There’s no doubt that the rise of AI has disrupted the SEO industry. But while some claim that SEO is finally dead¹ and have even invented new names (Hi, GEO 👋), others declare that nothing has changed and “GEO is just SEO”. And as is often the case with polemic positions, the truth probably lies somewhere in between (and might be more actionable than you think).

Let’s have a look at a selection of things that AI has actually changed in SEO, from the untroubled perspective of someone who has been working in the industry for roughly 15 years.

TL;DR

If you’re in a rush, here’s the short version:

• A lot of basic SEO work can now be automated: Elements like title tags or product descriptions can now be generated in bulk, achieving high quality and generating valuable organic search traffic.
• AI visibility monitoring tools – A new wealth of data at our fingertips: Tools that measure our brands’ visibility across AI platforms provide us with valuable data for our SEO efforts.
• Informational content no longer boosts vanity metrics: The traffic that many businesses have lost since the roll-out of AI overviews in Google’s search results is not coming back, but lots of businesses never needed it (the traffic, not the content) in the first place.
• LLMs value factual information over marketing fluff: If that press release boilerplate PDF from 2016 is the only document on your website that contains actual facts about your business, LLMs will find it and use it (and ignore your fancy corporate “About us” page).
• Bot management has become a new challenge: There are more bots than ever trying to crawl your website, and it’s a tough job to block the bad ones and let the good ones in.
• It has never been easier to get buy-in for SEO initiatives: People care more about “GEO” than they ever have about SEO, so it’s a great time to push those optimisation ideas that have been so hard to turn into realities for years.

End of TL;DR, but if you’re here for a longer ride, let’s dive right in:

A lot of basic SEO work can now be automated

The times of manually writing hundreds of title tags are probably over for most organisations. While I still recommend spending some time on optimising the most important pages one by one, high-quality title tags can be created in bulk for most pages with the right AI-supported processes.

And optimised title tags do still matter: A bit more than a year ago, we started a process with one of my clients where we created AI-generated title tags for all new publications on one of their content platforms. As quality and accuracy matter a lot to this business, all AI-generated title tags had to be approved by the numerous authors before they were implemented.

The friction of this additional human approval step resulted in only roughly a quarter of the AI-generated title tags actually going live, while the rest of the articles used the un-optimised article headlines as title tags.

This made for an interesting test² and the results were far better than we expected: Across ~650 analysed publications over a time frame of about 6 months, the articles that were published with AI-generated title tags achieved an average of 30% more organic search traffic than the ones that went live without optimised title tags.

While the data is clearly anecdotal and should therefore be taken with a grain of salt, this case is an example of AI being used to efficiently produce good results where resources were previously missing.

With another client of mine, this time an e-commerce business, we used AI to write tens of thousands of high-quality product descriptions based on product data that was already present on the product detail pages in the form of data sheets.

We created a detailed copy writer briefing and put a lot of effort into achieving the right tone, quality and messaging. For maximum transparency and compliance, we also made sure to place a clear disclaimer on each page that the content had been created with the help of AI.

The results were user-friendly and interesting product descriptions that added to the dry tables of data that were already present on the pages. This user-focussed addition to the product pages was rewarded by search engines with better rankings, more traffic, and more sales.

Again, anecdotal, but: We used AI to achieve what was previously impossible, due to resource and budget restrictions.

AI visibility monitoring tools – A new wealth of data at our fingertips

With the rise of AI, there has also come the advent of AI visibility monitoring tools³. Tools like LLM Pulse or Sistrix for Chatbots – still in beta at the time of writing (my two personal favourites out of the ones that I’ve tested) analyse different AI platforms’ replies to a set of prompts that you define. They then provide data about mentions for your brand and your competitors, sources used, and much more.

I will not dive too deep into the details of what can be done with this data, as this would fill an article of itself, but one aspect that stands out to me is citation data: AI replies contain information about the sources they use, and AI visibility monitoring tools surface them easily.

This is actionable data that businesses can use to improve their presence across different platforms, and it’s an interesting evolution from traditional link building or online PR, where it was often not clear nor in any way provable what particular links actually did for your rankings. With citation data from AI platforms, you do know which sources are referenced, and in many cases, you can directly influence whether your brand is mentioned there.

If you are not using an AI visibility monitoring tool yet, I strongly recommend you put it on your to do list.

Informational content no longer boosts vanity metrics

The rollout of AI overviews in Google’s search results caused significant traffic losses for businesses that relied heavily on informational traffic. In hindsight, at least for businesses who did not directly monetise the traffic, it turned out that traffic to informational content never was that critical in the first place.

This is not to say that the content itself had no value! Informational content is often a natural link magnet (I do not have to stress the importance of links for SEO here), and of course high-quality content can reach potential customers early on in their journeys and help with brand and trust building.

But a big part of the traffic that landed on informational content in the past was due to what I now refer to as “superficial research”. We used to search for a topic in our preferred search engine, open the first few links and skim through the content, only to close most of them immediately and spend a bit more time with maybe one or two pieces that seemed relevant to us.

AI features in search engines have taken over this “superficial research” and they now do the work for us. The result are fewer clicks to websites and therefore less traffic to informational content.

After the initial shock of seeing the vanity metric “traffic” nosedive, most businesses I deal with have managed to shift their attention to the kind of interactions that really matter: Users who visit our websites to become our customers, and there is so much SEO that can be done in this area!

LLMs value factual information over marketing fluff

One recurring theme that I’ve seen when analysing data from AI visibility monitoring tools is that LLMs, unlike Google, don’t care that much about the most user-friendly or popular resources. Instead, they go to where they find the information that matters.

On several occasions, I’ve seen AI replies cite press release boilerplates for factual information about companies, instead of taking it from the home pages or “About us” pages of the company websites. This is because corporate websites rarely contain any factual information at all, and too often resort to meaningless marketing fluff to describe their businesses.

LLMs seem far more interested in dates, numbers, and legal facts than in phrases like ‘leading’, ‘innovative’, or ‘customer-centric’.⁴

In my recent experience, it is astonishingly easy to convince people to add some purely factual bullet points to their marketing-oriented content by showing them this kind of eye-opening data.

Bot management has become a new challenge

In the good old days⁵, you were on the safe side if you made sure that Googlebot and Bingbot (and maybe a couple others) were able to access your website. Now, there are hundreds of important bots and unfortunately, lots of businesses block too many of them.

Bot traffic is expensive and by far outweighs human traffic on most websites. It can also be dangerous, depending on which intentions certain bots have. This is why IT and security teams like to block bots, either by setting up their own rules or by using out-of-the-box solutions.

As SEO practitioners, we now have to keep a careful eye on which bots are being blocked from our websites: If we are blocking AI crawlers that are relevant to our audience, we are losing out on a lot of potential and we might not even be noticing.

It has never been easier to get buy-in for SEO initiatives

People care about GEO, AEO, LLM SEO, LMAO, or whatever they want to call it. They care about it a lot. Some even care about it more than they have ever cared about SEO.

This is great for SEO, as many of the things that have always made you successful in SEO also make you successful in this new world of “O”s:

• Technical and structural accessibility: Make sure that users, crawlers, bots, AI tools and agents can access your website and process your content.
• High quality content with a clear focus: Content that wins with users also wins in search engines and in AI conversations.
• Presence on other platforms: Link building, digital PR and all other evolution phases of the discipline that was once called “off-page SEO” have always been important, and they still are (now powered by AI citation data).

So now is a great time to check that backlog of SEO ideas and projects that never made it to production, re-explain them from an AI-first perspective, and ship!

That’s it (for now)

Thanks for scrolling this far. If you enjoyed this read, or if you absolutely hated it, or if you couldn’t care less, I’m looking forward to hearing your thoughts.

Footnotes:

1. Again! ↩︎
2. This came out of a real-world setup rather than a controlled test, so the results are directional rather than definitive. ↩︎
3. If this sentence sounds like it was written by AI, I’m sorry! I wrote it all by myself (unlike footnote 2). ↩︎
4. This must be true, because I asked ChatGPT 5.2 to review my article and it suggested I add this sentence. I wrote the rest of the article all by myself, pinky promise! (except for, you know, footnote 2) ↩︎
5. Two years ago. ↩︎