The End of Traditional Search: AI Overviews, AI Mode, and What SEO Becomes

Google Search has been doing the same magic trick for two decades.
You ask a question; Google points at places that might know the answer; the internet gets paid in clicks and conversions.

If you run a business that lives on organic traffic, this feels like someone moved your shop inside a maze and started charging admission. Reasonable fear. The fix is not to rename SEO 20 different ways and light incense. It is to adapt your content, technical signals, and demand generation to an interface that answers first and clicks second. This piece explains what’s changing, what’s exaggerated, and what to do next.

That deal is changing – pray they do not alter the deal any further. Not because Google woke up and chose the dark side for fun, but because the interface has evolved into something that can answer in full sentences and hold a conversation. AI Overviews made “answer first” feel normal. Now Google is actively smoothing the path from “summary” to “chat.” Expand an AI Overview, ask a follow-up, and you are no longer really in “results”; you are in AI Mode. On mobile, “Show more” can sit on top of the SERP like a lid; you have to close it to get your links back.

The old “Web” tab is still there, like a manual gearbox. Useful; familiar; increasingly not the default. Also, something you may now need to deliberately drive back to, because the Overview can literally sit on top of the results, and the rumour mill has been running that AI Mode is soon becoming the default.

Somewhere in the distance, you can hear a familiar phrase being warmed up... Not yet. We will come back to it.

TL;DR for businesses

AI Overviews and AI Mode will take some informational clicks. You can still win by becoming the source the model cites; by making your site technically legible; by publishing proof-heavy content that cannot be safely fabricated; and by building demand outside Google so a UI change does not delete your pipeline.

AI Overviews and AI Mode: why publishers keep sweating

AI Overviews sit at the top of results and often satisfy the query before the user clicks. AI Mode is the more conversational version, and Google is making it easier to slide into AI Mode from Overviews, including by letting users ask follow-up questions directly from an Overview while keeping context.

Google’s stated reason is simple: people prefer an experience that flows naturally into conversation, and follow-ups that keep the Overview’s context feel more helpful. Fair. Also, every extra conversational turn is one less click a user needs to make.

The consequence is predictable: fewer clicks. The direction of travel is consistent across analyses. When an AI summary answers the question, fewer people need to click. The exact size varies by query type, but the pattern is most brutal on informational searches.

Google’s response is usually some form of “the clicks are higher quality.” That may be partly true. It is also exactly what you would say if you had just moved the goalposts and wanted everyone to calm down.

AI Overviews are no longer a feature; they are the gateway into AI Mode. If the interface trains users to stay inside Google’s conversation layer, your top-of-funnel content competes with the summary, then competes again with the follow-up thread. The remaining clicks skew toward later stages: people who need depth, proof, or a transaction.

This is mobile-first at the moment; which is how these defaults usually learn your habits before they become everyone’s problem.

The click is no longer the only win condition

This is why “traffic” as the single scoreboard starts to wobble. If Google can solve the problem on the results page, you will lose some top-of-funnel clicks. The clicks you do keep (you would hope) can be more valuable because they come from people who want depth, trust, or a transaction.

At this point, “ranking” also starts to feel like the wrong mental model. You are not only trying to appear; you are trying to be selected, summarised, and cited. Call it SEO if you like.

The operational change is simple: you are optimising for selection, citation, and ranking.

What this means for different businesses

• Lead gen services will lose some early research clicks. They will keep more high-intent clicks. Conversion rate and lead quality will matter more than sessions.
  
• E-commerce will find category discovery gets messier. Comparison and alternatives content becomes more valuable. Product pages need to answer objections cleanly.
  
• Local businesses will feel trust and entity signals matter more. Reviews, listings, and consistency become defensive moats.
  
• Publishers will feel the squeeze hardest on generic informational content. First-party reporting, opinions, tools, and communities serve as the hedge.

And that is when the familiar phrase gets a bit louder in the background.

BUT I’M NOT READY TO SAY IT YET.

If you have spent ten minutes on LinkedIn recently, you have seen the new names: GEO; AIO; AEO; LLMO; “AI Search Optimisation”; “Generative Search Optimisation”; pick your flavour. The naming chaos keeps happening because the interface is changing, and the industry loves a rebrand.

My charitable interpretation is: people are trying to describe a real shift.
My less charitable, more cynical interpretation is: the industry loves rebranding because it can sell “the new thing” at a premium while still doing most of the old thing in a different hat.

Here is the simple truth. At its core, it is still SEO. You are optimising for search behaviour, regardless of the interface. People still seek, still ask, still compare, still decide. SEO cannot die unless people stop seeking. Good luck with that.

What has changed is the shape of the search engine. It is less a list, more a narrator. So yes, GEO and AIO can describe tactics that increase your chance of being cited in AI answers. But do not let the acronyms hypnotise you. Most of the underlying mechanics look suspiciously familiar: clarity, credibility, structure, uniqueness, and technical hygiene.

What SEO becomes when the engine starts answering

Think of modern SEO as three jobs running at once.

1) Make your site machine-readable and trustworthy

This is the unglamorous bit that suddenly matters more. AI systems and crawlers need clean signals: what the page is about; who wrote it; whether it is consistent; whether it is accessible; whether it loads quickly; whether it duplicates other pages; and whether the site’s structure makes sense.

If your site is messy, you do not just rank worse; you also risk losing visibility. You become harder to cite, harder to trust, and easier to ignore.

Practical priorities:

• Clean indexation; no crawl traps; sensible canonicals.
• Internal linking that reflects priority; clear content architecture.
• Structured data that clarifies entities; consistent authorship signals.
• Fast pages; accessible UX; no duplication spread across near-identical URLs.

2) Publish things the model cannot safely invent

If your content is interchangeable, it becomes background noise. The model can paraphrase it without needing you.

The content that tends to win in an AI summary world has at least one of these properties:

• First-hand experience; proof; real examples.
• Original research; benchmarks; audits; data.
• Nuanced comparisons; trade-offs; constraints.
• Clear authorship and credibility signals.

You want to become the source an AI system points to because it cannot confidently produce the same value from generic material. If your page could have been written by anyone, it would be credited to no one.

3) Build demand outside Google so defaults cannot delete you

If Google answers more questions on Google, you need more people arriving anyway.

Brand demand becomes the hedge. Email capture. Partnerships. Digital PR. Paid search targeted at high intent. Content formats that spread without requiring Google’s permission.

If this sounds like “marketing, broadly,” yes. SEO has always been marketing. We just pretended it was a technical hack because that sounded cooler on proposal decks – we’re still dependent on the intent.

A sane order of operations: fix indexation and duplication; tighten internal linking; add entity and authorship clarity; publish one proof-heavy asset; then build distribution that does not rely on Google’s mood.

What Candy Marketing does so you stay visible

If you are reading this with a sinking feeling, here is the practical bit. The goal is not “traffic.” The goal is resilient demand and defensible visibility, even when Google changes the interface again.

Our job is not “rank you number one for ten keywords.” That is a weak promise in an interface that can change shape overnight. Our job is to increase your share of attention and demand across both human- and AI-mediated discovery – the bottom line, will always be the bottom line.

Technical eligibility, so you remain citeable

Indexation control; duplication cleanup; internal linking; performance; schema that clarifies entities; content architecture that makes your priorities obvious. You stay eligible to be crawled; understood; and cited.

Cite-worthy content assets, not filler

We plan content around what AI systems struggle with unless they have strong sources: decisions; costs; compliance; comparisons; edge cases; and local constraints. Then we build pages that answer follow-ups, not just the first question. That is how you win citations and the remaining clicks.

Entity and credibility signals across the web

If the engine is trying to understand “who is this,” we align authorship, About pages, credentials, citations, listings, and brand mentions so you look coherent and trustworthy. The engine understands who you are and why you deserve trust.

Measurement that survives the zero-click era

Leads and revenue. Brand query growth. Share of voice. Assisted conversions. Conversion quality on the traffic you still earn. You can prove revenue impact when sessions wobble. If you only measure sessions, you will spend 2026 yelling at graphs.

A blunt prediction

Expect more zero-click behaviour as Overviews and AI Mode become more embedded. Google is making the conversational path easier, including by encouraging follow-up questions from AI Overviews that jump users into AI Mode conversations. The direction of travel in reported analyses remains depressed click rates when AI summaries satisfy the query.

This is annoying. It is also workable if you stop chasing the old model of volume by default and start building a system that wins selection and converts the traffic you do get.

SEO IS DEAD

There I said it, are you happy?

The King is dead.
Long live The King.

I’ve always wanted to write the “SEO is dead” article. The industry runs it like an annual ritual; someone declares it, everyone nods gravely, then we all go back to chasing the spotlight like the rockstars we are, or as the professionals call it, “Visibility”. This year, the argument feels less like theatre and more like a genuine identity crisis.

The real question is a Ship of Theseus one: how many planks can you replace before you admit you are sailing a different vessel? Same destination; same sea; new hull, new rigging, and now there’s a concierge at the gangway giving everyone the “quick summary” before they even bother to come aboard.

What dies is the comfy version; the version where “publish broadly similar content and wait for traffic” counts as a plan. What survives is closer to engineering and reputation: make your site legible; make your expertise undeniable; publish things the model cannot safely invent; build demand outside the SERP; then make conversion painless when users do arrive.

Search does not end. People still seek. They always will.

The interface just got cocky and started talking back.