Every few weeks a firm owner tells me some version of the same story: the Google traffic chart looks worse than last year, but the calls coming in are somehow better qualified. That isn't a paradox. It's what it looks like when a slice of your buyers stops clicking links and starts asking ChatGPT instead.
The numbers behind that shift are hard to ignore. Seer Interactive looked at their client data and found ChatGPT referrals converting at 15.9 percent, against 1.76 percent for Google organic. Ahrefs saw AI-sourced visitors making up half a percent of traffic but 12.1 percent of signups. Fewer visits, dramatically better ones. So the question I keep getting, "do I still need SEO if everyone's using ChatGPT," deserves an actual answer rather than a slogan.
How an AI answer gets assembled
When someone asks an assistant for "a fee-only financial advisor in Raleigh," the model does two things, roughly. It leans on what it absorbed during training, years of the public web where certain firms kept turning up in reviews, directories, local press, and everything else it read. And increasingly it runs a live web search behind the scenes and synthesizes the top results into one confident paragraph, sometimes with citations attached.
Notice what sits inside that second step: a search, often literally against Google's or Bing's index. The AI didn't replace the search engine. It sat down on top of it and started eating the clicks. Google's own AI Overviews now appear above the traditional results on a large share of searches, which means even people who never open ChatGPT are getting the assembled-answer experience whether they asked for it or not.
Your old SEO work is the raw material
This is the part that surprises owners who were braced to throw everything out. Nearly everything classic SEO rewarded still matters, because it's exactly what the models ingest: crawlable pages with clear service language, a maintained Google Business Profile (still the heaviest input for local firms, and I wrote about that side of things in /blog/google-maps-local-seo-law-firms/), reviews that double as training data, and links from real sources.
Think of it this way. If you rank third for "estate planning attorney Franklin TN," you're in the pool of pages the AI reads before it answers. If you rank forty-third, you were invisible before and you're invisible now, just one layer further removed from the buyer.
Where the two genuinely diverge
Citations, not positions
In classic SEO, position three was a perfectly good business. Ten blue links meant ten winners on every page. An AI answer names a couple of firms and moves on, so visibility becomes winner-take-most and the consolation prizes largely disappear. That changes the economics of being "pretty good" at this. For a firm that's spent years comfortably holding position four in its market, the map hasn't shifted so much as the prize structure has, and the difference only shows up when you look at where new clients say they found you.
Being the answer beats being the link
Rankings rewarded pages that attracted clicks, so for fifteen years we all wrote long pages engineered for dwell time. Models reward content they can lift a clean answer from: a page that states your fees plainly, or a FAQ that actually answers "what does a revocable trust cost in Texas" instead of dancing around it for eight paragraphs. There's a small irony here. The direct, unclever page your compliance officer would love is now the one machines prefer.
What others say about you outweighs what you say
A model deciding whether to name your firm weighs the public record at least as heavily as your own site. Review volume and recency, mentions in local press, directory consistency, even how often your name appears near your practice area in text it trusts. In classic SEO you could partly ignore all that if your on-page work was strong. You can't anymore. I walked through the full fix-it order, and what to do first, in /blog/get-recommended-by-chatgpt/.
What to actually do differently this year
- Keep the classic engine running. Rankings feed the answers, so cancelling your local SEO to "pivot to AI" is like tearing out the plumbing to improve the water.
- Restructure your key pages to be liftable. Plain answers near the top, real prices where your regulator allows it, FAQ sections with schema markup, and headings phrased the way clients ask.
- Shift content budget from volume to citability. Five pages worth quoting beat fifty generic posts, and a single mention in a regional publication beats both.
- Put reviews on a system instead of a memory. They're now training data, not just social proof, and steady velocity matters as much as the total.
- Change how you measure. Ask every new client how they found you, and actually check what the assistants say about your market monthly, because none of it shows up in Search Console.
The uncomfortable timing part
The firms being named by assistants today mostly built their signals over the past several years without knowing this was coming. The window where you can catch up cheaply is the window where your competitors haven't noticed yet, and in most local professional markets that window is still open. I wouldn't count on that lasting through next year.
If you want the unvarnished starting point, our free AI Visibility Audit at https://scalewithquail.com/ai-visibility/ shows what the assistants currently say when someone asks for a firm like yours in your city. Most owners are surprised by it. Not always pleasantly, but that's rather the point of looking.
Questions people ask
Do I still need SEO if people use ChatGPT instead of Google?
Yes, arguably more than before. AI assistants build answers from search indexes and from the same public signals SEO has always shaped, so a firm that stops ranking also stops getting cited. The channel changed shape, but the inputs didn't.
What is generative engine optimization (GEO)?
It's the work of making your firm something AI systems can read and confidently recommend. In practice that means crawlable pages, consistent business data, a strong review base, and mentions from sources the models treat as trustworthy.
How is being cited by AI different from ranking on Google?
A results page has ten winners, while an AI answer usually names a couple of firms, so visibility is closer to winner-take-most. The traffic is smaller but far better, with Semrush measuring AI search visitors converting at about 4.4 times the rate of organic.