors are 23 times higher than from conventional organic search visitors. Now that really is impressive and cannot be ignored!
How do we define this in words we can understand?
Well, one way is to say that the old ‘search’ economy has morphed into the new ‘discovery’ economy. AI-referred visitors are already warm: they have a good idea of what they are looking for, and the AI’s answer reinforces client confidence. The link gives them the source of the answer, increasing the likelihood for conversion!
There was a time when there was a close relationship between the number of impressions and the number of effective clicks into the website – AI overviews changes this. Many users may now be faced with your information before entering the website, and often that information proves enough to the user. It becomes even more important to have your website’s information structured for machine reading – the dominance of UX has been replaced by the new requirements of AI.
AI are ‘probability’ engines, not ‘search’ engines. The richer your website content is, and the better organised, the higher likelihood of appearing on AI overviews. Having highly informative/didactic sections of your website is more important than ever.
What does this mean for brand owners and/or agents?
There are three things you can and should do immediately:
- The website technology must be optimised for AI ingestion, not just clicks, by formatting the content into clear information clusters so AI models can easily parse and cite data.
- The schema structure must be clear.
- Not all bots are the same! AI overview crawlers should not be blocked.
This therefore leads to an ‘answer-first’ architecture – with this modular cluster formatting and the maximisation of information density, avoiding ‘fluff’ – a shift from where SEO managers have been taking their websites traditionally.
Another big change is the expected update frequency. Where a 6-month update window once proved enough, now update frequency is recommended at 3 months. Having AI tools in their stack means many competitors will update regularly – and it is essential to not be left behind in the race.
What does it mean for website owners?
Essentially, it’s about moving from traditional transaction advertising to a decision-support model, providing honest, informative content instead of aggressive sales pitches. So, phrases like ‘buy now’ are out and ‘see if this is right for you’ is in. In old fashioned terms: brand before response.
It has implications for your strategic placement as well, as you’re looking for impressions rather than conversions – assuming you have landing pages set up to convert the visitors that these advertisements attract!
The shift of search habits from browsing to problem solving must be understood for your advertising to be effective. As searcher intent has changed, so must your advertisements and media buying habits.
The fact is that AI is an ‘influencing’ machine, not a ‘buying’ machine. At a superficial level the same rules apply as for SEO, only more so. However, at a deeper level there is much that needs to be done to get to the hallowed position of an AI influencer.
AI creates new demands on your website, so your search strategy must change to reflect it. If not, you will lose out to those that do – that 23X conversion rate is coming from somewhere; it’s being stolen from you!
