How AI Is Changing SEO: What I Have Seen Change in 10 Years

Every client conversation I have these days seems to come back to the same thing. They have seen ChatGPT, they have seen AI Overviews appear above the normal Google results, and they want to know what it means for their website. Some are worried SEO is finished.

Others have been told by another agency that nothing has really changed and it is business as usual. Neither of those answers is honest.

I have been working in SEO for 10 years, and I have watched search go through several genuine shifts, but nothing has moved as quickly as the changes over the last 18 months.

In this article I want to walk through what has actually changed in how search works, what I am doing differently with clients at Nexa Growth as a result, and what I think is overblown.

Quick Answer: AI is changing SEO in three main ways. Search engines now generate AI summaries directly on the results page, which is reducing click-through rates for many queries. Search has expanded beyond Google to include AI platforms like ChatGPT and Perplexity, which cite sources differently to traditional search. And AI tools have changed how SEO work itself gets done, with much of the manual research and reporting now automated. The fundamentals of SEO, meaning relevance, authority and genuinely useful content, still matter. What has changed is where and how that content gets found and used.

What Has Actually Changed in Search Results?

The most visible change is AI Overviews appearing at the top of Google’s results for a huge range of queries. When I search for something now, more often than not, there is a generated summary sitting above the traditional list of links, often with a handful of source citations underneath it.

This is not a small cosmetic change. Industry data shows Google impressions are up significantly year on year, while click-through rates are falling as users increasingly get their answers directly from AI summaries instead of clicking through to a website.

For my clients, this has meant a noticeable shift in how I report on performance. A page can be getting more impressions than ever in Search Console while seeing fewer clicks, and a client who only looks at the click numbers will assume something has gone wrong.

I have had to spend more time explaining this dynamic over the last year than almost anything else.

What this means practically is that being cited within an AI summary is now genuinely valuable, even if it does not always result in a click.

For some businesses, particularly those selling services where brand visibility matters, appearing in that summary box is itself a form of marketing, even without the traffic.

Has SEO Become More About Being Cited Than Being Clicked?

This is the question I get asked most by clients who have read something about “AI Overviews” or “Generative Engine Optimisation” and want to know if their entire strategy needs ripping up.

My honest answer is no, not entirely, but the priorities have shifted. Generative engine optimisation has emerged specifically because content needs to be structured in a way that allows AI answer engines to lift the right text for snippets and summaries.

In my experience, the content that gets pulled into AI summaries tends to be the same content that was already ranking well for clear, specific, well-structured answers to questions.

If your content was vague, padded out, or written purely to hit a word count, it was never going to perform well in traditional search either, and it certainly will not get picked up by an AI summary.

I had a client in the home services sector whose page on a specific repair process started appearing inside AI Overviews for several related queries. We had not done anything specifically for “GEO”.

What we had done was restructure the page with a clear, direct answer near the top, followed by detail. That same structure that helps a human skim the page also happens to be exactly what these AI systems look for when deciding what to lift.

So my view is that being citable and being clickable are not in conflict. The content that does one tends to do the other. What has changed is that there is now a third audience for your content beyond Google’s ranking algorithm and human readers, which is the AI systems summarising your page.

How Has AI Changed the Day to Day Work of an SEO?

This is the part of the conversation that does not get nearly enough attention, and it is the change I have felt most directly in how I run my own business.

Ten years ago, a huge proportion of my time went on manual tasks. Pulling keyword data, building content briefs from scratch, writing meta descriptions one by one, putting together monthly reports by hand. A lot of that has changed.

I now use AI tools throughout the research and reporting stages of my work. Tools like the Ahrefs AI Content Helper analyse what is already ranking for a topic and help build out a content brief in minutes rather than hours. Reporting platforms generate plain-language summaries of performance data automatically.

The honest truth is that the bottleneck used to be the volume of manual work an SEO could get through in a day. That bottleneck has largely moved.

The work that remains, and the work that actually matters, is the strategic thinking: deciding what to prioritise, understanding a client’s business and customers well enough to know what content will genuinely help them, and making the judgement calls that AI tools cannot make on their own.

I would say to anyone running an agency right now that if your team’s value is purely in the manual execution of tasks, that is the part most at risk. If your value is in strategy, judgement and client relationships, AI tools make you more productive rather than replacing you.

Is Traditional Keyword Research Still Relevant?

Yes, but the way I approach it has changed quite significantly.

For years, keyword research was largely about finding the exact phrase people typed into Google and making sure that phrase appeared on the page. That approach is becoming less useful for a simple reason: people are not always typing short keyword phrases anymore.

Search behaviour has shifted from inputting keyword queries to having conversations, with users now asking fuller questions and following up within the same session rather than typing isolated search terms.

What this means in practice is that I now spend more time thinking about the full range of questions someone might ask around a topic, rather than just the single highest-volume keyword.

If someone searches “best accountant for a small business” and then follows up with “how much should I expect to pay”, a page that only answers the first question is missing half the opportunity.

Tools like Ahrefs and SEMrush still give me the underlying data on search volume and competition, and that data is still useful.

What has changed is that I am building content around topics and the full set of questions a person has, rather than building separate pages for slight keyword variations, which was a tactic that was already losing effectiveness before AI search arrived.

Does AI Generated Content Actually Hurt SEO?

I get asked this constantly, usually by clients who have either been using ChatGPT to write their blog content themselves, or who have been told by a previous agency that AI content will get them penalised.

The honest answer is more nuanced than either extreme. Google’s own guidance does not explicitly penalise AI generated content, but the language they use focuses on manipulation and poor quality content rather than the method used to produce it.

What I have seen with my own eyes is that thin, generic AI generated content does not perform well, not because it was written by AI specifically, but because it tends to lack the things that make content genuinely useful: real examples, specific detail, a clear point of view, and information that could not have been written by anyone without direct experience of the topic.

I had a client who came to us after a previous agency had produced a large volume of AI generated blog posts for them. The content read fine on the surface, but every post sounded exactly like every other agency’s blog.

None of it ranked, none of it got cited anywhere, and worse, it was actively diluting the handful of genuinely good pages on the site by burying them in a sea of generic content.

My approach now is that AI tools are useful for structure, research and first drafts, but the final content needs a layer of genuine expertise and specific detail that only a person with real experience of the topic can add. That is not a moral stance about AI, it is simply what produces content that performs.

What Should an Agency Actually Be Doing Differently Right Now?

Given everything above, here is what has changed in how we operate at Nexa Growth, practically speaking.

We now check AI Overview visibility for important client queries as part of regular reporting, alongside traditional rank tracking. This is not yet a precise science, the results vary by region and by individual search session, but it gives clients a picture of where they stand beyond the traditional results page.

We structure content with clear, direct answers near the top of sections, particularly for question-based queries. This was good practice before AI search existed, but it matters more now.

We have shifted some of our content strategy conversations toward topics and question clusters rather than individual keywords, reflecting how people actually search using AI tools and voice assistants.

We use AI tools heavily for research, briefs and reporting, which has freed up time for the strategic work that clients are actually paying for: understanding their business, their customers, and what will genuinely move the needle.

And we are honest with clients about what is measurable right now and what is not. AI visibility tracking tools exist and are improving, but anyone claiming to have this fully figured out is getting ahead of where the technology actually is.

Will Search Engines Like Google Become Less Important?

This is the big question behind a lot of the “is SEO dead” headlines, and I think it deserves a straight answer.

AI chat platforms like ChatGPT remain reliant on external search engines for real-time information, accessing search results from providers like Bing in order to answer queries about current events or recent information.

This matters because it means that if your website is not visible in search engines, it is also less likely to be surfaced through AI platforms that rely on those same search results.

In practical terms, this means traditional SEO fundamentals, meaning a site that is crawlable, well structured, fast and authoritative, remain the foundation for visibility everywhere, not just in Google’s traditional results. AI search has not removed the need for that foundation. If anything, it has added a layer on top of it.

What I tell clients is that I would be cautious about anyone suggesting you can ignore traditional SEO and focus purely on “AI optimisation” as a separate discipline. The two are connected, and for the vast majority of businesses, getting the fundamentals right remains the highest-value activity.

Frequently Asked Questions

Conclusion

AI has genuinely changed search, both in how results are presented and in how SEO work gets done day to day. But the core principle has not changed: content that genuinely helps people, backed by real expertise and structured clearly, performs well whether it is being read by a person, ranked by an algorithm, or summarised by an AI. If you want help working out what this means specifically for your website, take a look at our SEO services page or get in touch for a chat.