The StartingUp Summer, continued. We're using the season to take a step back on a year that shook up online visibility. After laying out what GEO actually covers, it's time to settle the question everyone keeps asking us: SEO or GEO, should you still write for Google?
The subject now finds its way into every marketing meeting. On one side, hasty funerals: "SEO is dead, everything now happens in ChatGPT." On the other, comfortable denial: "just a fad, our Google rankings are enough." Two opposite postures, one thing in common: both let you off the hook of thinking. And the numbers, which we checked one by one, don't side with either.
The short answer: SEO isn't dead, and GEO doesn't replace it, it stacks on top of it. Google still sends roughly 190 times more visits to sites than ChatGPT, but nearly two searches out of three already end without a click. Writing for Google therefore remains the foundation; becoming citable by generative engines is the new floor above it.
This article settles this false debate with facts, then offers a simple grid: depending on whether you're after awareness, leads, sales or support, the balance between SEO and GEO isn't struck in the same place.
"SEO is dead": the corpse is doing just fine
Let's start with the order of magnitude the obituaries forget. Google handles on the order of 5 trillion searches a year, more than 13 billion a day, and that volume is still growing year over year. ChatGPT receives billions of daily prompts, but only a minority of them reflect a search intent: depending on the calculation method, its "search" usage represents anywhere from a few percent to a bit over 10% of Google's volume. Considerable for such a young product; marginal next to the main faucet.
More importantly, the traffic actually sent to sites remains in a different league: according to figures published by Ahrefs in early 2026, Google still sends about 190 times more visits to websites than ChatGPT. And overall search usage, assistants included, keeps growing: AI hasn't eaten the pie, it has made it bigger.
The second argument is more structural: generative engines feed off search indexes. ChatGPT Search relies mainly on Bing's index, supplemented by its own crawler; AI Overviews draw, unsurprisingly, from Google's index; Perplexity overwhelmingly cites domains present in Google's top 10 for the same questions. A poorly indexed, poorly structured site closed off to robots isn't just invisible on classic search engines: it doesn't exist for any of these machines. Being findable on Google and Bing is no longer an end in itself, but it remains the ticket to entry, as we detailed in our levers to appear in LLM answers.
"GEO replaces SEO": the novelty that recycles
The opposite argument deserves the same treatment. Listening to some peddlers of certainty, you'd think you had to throw out twenty years of practice and adopt an entirely new discipline. Yet when you line up the recommendations that reach consensus in GEO, what you find is... the fundamentals of serious SEO:
- Structured content: hierarchical headings, sharp answers up front in each section, tables and lists that machines and rushed readers grasp in one glance.
- Citable data: dated and sourced figures, straightforward definitions, signed positions, everything a model can reuse while crediting you.
- Authority: being mentioned in the press, comparison sites and your industry's forums, what SEO already called reputation and link building.
- A technically sound house: crawlability, speed, structured data, robots allowed in.
This overlap is no accident: models learned to judge quality on a web already shaped by Google's criteria. What GEO genuinely adds (optimizing for citation rather than for the click, the weight of brand mentions with no link, a measurement discipline still in its infancy) makes it a demanding extension of the trade, not its replacement. Whoever sells you one against the other is selling you the same thing twice, or half of each.
"Nothing changes": the comfort that costs the most
That leaves the third posture, the most widespread because it's the most restful: since SEO isn't dead, let's keep doing what we've always done. That means ignoring three well-documented shifts.
The share of no-click searches keeps climbing. About two out of three Google searches now end with no click to a site at all, a proportion that has sharply accelerated since generative summaries rolled out. And when an AI Overview shows up, the click-through rate of the first organic result drops by anywhere from a third to more than half, depending on the study. We broke down this mechanism and how to counter it in our analysis of the impact of AI Overviews on your traffic.
Ranking no longer guarantees citation. An Ahrefs study covering more than 18,000 queries measured it: only 12% of the URLs cited by AI assistants appear in Google's top 10 for the original question. The domains that hold authority remain largely the same, but the pages chosen differ: ranking first is no longer enough, being citable becomes a property of each individual page.
Measurement is shifting, and some formats are fading out. Rankings and organic sessions tell a shrinking share of the story; you now also need to track your mentions and citations in generative answers, with tooling that's still young. As for content that only ever lived off the quick-information click (generic definitions, interchangeable listicles, copied FAQs), the machine answers it directly: its reason for existing disappears.
SEO or GEO: where to invest based on your objective
So do you have to choose? Yes, but not between SEO and GEO: between your objectives. The right balance depends on what your site needs to produce.
| Your objective | What's changing | Where to invest first |
|---|---|---|
| Awareness | The brand is increasingly built with no click, inside answers composed by machines | Press mentions and comparisons, signed citable data, opinion content; listen to what assistants say about you |
| Lead generation | High-intent queries keep a good click-through rate; visitors arrive later, better informed | Deep expertise pages, case studies, honest comparisons: SEO and GEO converge here almost entirely |
| E-commerce | Product summaries and shopping agents are emerging; classic search and Shopping remain massive | Flawless technical SEO, structured product data, up-to-date feeds and reviews: the foundation serves both channels |
| Support and after-sales | Documentation gets summarized without a visit, and that's actually good news | Public, clear, structured documentation; measure tickets avoided rather than traffic |
The overall reading: the higher your objective sits in the funnel (awareness), the bigger the GEO share of the effort grows; the more transactional it is, the more the SEO fundamentals stay decisive. No row in the table settles for just one of the two.
Writing for Google remains the foundation; becoming citable by generative engines is the new floor above it.
Writing for whom, then?
Neither for Google, nor for LLMs. The question itself is a trap inherited from twenty years of tactical SEO, where you wrote for the algorithm hoping the reader would follow along. Generative engines, paradoxically, reward the opposite: they cite what helps them answer, meaning lived experience, the sourced figure, the position you actually own, the page that says something a thousand others don't.
So write what both want to cite: content that carries authority. That's more demanding than yesterday's SEO, which tolerated optimized filler; it's also fairer. For the first time in a long while, the machines' interest and your readers' interest point in exactly the same direction.
