AI search didn’t “kill SEO.” It just changed what SEO is for.
For the last couple of decades, search was pretty straightforward. You typed a question into Google, you got a list of links, and the game was to get your page as close to the top as possible. If people clicked your result, you won. If they didn’t, you didn’t.
That model still exists, but it’s no longer the whole experience.
More and more often, Google (and other tools) will show an AI-written summary before you ever see the list of links. And when that summary shows up, people click less and they stop digging sooner. In one Pew study, users clicked regular search results about 8% of the time when an AI summary appeared, compared to 15% when it didn’t.
That might sound like a small behavioural shift, but it changes the incentives in a big way.
We’re moving from “here are some websites you can read” to “here is the answer.” And once search becomes an answer-first interface, the question isn’t “How do I rank?” so much as “How do I get included in the answer?”
What’s actually changing
1) There isn’t one “top result” anymore
Traditional search gave you something you could point to. If you were #1 for a keyword, you could track it, report it, and celebrate it.
With AI search, that clarity starts to dissolve. Two people can ask the same question and get different answers, because the system may shape the response based on personal context, preferences, or past behaviour. Google has been pretty public about moving in this direction, where results aren’t always one-size-fits-all.
So visibility becomes more personal, and less like a single leaderboard everyone shares.
2) SEO still matters, but it’s not the finish line
None of the basics go away. Your site still needs to be crawlable. Your pages still need to be structured clearly. You still need signals of trust, authority, and relevance. If the site can’t be discovered or understood, you don’t even get a seat at the table.
But ranking is no longer the main prize. Think of it more like a ticket to be considered.
The bigger question becomes: will the AI pull your information into the answer?
And here’s the uncomfortable part for brands. AI systems often lean on sources that look independent and verified. That usually means third-party material: journalists, reviewers, analysts, industry publications, and other reputable sites.
So even if your own website is polished and technically perfect, you can still lose visibility if the wider web does not describe you clearly, consistently, and credibly.
3) Some “AI search” is basically regular search plus a summary
A lot of AI search still follows a familiar recipe:
- Run a normal search.
- Pull a set of top pages.
- Ask an AI to summarise what it found.
In that world, you still benefit from classic SEO because you still need to be one of the pages that gets chosen.
But the bigger shift shows up when AI stops treating a page as a single unit.
4) AI doesn’t always use whole pages. It uses pieces of pages
Many systems break pages into chunks. It might be a paragraph, a definition, a few sentences, a table row. Then they retrieve only the most relevant chunks and assemble an answer from those pieces.
So instead of “winning with a full page,” you win by having clear, useful snippets that stand on their own.
That’s why “chunking” (how content is split and retrieved) has become such a big topic in modern AI search and retrieval systems.[5]
5) “Correct” increasingly means “widely supported,” not “sounds clever”
In the old content world, a lot of strategy was about saying something impressive or distinctive. You wanted a fresh angle, a bold claim, a unique point of view.
In AI search, what matters more is whether the system can retrieve enough good evidence to keep the answer accurate.
AI answers are heavily shaped by what gets pulled into the system’s “context window,” which is just the information the model is allowed to consider while responding. Context windows are getting larger, but they still have limits and tradeoffs.[6]
If retrieval is strong, the AI has less room to improvise.
If retrieval is weak or messy, the AI can still sound confident while being wrong.
This is also where PR and earned media start to matter in a new way. If a brand is mentioned across multiple reputable, independent publications, AI systems have more “outside evidence” to pull from. That tends to look less biased than a brand describing itself, and it gives the AI more confidence that the claim is real.
6) The old “search deal” is breaking
For years, there was an unspoken trade:
- Google sends traffic.
- Websites create content.
- Everybody benefits.
AI summaries strain that deal. You can get visibility without getting clicks, because people get what they need right there in search.[1][7]
That’s painful for publishers, but it also affects any business that relied on a predictable funnel of:
impressions → clicks → conversions
McKinsey has argued this shift could mean meaningful traffic drops from traditional search, especially for brands that aren’t prepared.[8]
7) People are using AI to decide earlier
One reason clicks are dropping is simple: people are forming opinions before they ever land on a website.
Research suggests plenty of consumers intentionally use AI search tools, and many treat them as a primary source when making buying decisions.[8][9]
So by the time someone reaches your site, they may already:
- trust you, or
- have ruled you out.
8) The real competition is: “Will the AI retrieve you?”
In the old world, competition was visible. You could literally look at the rankings.
In the new world, a lot of the critical action is hidden. Two AI systems might look at the same internet, but they can retrieve different snippets, trust different sources, and produce different answers.
So the questions that matter start to look like this:
- Will your content break into clean, reusable chunks?
- Do your key statements still make sense when taken out of context?
- Are your claims repeated and supported across other trustworthy sources?
- Are your brand, product, and category facts described consistently across the web?
- Do reputable third parties describe you in the same terms you use about yourself?
In practice, a lot of “AI visibility” is won off-site.
Strong owned content helps, but consistent, credible mentions across independent outlets often carry more weight than a single company web page.
Put simply, it’s not just about being indexed anymore.
It’s about being retrieved and used.
The plain-English takeaway
If you only remember a few things, make it these:
- Small pieces of content matter more than whole pages.
- Ranking matters less as a “win” and more as a “ticket to be included.”
- Clicks are no longer guaranteed.
This isn’t a moral judgement. It’s just what happens when the interface changes.
Search engines are increasingly comfortable giving the “so what” directly, and only pulling in the evidence they choose.
As that becomes the default, everyone will have to adjust how they measure success and how they create content that actually shows up inside answers.
Sources cited
- Pew Research Center: https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/
- Seer Interactive (2025)
- Databricks: https://community.databricks.com/t5/technical-blog/the-ultimate-guide-to-chunking-strategies-for-rag-applications/ba-p/113089
- IBM: https://www.ibm.com/think/topics/context-window
- McKinsey “New front door to the internet: Winning in the age of AI search:” https://www.mckinsey.com/capabilities/growth-marketing-and-sales/our-insights/new-front-door-to-the-internet-winning-in-the-age-of-ai-search
- Adobe: https://business.adobe.com/blog/generative-ai-powered-shopping-rises-with-traffic-to-retail-sites