Ten Blue Links Are Over. What Replaces Them Decides Who Gets Found.

The ten-blue-links era of search is over — not because Google shut it down, but because a growing share of questions now get answered instead of listed. What replaces the links is a scatter of AI-generated answers across Google, ChatGPT, Perplexity, and Gemini, each naming a handful of sources and businesses where a results page once offered ten. Whether you are one of the names those answers use is quickly becoming the difference between being found and being theoretical.

What actually replaced the links?

Answers did. Google now puts AI Overviews above the results for a widening set of queries and resolves more searches on the page itself. ChatGPT answers the questions people used to type into a search box, and recommends businesses when asked. Perplexity made citation-studded answers its entire premise. Gemini sits inside products half the planet already uses. Search did not die — it dispersed. The underlying behavior — have a question, consult a machine — is healthier than ever. It just no longer reliably ends in a click, and it no longer happens in only one place.

This is not a temporary interface experiment to wait out. An answer is simply a better format for most questions than a list of documents, and users have voted with the only currency that matters: they keep asking. Ten blue links survived for decades because nothing better existed. Something better exists now.

Why are recommendations winner-take-most?

A results page was a democracy of sorts. Ten listings, each visible, each clickable; position three did fine, position seven survived, and everyone got a shot at the scan-and-click. An answer is not a democracy. It names two or three businesses and moves on. There is no position seven in a paragraph.

That changes the economics of visibility. The difference between ranking third and fifth was a gradient — fewer clicks, same game. The difference between cited and not cited is binary: you are in the customer's consideration, or you do not exist for that question. Winner-take-most dynamics reward the businesses an engine can confidently name — the well-documented, well-reviewed, consistently described ones — and quietly starve everyone else. There is no page-two consolation prize anymore. There is not even a page.

Why can't your dashboard see any of this?

Because every instrument was built for the old geometry. Rank trackers measure your position on a page that fewer customer journeys pass through. Analytics undercounts AI-referred visits and cannot see answered-in-place questions at all. You can hold the top organic spot — verified, screenshotted, celebrated in the monthly report — and be absent from every answer your customers actually read.

You can rank first on a page your customer never sees, in a journey that ended inside an answer you are not part of.

The measurement that matches the new behavior is blunt: ask the engines your customers' questions — in varied phrasings, on a schedule, across Google, ChatGPT, and Perplexity — and record who gets named, who gets cited, and how that changes over time. Visibility is now an empirical question you put to the engines themselves, not a number you read off a rank tracker. That is the premise Speak Local's platform is built on, and the founding conviction of this blog.

What do you actually do about it?

The work sorts into two crafts wearing a dozen names. Being quotable: pages structured so engines can lift clean answers — the discipline of AEO. Being cited: an entity engines trust enough to name, built from reviews, consistency, and reputation — the discipline of GEO. Both stand on the same unglamorous fundamentals search has always rewarded: crawlable sites, accurate listings, real answers to real questions. And if you run an agency, this is the next thing clients will ask you to prove. "Do the AIs recommend us?" is already coming up in pitch meetings, and the agencies that can answer it with data will eat the lunch of the ones who answer it with vibes.

The age of answers

Every era of search had a gatekeeper and a game. Directories rewarded being listed. Ten blue links rewarded ranking. Answers reward being known, trusted, and quotable — which, you may notice, is suspiciously close to simply being good and being legible about it. That is the optimistic read of this whole disruption, and we hold it sincerely: the age of answers tilts toward businesses that are genuinely well-documented, well-reviewed, and clear about what they do, not toward whoever bought the most ads. A small firm that answers the question precisely can be the citation. That was rarely true of rankings.

Ten blue links had a good run. What replaces them decides who gets found — and for the first time in a long while, that decision is being made somewhere most businesses are not even looking. Start looking. This post is the first of many about how.

Speak Local
The Speak Local TeamWe measure how machines see local businesses — and write down what we learn.

Quick answers

Is SEO dead now that AI answers exist?
No — it has been demoted from strategy to foundation. AI engines retrieve, read, and cite web pages, so crawlability, indexation, and content quality still gate everything. What died is the assumption that ranking well is the same thing as being found. Today a business can hold the top organic spot and still be absent from the answers a growing share of customers actually read.
What does winner-take-most mean in AI search?
A classic results page spread attention across ten listings; an AI answer names a few businesses and moves on. Being fifth used to mean fewer clicks — now it often means absence. The gap between cited and not cited is binary in a way rankings never were, which makes earning and keeping a place in answers disproportionately valuable for whoever gets there first.
How do I measure visibility in AI answers?
Not with rank trackers alone. The measurement that matches the new behavior is asking the engines themselves — Google's AI Overviews, ChatGPT, Perplexity — the questions your customers ask, in varied phrasings, on a schedule, and recording who gets named and cited. Because answers vary between runs, single checks mislead; the honest signal is the trend across weeks.