I'll be honest; this post is later than I meant it to be. BrightonSEO was back in April, and I've been chewing over my notes ever since. Part of that is simply that things have been busy. Mostly, though, it's because it took me a while to work out what I actually made of it.
Part of mulling it over is that there was still great debate between those attending about whether AI has actually moved the goalposts for SEO and marketing people, or whether it's just clarified where we all should have been shooting towards anyway - almost a topical World Cup tenuous link! I'm still not sure there's a single, tidy conclusion to draw. There were several points of view, and plenty of them didn't agree with each other. When talking to attendees and agency owners, I heard the full spectrum, from "everything has changed" to "nothing has really changed at all." It's the same tension I picked at in an earlier post; is SEO dead, or is it carrying on much as it always has?
So among the attendees, that argument is nowhere near settled. From the stage, it was clearer. The industry leaders speaking this year were fairly united on one point: AI has changed the old model. Or at the very least, it's changed the emphasis, where your effort now needs to go.
For me, that change in where the effort needs to go points to something I've thought for a long time. Ranking was never really the point. The job was always to win the outcome, and AI has just made that harder to ignore.
The click continues to disappear, and "winning" needs a new definition
A headline stat came from speaker Matt Beswick: the share of a Google results page taken by the top organic result has fallen from 15.8% in January 2024 to just 4.3% in January 2026. Around 60% of Google searches now end without a click at all. People get their answer on the results page or from an AI assistant, and move on. Whilst this isn't new news, it really drives home the seismic shift in the numbers. That change is thousands, millions, maybe billions of clicks that just vanished.
Ashley Liddell summed up the consequence in a line: "Ranking is no longer the problem. Recognition is the problem." When an AI answer synthesises a response instead of handing over ten blue links, the question you should be asking changes. It stops being "which of your pages ranks?" and becomes "is your brand the one that gets mentioned?". Again, this is not new news, and was very much the direction of travel highlighted in last Autumn's BrightonSEO conference, but it continues to press the point, brand, brand, brand....brand.
The real change is one of attitude: from "your website" to "your reputation, everywhere"
This was the thread running under everything else. According to Christian Desert, roughly two-thirds of the sources behind AI answers aren't on your own website. Callum Steenson made the same point from another angle: AI answer engines build a picture of your brand by pulling from lots of third-party sources at once, from Reddit and YouTube to review platforms like G2 and Capterra, respected media and Wikipedia.
Connie Ballantyne put the shift plainly: "AI is not a website, it's an entity." It wants to see your brand backed up across several credible, relevant sources before it will confidently recommend you.
"AI is not a website, it's an entity."
That's an uncomfortable idea if you've spent a decade optimising pages inside your own site, because it means you can no longer control your visibility purely from the inside. Digital PR, genuine community presence and earned coverage move from nice-to-have to the main event. The data at the conference largely agreed on this point; press distribution earned citations within days, Wikipedia and YouTube both punched well above their weight, and the fashionable hacks did nothing (more on those shortly).
Trust is the decider, and you can't fake it
A theme close to my own view: as AI floods the web with content that's competent but hollow, provable trust becomes the thing that sets brands apart. Annika Haataja cited research showing 62% of people name trust as a key factor in who they buy from, ahead of familiarity or price. So yes, AI wants trust signals, but humans want them too.
Sean Barber's talk, drawn from his work at Macmillan Cancer Support, was the most sobering of the two days. AI models, he reminded the room, "infer credibility from patterns, not truth." That's fine when you're choosing trainers and dangerous when you're giving health advice. His answer was radical transparency: name your authors and their credentials, say who reviewed the content, be honest about where and how AI was used, and treat accuracy as an organisation-wide responsibility rather than an SEO afterthought.
For anyone operating in a regulated or high-stakes field, that is both good ethics and how you earn the credibility signals that both people and machines now look for.
If attitudes change, so must the metrics
Here's where the "tools can't keep up" story actually belongs. Clinging to old metrics is a symptom of the old mindset, and several speakers were blunt about how far behind the industry has fallen.
It's an approach I've long taken with clients: judge a website by outcomes like enquiries, conversions and revenue, not by rankings or raw visit counts. It was good to hear the case made so plainly from the main stage.
Judith Lewis was the most direct: "Rankings, they're fun, but don't report on them. They're a waste of your time and money." Report instead on conversions, brand awareness, profitability, cost per acquisition, and a newer one: how accurately AI models actually describe your brand.
Aleyda Solis offered the most complete framework of the conference, built around three questions. Presence: Are you showing up and being recommended? Readiness: Is your content built to be found and trusted? Business impact: What is it actually worth? Her guiding principle is one I'd happily put on the wall: "The goal is not perfect attribution, it's an honest reporting model without overclaiming."
It wasn't all organic search at BrightonSEO. On the PPC side, John Readman made the commercial case for end-to-end understanding of PPC and campaign spending and how that's even more important with AI in the mix. This was more of a cautionary tale where he gave an example of two campaigns that posted identical 4.0x return on ad spend (ROAS), however one actually lost money, and the other turned a healthy profit, because ROAS typically ignores margin. His recommendation: measure profit on ad spend, and treat AI as the real, converting channel it now is. In his portfolio data, AI referrals converted at 4.8%, roughly twice the rate of paid search. As he put it, "nobody has a budget line for it.".....yet.
Beware the hype, and the people selling it
For all the genuine change, there's a lot of noise. Dan Liddle's warning stuck with me: around $1.5 billion has recently gone into AI-visibility tools, and the vendors selling them have every reason to make the problem sound as frightening as possible. He called the result "operational theatre," activity that looks like progress until someone asks what it actually delivered.
The experiments he showed backed his scepticism.
- The much-hyped llms.txt file? Reality... involved in roughly 0.1% of AI bot traffic.
- Duplicating your pages as markdown? Reality....Zero crawler visits.
- Schema markup? Brilliant as it is for Google's AI features, this made no measurable difference to citations in ChatGPT, Gemini or Copilot.
That last one, Schema, had some conflicting views across other talks, but he had the data, and I'm a data person! Don't ditch Schema just yet, though - he also said it's brilliant, and it wouldn't surprise me if we hear that Schema is equally brilliant for AI citations at the next BrightonSEO.
Final piece of advice on hype came from Thomas Peham. Test and measure rather than blanket-adopt whatever's trending on LinkedIn this week. Every man/woman and his/her dog is touting the silver bullet to SEO and AEO on LinkedIn, Instagram and TikTok. Some are great, most are entirely unproven. Stick to our marketing principles and test, as these tactics behave differently on every engine, in different sectors, for different businesses etc, etc.
Probably most importantly, it's a team sport now
The final thread tied the rest together. If AI builds its view of you from your website, your PR, your social presence and your reviews all at once, then running those functions as separate silos guarantees a muddled, contradictory brand story.
Callum Steenson described most marketing teams as "out of shape and out of position," with SEO chasing keywords, PR chasing headlines and social chasing engagement while nobody reinforces the same message. Everest Nwagwu made a similar case for paid and organic, pointing out that plenty of brands spend a good slice of their PPC budget bidding on terms they already rank for organically, paying twice for the same visibility.
The brands that come out ahead over the next few years won't be the ones with the cleverest single tactic. They'll be the ones whose teams and whose message pull in the same direction.
My conclusion
What struck me most wasn't any single stat. It was how much the winning approach now sounds like the advice good marketers have always given, only with higher stakes. Be genuinely useful. Be genuinely trusted. Show up where your customers already are. And that's the reassuring part: AI hasn't overturned the fundamentals of good marketing, it's just made them harder to ignore.
I came away more convinced than ever that this is a joined-up problem. The brands that thrive will be the ones that connect SEO, digital PR, content and social so their story holds together wherever an AI or a person runs into it. It's probably a good time to ask yourself... where do you actually show up, how would you even know, and what should you do about it?