Every few years this industry produces a wave of panic that Google is dying, and every time the actual answer turns out to be "search is changing shape, not disappearing." AI Overviews and chat-first search are the real version of that this time — the shape genuinely is changing — but the panic is still mostly overcorrection. Here is the level-headed version.
What is actually shrinking
AI Overviews sit at the top of a meaningful share of informational queries now, and for a lot of purely informational searches — what is the boiling point of X, how do I convert Y to Z — the answer satisfies the user before they ever click a blue link. That traffic is genuinely gone for the sites that used to capture it purely by answering a simple factual question well. Chat-first tools pull some of that same informational research phase out of search entirely, happening in a conversation instead of a series of queries.
The informational long tail — the broad, top-of-funnel content that used to be an SEO strategy's bread and butter purely for traffic volume — is the hardest hit. If your content strategy was built entirely around ranking for "what is [broad topic]," that specific play is weaker than it was three years ago.
What still works, plainly
- Brand search. When someone searches your business name, they already decided you might be the answer. AI Overviews barely touch this category, and it converts at a rate nothing else does.
- Local intent."Near me," map pack, and service-area searches remain almost entirely untouched by AI Overviews, because the answer genuinely requires a business with a real location and real availability, not a synthesized paragraph. This is most of why we still lean hard into local SEO for service-based OC clients.
- Transactional queries."Buy," "book," "hire," "quote" — anything with clear purchase intent still sends a click, because an AI summary cannot complete the transaction for the user.
- Genuinely differentiated expertise. Content that reflects real experience, a specific point of view, or original data still gets cited and still gets clicked, because it is the thing a generic model answer cannot fully replicate. Generic, reworded "top 10 tips" content is exactly what AI Overviews now do better and faster than most websites ever did.
Being the cited source instead of the ranked link
A newer, real metric worth tracking alongside rankings: is your content the one AI Overviews and chat tools cite when they answer a question in your space? The mechanics for earning that citation overlap heavily with good SEO fundamentals — clear structure, direct answers near the top of the page, credible sourcing — but reward specificity and originality more than older SEO did. A page that says something no other page says, backed by real data or real experience, is more likely to get pulled into a synthesized answer than a page that competently repeats consensus information.
This is not a reason to abandon ranking for clicks — most transactional and local traffic still comes through a click, as above — but it is a reason to write content that would survive being the only source an AI model has to draw from, rather than content that is a slightly reworded version of the top five existing results.
A concrete example, because the abstract version is easy to nod along to
Take a home services client with a page targeting "how often should I replace my AC filter." Two years ago that page earned a steady trickle of informational traffic and occasionally converted a reader into a service call. Today, a good portion of that exact query gets fully answered in an AI Overview, and the click-through for the purely factual version of that question has genuinely dropped. The fix was not to abandon the page — it was to rebuild it around the actual decision a homeowner in Orange County's climate faces, with specifics an AI summary is less likely to fully capture: how coastal salt air and Santa Ana wind dust load filters faster in some OC neighborhoods than the generic "every 90 days" advice accounts for, paired with a direct path to booking a filter check. The informational half of that page lost some traffic. The decision-support half of it, aimed at someone already close to calling, held up fine — because that is the half an AI Overview answers worst and a real local business answers best.
Structured data and entity clarity matter more, not less
Both traditional search and AI systems lean heavily on being able to confidently identify what a page is about, who wrote it, and how it relates to other known entities. Clean structured data (Organization, Person, Article, LocalBusiness schema), consistent entity naming across your site and citations, and a clear internal linking structure all make it easier for both Google's ranking systems and AI summarization to understand and correctly attribute your content. This is groundwork, not glamour, but it is more valuable now than it was five years ago, not less.
What we track differently now
We added two things to standard monthly reporting that were not there three years ago: a rough log of which target queries now trigger an AI Overview above the organic results (Search Console does not break this out cleanly yet, so this is still partly manual spot-checking), and a periodic check of whether our own client content gets cited when we ask common questions in the tools their customers are likely using. Neither is a precise science yet. Both are more useful for spotting a real trend early than waiting a year to notice traffic to an entire content category quietly eroded.
The new snake oil
"AI SEO" and "GEO" (generative engine optimization) packages are, largely, the "we'll submit your site to 500 directories" offer of this decade — a label slapped on fundamentals that were already best practice, sold at a markup to companies scared of a headline. There is no secret technique for getting cited by an AI Overview that is meaningfully different from writing genuinely good, clearly structured, honestly sourced content. Be skeptical of anyone selling a specific "AI ranking factor" product with confident specificity — the systems change too fast and are too opaque for anyone outside the labs building them to have that level of certainty.
What we are actually changing for clients
- Weighting content strategy more toward transactional, local, and genuinely expert content, and less toward broad informational traffic plays that AI Overviews now intercept.
- Tightening structured data and author/entity signals across existing content, which was previously a nice-to-have and is now closer to table stakes.
- Tracking brand search volume and citation appearances in AI tools alongside traditional rank tracking, because rank position alone is no longer the full picture of visibility.
What we are not changing
We are not abandoning organic content strategy, we are not telling clients rankings do not matter anymore, and we are not selling anyone a GEO package with a straight face. The fundamentals — a fast, well-structured site, content that answers a real question better than the alternatives, and a strong local and brand presence — are the same fundamentals that made a site resilient to every previous Google update, and they are what makes a site resilient to this one too. If you want a plain read on how AI search is actually affecting your specific traffic rather than the industry in general, that is a conversation worth having before you rebuild a strategy around a headline.