The Pyrrhic Victory of GEO

GEO has won the battle for attention. But if businesses treat it as a replacement for SEO, they neglect the things that make them visible in AI search in the first place.

Apparently, SEO is old news again. This time, the replacement is GEO: Generative Engine Optimisation. Or AEO. Or LLMO. We still have not agreed on the name, although GEO seems to be winning the race. That has not stopped people selling it as if they had found the Holy Grail.

I am not dismissing AI search. People use ChatGPT, Gemini, Perplexity and Google’s AI features to research products, compare businesses and make decisions. Companies need to understand how they appear in those answers.

What I question is the idea that GEO has replaced SEO, rather than expanded the job SEO and the wider marketing team need to do.

The concern is what happens when businesses move money and attention towards AI visibility while pushing the less exciting SEO work aside. If that means neglecting the technical foundations, content, evidence and customer understanding on which AI visibility also depends, GEO may win the internal argument while the business loses something rather more useful. Possibly money.

That would be a Pyrrhic victory.

GEO has won plenty of attention

A Clutch survey of 600 marketing professionals found that 78% of the companies represented were funding GEO. That was almost identical to the 77.5% funding SEO and PPC.

Nearly two-thirds expected to increase their GEO investment during the following 12 months. At the same time, 43% of respondents thought SEO was losing ground.

This was one industry survey, conducted through Pollfish in August 2025. It is not evidence that the entire business world has abandoned SEO. It does show that GEO has moved into budget discussions surprisingly quickly.

That is understandable. SEO has had more than 25 years to accumulate baggage. Depending on your experience, the term may bring to mind technical backlogs, endless keyword lists, content nobody reads or monthly reports celebrating movements that had no noticeable effect on the business.

GEO arrives without most of that history. It comes with colourful dashboards, prompt tracking and the promise that your company can be recommended by ChatGPT. That is easier to take into a budget meeting than another discussion about crawlability or information architecture.

It is not necessarily more important. It is simply easier to make sound new.

The problem is not that businesses are investing in GEO. They should be looking at AI-driven discovery. The problem is what they think they no longer need.

The new work looks rather familiar

Strip away the terminology and much of GEO consists of work we should have been doing already.

A business needs to explain clearly what it does. Its important information should be accessible to machines as well as people. Claims should be supported by evidence. Facts about the company should be consistent. Other credible sources should know that the business exists and have something useful to say about it.

All of this was important before ChatGPT appeared.

Good SEO was never just putting keywords into title tags. It included technical accessibility, understandable site structures, useful content, internal linking, authority and a reasonable attempt to match what people were actually looking for.

Digital PR, brand building and reputation were not invented by GEO either.

The systems and interfaces have changed. Some of the analysis needs to change with them. But we have not discovered a completely new branch of physics.

AI systems still need something to work with

AI search does not float above the web collecting wisdom from the atmosphere.

It needs information it can retrieve, process and use. The exact process varies by platform and is mostly hidden from us. How fully even the people building these systems understand every part of them is another question. ChatGPT does not work in exactly the same way as Gemini. Perplexity does not choose sources in exactly the same way as Google’s AI Overviews.

Anyone offering a single formula for all of them is selling snake oil.

For Google’s own AI features, the relationship with SEO is at least clear. Google says existing SEO best practices remain relevant to AI Overviews and AI Mode. To appear as a supporting link, a page needs to be indexed and eligible to appear in Google Search.

Google recommends allowing crawling, using internal links, making important information available as text and ensuring structured data matches what people can actually see.

In other words, the boring things still matter.

If the real problem is crawlability, indexation, rendering or site architecture, a focused technical SEO audit is likely to be more useful than buying another GEO dashboard.

There is no special AI schema. There is no magic file that turns a weak website into a trusted source. Adding a summary box to an article does not repair a confused business proposition.

For other LLM-based systems, conventional Google rankings are not a universal prerequisite. But those systems still need source material. They still benefit from clear information, consistent facts, credible evidence and recognition beyond the company’s own website.

The relationship is not a clean ranking formula, and it will differ between systems. But as a rough way of showing what both conventional and AI-driven visibility depend on, this is useful:

If the underlying material is poor, renaming the activity will not improve it.

A citation is not necessarily a win

We should also be careful about what GEO success means.

The SEO industry spent years training businesses to celebrate rankings and traffic without asking whether either contributed much to the business. We now seem determined to repeat the exercise with mentions and citations.

A citation tells you that an AI system used a source. That can be useful. It does not necessarily mean the system recommended the company behind that source.

Your article might be cited as evidence while the answer recommends three competitors. The GEO dashboard records another citation. The prospective customer remembers somebody else.

The reporting looks healthy. The commercial result may be rather less impressive.

Mentions, citations and recommendations are different things. They should not be bundled into one reassuring visibility score.

There is also the problem of stability. AI answers are generated dynamically and can change when the prompt, user, model or retrieval process changes.

An AirOps study of more than 45,000 citations found that only 30% of visible brands remained present in consecutive responses. Only one in five stayed visible from the first tested run to the fifth.

AirOps sells AI-search software, so this is vendor research and should be treated as such. Nevertheless, the result illustrates something anyone testing these systems will quickly notice: the answers move around.

This does not make tracking useless. It means we should treat the numbers as signals, not rankings with a different label.

What matters is whether the brand is represented accurately, appears in relevant commercial conversations and is recommended for the right reasons. Eventually, we also need to know whether any of it contributes to enquiries, sales, branded demand or another worthwhile result.

Otherwise, GEO becomes the same old reporting theatre with newer screenshots.

Fast growth is not the same as large scale

AI referral traffic is growing quickly.

Similarweb estimated that AI platforms generated around 1.13 billion referral visits in June 2025, up 357% from the previous year.

That sounds enormous, and it is. In the same analysis, Google Search generated an estimated 191 billion referrals.

Referral visits do not tell the whole story. An AI answer may influence somebody without generating a click. Google also answers many queries directly, particularly through its own AI features. We should not pretend these figures measure the full influence of either system.

They do provide some perspective.

AI discovery matters, but conventional search has not disappeared. Businesses are still finding products, services and information through Google in very large numbers.

Moving all the attention to GEO because it has the faster growth rate would be like abandoning a profitable shop because a new market opened down the road.

Visit the new market. Understand it. Sell there if your customers are there. Perhaps keep the shop open as well.

SEO provides something GEO cannot currently replace

Traffic is only part of the value of SEO.

Search data helps businesses understand how people describe their problems. Queries reveal comparisons, objections, demand and the language customers use before they know the approved terminology from the company brochure.

That information can improve more than a search campaign. It can influence product positioning, website navigation, sales material and customer support. It is market intelligence, not just channel reporting. I have written separately about using search data that way rather than leaving it buried in an SEO report.

AI conversations are largely hidden from businesses. We can create prompt sets and test how systems respond, but we cannot see a dependable, platform-wide record of what prospective customers are asking.

GEO tools usually monitor questions chosen by the marketer, supplied by the client or generated by the tool. That can be useful, but it is not the same as observing the market.

There is a danger that companies will gain an AI share-of-voice score while losing interest in the much less glamorous work of understanding actual customer demand.

They will know that they appeared in 37% of their own test prompts. Whether those prompts reflect real buying behaviour is another question.

Some SEO work deserves to lose its budget

That may sound surprising coming from an SEO consultant. It should not. This is not a nostalgic defence of everything sold as traditional SEO.

If SEO means producing hundreds of interchangeable articles, chasing traffic with no commercial relevance or delivering reports nobody uses, cutting the budget is entirely sensible.

AI search is also exposing weaknesses that conventional SEO reporting sometimes hid. A company can rank while remaining poorly understood. It can attract traffic without being trusted. It can publish endless claims about itself while nobody independent supports them.

GEO brings some useful questions into the room:

• Do AI systems understand what the company does?
• Are important company facts consistent?
• Is the brand mentioned when people compare solutions?
• Do independent sources support its claims?
• Is it being recommended, or merely used as a reference?

Those are good questions. They should widen the SEO and organic visibility conversation. They should not be used to pretend the old foundations have become irrelevant.

GEO and SEO are not opposing channels

A Semrush survey of 481 marketers, business owners and SEO professionals found that 38% planned to invest in AI-search optimisation, compared with 36% planning investment in traditional SEO.

Only 22% said their SEO and AI-search activity was fully integrated across strategy, execution and reporting. Interestingly, 77% already described AI search as sitting inside or extending SEO rather than being a completely separate channel.

That is probably the more useful finding.

Customers do not care how we divide the work internally. They may discover a company in an AI answer, check it on Google, read reviews, visit the website and then ask somebody in a community whether it can be trusted.

That is one customer journey, not five marketing channels politely waiting for their turn.

I do not think businesses should choose between SEO and GEO. That distinction may be useful when discussing platforms, tools or specific pieces of work, but customers do not behave according to our channel definitions.

The practical job is broader: make the business technically accessible, explain it clearly, support its claims and understand what customers are trying to find. Then examine how that information is represented in conventional search and AI-generated answers.

Deciding where SEO, GEO and wider brand activity deserve investment is a strategic SEO question before it is a tooling question.

GEO deserves investment where the audience and commercial opportunity justify it. But if that investment comes at the expense of technical foundations, useful content, authority and customer understanding, the business may win a new visibility score while weakening the work that made it visible in the first place.

That would be the Pyrrhic victory of GEO.

---