SEO in the Age of AI Search: What Still Compounds

Two numbers, before any opinion. First: across the client accounts we manage, the share of Google queries returning an AI Overview has climbed steadily for two years and now sits well above half for informational queries — the "how does," "what is," "why do" category. Second: click-through rate on traditional organic results below an AI Overview has dropped, but not to zero, and not evenly. Local and transactional queries — "roofer near me," "book a table Golden CO" — have held up far better than pure informational ones. Those two numbers together are the whole story. Everything else is commentary.

The commentary has mostly been unhelpful, splitting into two camps that are both wrong in the same way: the panic camp ("SEO is dead, pivot to something else immediately") and the denial camp ("nothing has changed, keep doing exactly what you were doing"). Both are reading the data selectively. The panic camp extrapolates from the worst-hit content category — generic informational blog posts — and assumes every business looks like that category. The denial camp looks at its own local rankings, sees they've held up fine, and assumes nothing structural has shifted anywhere. Neither conclusion survives contact with the full data set. Here is the less exciting, more accurate read, in the order I'd brief a client on it.

1. What actually changed

Zero-click behavior isn't new — Google has been answering simple queries directly in the SERP since featured snippets arrived years ago. What's new is the volume and the format. AI Overviews synthesize across multiple sources into one answer block, which means a page can contribute to an answer without earning the click that page used to get for ranking first. For a business whose organic strategy was built entirely on informational content — blog posts answering generic questions with no local or transactional intent — this is a real hit to top-of-funnel traffic. For a business whose organic strategy was built on local and transactional intent, the hit is smaller than the headlines suggest.

2. Why extractable content wins the citation

AI Overviews and answer engines pull from sources that make the answer easy to lift cleanly: a direct question restated as a heading, a concise answer in the first sentence or two beneath it, then supporting detail. Content that buries the answer under three paragraphs of preamble is harder to extract and gets cited less, regardless of how accurate it is. This isn't a trick — it rewards the same clarity that has always helped human readers scan a page, just with a machine doing the scanning first. Structure your headings as the actual questions a person would ask. Answer them immediately. Elaborate after.

3. Entity and brand signals matter more, not less

Answer engines are increasingly deciding whose answer to trust based on entity recognition — does this business have a consistent, verifiable identity across the web — rather than page-level keyword matching alone. That means your Google Business Profile, your consistent name-address-phone data across citations, your structured data, and your presence in industry-relevant directories all feed a trust signal that's larger than any single page. A perfectly written page from an entity Google can't clearly identify and verify is worth less than a good page from a well-established one. In practice this means the unglamorous work — claiming and fully completing a Google Business Profile, matching your business name and address exactly across every citation, linking your social profiles back to a consistent canonical site — now sits upstream of your content strategy rather than beside it. We've moved entity cleanup earlier in new engagements for exactly this reason.

4. Structured data's job has changed, not shrunk

Schema markup used to be mostly about earning rich results — star ratings, FAQ dropdowns, that kind of visual real estate in the SERP. It still does that, but its more important job now is machine legibility: giving an AI system an unambiguous, structured version of the facts on your page (services offered, service area, pricing range, business hours) so it can be cited with confidence instead of skipped for being hard to parse. Sites without clean structured data aren't just missing rich snippets anymore. They're harder for an answer engine to trust at all.

5. Local intent is the most defensible ground on the map

Here's the number worth sitting with: "near me" and city-qualified searches convert to a click or a call at a rate the purely informational category simply doesn't, because an AI Overview can summarize what a service is, but it can't drive to your shop or take your appointment. The map pack and local pack results have remained comparatively insulated from AI Overview cannibalization because the intent behind them is inherently transactional — a person searching "HVAC repair Lakewood" wants a business, not an essay. If your organic strategy leans local, you are standing on the most defensible ground available right now.

6. The Front Range angle: why local businesses are less exposed than SaaS blogs

Worth naming directly, since most AI-search commentary is written from the perspective of a national content publisher, not a Denver contractor. A software company's blog post explaining a generic concept is exactly the content type an AI Overview replaces most efficiently — the answer is portable, unbranded, and equally valid coming from any source. A page explaining what a Boulder outdoor brand actually stocks, what a Golden brewery's taproom hours are this weekend, or which neighborhoods in Lakewood a roofing crew services isn't portable in the same way. The answer depends on the specific business, which is precisely the category of query an answer engine still has to send somewhere to be resolved. This is a structural advantage for the kind of business we work with, and it's one that doesn't show up in the aggregate industry statistics, because those statistics are dominated by content categories more exposed than local service businesses are.

7. Measuring AI-referred traffic honestly

Most analytics platforms still attribute AI Overview clickthroughs and answer-engine referrals inconsistently, which means the honest answer to "how much traffic do we get from AI search" is currently "less precisely than we'd like." What we track instead: referral traffic from known AI platforms where it's identifiable, branded search volume over time (a decent proxy for whether AI-mediated exposure is building brand awareness even without a click), and — the metric that actually pays the bills — leads and bookings by channel, checked quarter over quarter rather than week over week, because the noise-to-signal ratio at shorter intervals isn't worth the anxiety it produces.

8. What still compounds regardless of the format war

A few things haven't changed and, on current evidence, aren't going to: a fast, technically sound site still gets crawled and trusted more easily than a slow one. Original, specific content still outperforms generic content in both traditional rankings and AI citation rates, because both systems are ultimately trying to reward the same thing — genuine expertise over reformatted competitor content. And a strong local reputation, consistent NAP data, and real structured data compound in value precisely because they're the inputs every new search format, from featured snippets to AI Overviews to whatever comes after them, has leaned on more, not less. Betting on fundamentals has a pretty good track record of not aging badly.

None of this requires a new strategy so much as tighter execution of the one that was already correct. The panic camp and the denial camp will keep arguing about whether SEO is dead. The data suggests it just got pickier about who it rewards — which, if you've been doing the fundamentals properly all along, is good news dressed up as an existential threat.