Audio version
Don't have the time to read the post? You can listen to the full blog here
Having worked in marketing for the best part of 25 years, I can’t begin to count how many times I’ve had the conversation with business leaders around SEO.
Where are we ranking? Who’s above us? How do we get to first place on Google? (Usually for the most generic of search terms!).
Up until now, the website’s role in search was relatively clear: rank highly, get clicks, convert the visit.
As with a lot of things, AI has changed how marketers approach search performance. For a while now, some in marketing have been quick to declare SEO “dead”, but that’s not strictly true. Search is still key. It’s just no longer the full picture.
Above all, user behaviour has changed. Google says AI Overviews are leading people to ask longer, more complex questions, while also creating more opportunities for websites to surface through supporting links. At the same time, its guidance for site owners is crystal clear: the fundamentals of SEO still apply in AI search experiences. Pages still need to be indexed, technically eligible for Search, and built around helpful, people-first content.
That’s the shift businesses need to get their heads around. Your website is no longer only competing for clicks in a list of blue links - it’s also competing to be found, understood and trusted inside AI-generated answers.
So yes, SEO still matters. Your website just has more to prove.
More AI Content Is Not the Answer
If AI has changed search, the obvious temptation is to respond by producing more content, faster.
That’s where a lot of businesses are going wrong.
They see AI search, assume the answer is scale, and start packing their websites with more pages, blogs and FAQs - content that sounds polished enough on the surface, but says very little underneath.
Rather than supporting the shift, it usually just makes the website easier to ignore.
Google’s position on this is actually quite straightforward. It doesn’t reward content because it was written by a human, and it doesn’t penalise content simply because AI was involved. What it is looking for is content that’s helpful, reliable and created for people. It has also been clear that using automation, including AI, to churn out low-value content for the purpose of ranking higher is against its spam policies.
If your website is going to be surfaced, cited or used as a source, the content has to be worth selecting in the first place. It needs to offer clear answers and genuine usefulness, which generic content created at scale usually can’t match.
So no, the answer is not to publish ten AI-written pages where one well-written, genuinely useful page would do the job better.
The answer is to keep the website human-first, written for humans, and valuable enough to earn attention in the first place.
What Exactly Is LLM Optimisation?
“LLM optimisation” is quickly becoming one of those phrases that sounds impressive and slightly intimidating in equal measure, but it’s far less mysterious than it sounds.
Sometimes referred to as Generative Engine Optimisation (GEO) or Answer Engine Optimisation (AEO), Adobe defines it as how you make your brand and content visible, trustworthy and retrievable within AI-generated answers.
Put simply: if traditional SEO helped you win page-one rankings, LLM optimisation helps you win AI citations and visibility inside the likes of ChatGPT.
Still, it comes back to some fairly basic questions. Can AI systems access your website at all? Can they understand what the page is about? Is the content clear enough, useful enough and open enough to be crawled, indexed and cited as a source?
Essentially, the key is making your website easier to crawl, interpret and, critically, trust by both humans and AI platforms.
Google’s own guidance says there are no special requirements to appear in AI-powered searches. Pages still need to meet the usual technical requirements for standard search, which means they must be crawlable, indexable and built around helpful, people-first content. OpenAI is saying something similarly practical from a ChatGPT search perspective: publishers that allow OAI-SearchBot can appear in ChatGPT search answers, while publishers that block it won’t.
So when people talk about optimising for LLMs, the answer is often less confusing than they expect.
Can your site be found?
Can it be understood?
Can it be trusted?
If the answer is no, the problem is that your site still hasn’t done the basics well enough.
A sensible place to start is with the basics. Make sure the site can actually be crawled, that key pages are indexable, and that structured data is in place where it helps search engines interpret the content properly. Schema is not a magic fix, but it is one of the clearer ways to help search systems understand what a page is, who it is for and how it fits into the wider site. Google’s AI search guidance still points back to the same foundations, while Bing has introduced AI Performance reporting to show how content is surfacing across Copilot and other AI-powered experiences.
What Your Website Needs to Do Now
If websites are now being judged in more ways than rankings alone, you need to get much clearer on what the website actually offers users.
A lot of websites still make life harder than it needs to be. Some businesses love to shout about what makes them “unique” or “special”, but that’s much less useful than being clear on the value you offer and the problems you solve. Service pages become vague, key information gets buried, and the site ends up saying what the business wants to say rather than answering what the customer actually needs to know.
If that approach wasn’t strong enough for traditional SEO, it becomes even more of a problem in AI search. Google’s guidance still points back to the same core principle: helpful, reliable, people-first content.
So in practical terms, what does that mean?
- Clear service pages.
- Direct answers.
- Less jargon.
- Stronger internal structure.
- Content that says something useful quickly.
- And a website built around the questions your ICPs are actually asking, not only the things you want to say about yourself.
Your website needs to say something useful quickly, because users are increasingly seeing answers inside AI Overviews and other search experiences without ever clicking through in the old way. That means brands should not panic if clicks and GA4 patterns start changing. We’re living through an inflection point where old and new search behaviour are working side by side. The site still has to earn visibility, but it also has to add value immediately when it is surfaced.
If your website has strong foundations - original thinking, clearer positioning, and content that sounds like it came from a business that knows what it’s talking about - that is where you’ll give yourself a much better chance.
Structure starts pulling more weight here too. Strong internal linking, sensible page hierarchy, consistent metadata, schema where it adds clarity, and technically accessible content all make it easier for search systems to interpret the site properly. And this is becoming easier to monitor. Bing now offers AI Performance reporting, while GA4 remains a useful place to track how organic behaviour is changing. The point is not to obsess over one dashboard. It is to use those signals to build a smarter content and visibility strategy around your actual business goals.
Think about the customer journey properly. Who are your ICPs? What are they actually searching for? What do they need to know, and which audiences matter most to the business? A website rarely needs to speak to every audience equally. That comes down to strategic planning.
If you don’t already have an ICP audit, that’s a good place to start. It gives more structure to what the site is trying to do, who it needs to speak to, and what content will actually support the wider strategy.
Visibility Now Means More Than Rankings
For years, visibility was mostly discussed in one way: where do we rank, how much traffic are we getting, and what keywords are we showing up for?
Too often, that then turned into an obsession with what competitors were doing.
That is the wrong frame. The sharper question is what your ICP audiences are doing, what they are searching for, and whether your website is built to serve them properly.
AI-powered search is already operating at huge scale. Google’s AI Overviews had reached 2 billion monthly users by mid-2025, while ChatGPT had more than 700 million weekly active users in January 2026.
So it’s no longer only about whether someone clicks your result from a search page. It’s also about whether your site is being selected, surfaced or used as a source in the AI answer itself.
What Businesses Should Stop Doing
A lot of businesses are still responding to this shift by publishing more low-value pages, stuffing their sites with generic FAQs and churning out blogs that could have come from almost any company in the sector, whether they offer the same services as you or not.
Let’s be clear: that’s not adaptation. It’s dilution.
You need to be identifiable, clear about what you do, and clear about how you solve problems, especially in the service industry.
Google has been clear that content should be created for people first, and that automation used to produce low-value pages at scale for rankings is against its spam policies. So AI search is not a loophole for weaker content. If anything, it raises the bar on what your website needs to be.
The job now is straightforward: make the website easier to understand, more specific to the people you want to reach, and easier to trust.
If you need help making your website work harder in search and in AI-driven discovery, get in touch with Marmalade Marketing.

Jo Perrotta