SEO in the Age of AI Search

Eight years behind a record store counter taught me something nobody puts on a résumé: the store doesn't die when a new format shows up. It dies when the inventory is mislabeled and nobody can find what they came in for. Vinyl didn't kill the store. CDs didn't kill the store. Streaming changed what people bought, not whether the crates needed to be alphabetized correctly. AI search is the new format. Same rule applies.

A lot of what's being said about AI Overviews and answer engines right now is either doom or hype, and both versions are wrong for the same reason: they treat this like a format change that rewrites the rules, instead of a format change that rewards the shops that already had their inventory in order. Here's the technician's read, no theatrics.

What actually changed

Informational queries — the "what is," "how does," "why does" searches — increasingly get answered directly on the results page or inside a chat interface, without a click. If your traffic depended on ranking for broad, top-of-funnel questions, some of that traffic is gone, and it's not coming back by writing a longer version of the same article. That's the real change, and it's worth saying plainly instead of hedging around it.

But the click that survives is a better click. Someone who still leaves the answer engine to visit a site is doing it because the summary wasn't enough — they want more depth, they want to compare options, or they're ready to act. That's a higher-intent visitor than the person who used to land on your page, skim one paragraph, and bounce. Fewer visits, better visits. That trade isn't a catastrophe. It's a different inventory mix.

Zero-click pragmatism, not zero-click panic

A search that ends without a click isn't automatically a loss. Someone who asks an answer engine for your business hours, your address, or whether you're open on Sunday and gets a correct answer instantly has had a good experience with your business, even though nothing hit your analytics. That query was never going to convert into a session worth obsessing over. What you actually want is for the zero-click answer to be correct, which depends entirely on whether your structured data and your Google Business Profile agree with each other in the first place.

The pragmatic move is separating queries by what a click was ever going to be worth. Chasing traffic on a query that was always going to be a zero-click lookup — store hours, a phone number, a yes/no fact — is chasing a number that was never revenue to begin with. Spend the effort instead on the queries where a visit was always the point: comparing options, checking your work, deciding whether to book. Sorting your own query list this way, honestly, tells you where AI Overviews actually cost you something and where they just automated a lookup nobody was going to convert on anyway.

What didn't change

Crawlability didn't change. If an AI system or a search engine can't crawl your site cleanly — broken internal links, blocked resources, a JavaScript-rendered mess with nothing in the initial HTML — it can't cite you, summarize you, or rank you. That was true in 2015 and it's true now. Nobody gets to skip the crawl.

Entity clarity didn't change either. Answer engines still need to understand who you are, what you do, and where you do it, unambiguously. That's structured data, consistent business information, and a site architecture that doesn't make a machine guess. Being the source worth citing was always the job. It's just gotten more literal — you're now sometimes cited by name in a generated answer, which is a direct, visible reward for exactly the fundamentals that used to only pay off indirectly through rankings.

Structured data and consistent NAP are how a machine reads your shelves

Schema markup, consistent name-address-phone data across your site and every directory listing, clear service and pricing information in machine-readable form — this was always good practice. Now it's closer to mandatory, because an AI system summarizing local options doesn't have the patience a human has to piece together your hours from three different pages that don't agree with each other. If your NAP is inconsistent across your site, your Google Business Profile, and your directory listings, you're not just confusing Google's local algorithm the old-fashioned way — you're giving an answer engine a genuine reason to cite a competitor whose data is clean instead of yours.

Think of it as alphabetizing the crates. Nobody thanks you for it directly, but the person looking for the record finds it faster, and the system that's summarizing your inventory for somebody else can actually do its job.

Local intent is more insulated than most categories, and here's the actual reason

An AI Overview can summarize the history of hardwood floor refinishing in three tidy paragraphs. It cannot refinish your floor. It cannot show up at your house Tuesday. It cannot fix your leaking pipe, fill your cavity, or show a bachelorette party to their table. Local, service-based, transactional intent survives the shift better than broad informational intent because the answer engine can compress information, not perform a service. The demand for the actual thing — a plumber, a dentist, a restaurant table — still has to resolve to a business somewhere, and that resolution still runs through local search and Google Business Profile.

That doesn't mean local businesses are exempt from any of this. It means the informational content wrapped around a local business — the blog post explaining what a service is — is more exposed than the transactional pages actually driving bookings. Know which of your pages are which before you decide where to spend effort.

What to stop doing

Stop writing thin informational content whose entire purpose was to rank for a broad question that an answer engine now handles better than a 600-word blog post ever did. "What is local SEO" as a standalone page competing on pure definition is a fight you're increasingly fighting against a machine built specifically to win it. If a page's only job was to catch that kind of query and funnel a click, expect that funnel to keep narrowing, and don't keep pouring production budget into more of the same shape of content hoping volume fixes it.

Stop treating keyword-stuffed definitions as a content strategy, too. Five hundred words restating what a service is, with the target phrase wedged in every third sentence, was always weak content — it just used to squeak by on volume alone. An answer engine reads that page exactly as thin as it is, which means it's neither getting cited nor getting the click. It's occupying a slot in your content calendar that a genuinely useful page should have instead.

What to double down on

First-hand expertise that a summarizer can't fabricate: real case detail, real photos of real work (not stock), specific numbers from specific jobs, the kind of texture that comes from having actually done the thing. That content is harder to compress into a generic answer because it isn't generic to begin with, and it's exactly what still earns a citation instead of getting summarized around.

And double down on the pages that convert the clicks you still get, because you're getting fewer of them and each one matters more. A contact page with friction, a service page without a clear next step, a slow page that loses someone before it loads — those cost more now than they did when traffic volume was higher and could paper over a weak conversion path. Tighten the pages doing the actual business, not just the pages chasing the traffic.

None of this is a five-alarm fire and none of it is nothing. It's a format change. The stores with clean inventory and staff who know where things are keep selling records. The ones that were coasting on foot traffic and messy shelves were always going to have a bad year eventually — AI search just moved up the timeline.