Case 1: The med spa — the accidental discovery
The first test wasn’t a test; it was family. My wife runs a med spa, and I rebuilt its online presence the way the playbook now describes: made every listing agree with every other listing, rewrote the site so each treatment had a plain-language page a machine could quote, added structured data, and built a steady review habit where clients mentioned the specific treatment they’d had.
No new funnel. No ad spend. For weeks, nothing — and then new clients started mentioning, unprompted, that ChatGPT had recommended the spa when they asked where to go. Not as a one-time novelty: it became a daily, dependable source of new appointments. The leads arrived pre-sold — they weren’t comparison shopping, because the comparison had already happened inside the assistant.
What made the difference
- Treatment pages an assistant could quote (“what does a hydrafacial cost here?” had a real answer).
- Reviews that named treatments, teaching assistants what the spa is known for.
- Total consistency of the boring facts everywhere the spa appeared.
Case 2: The detailing company — ruling out the fluke
One success proves nothing; maybe the spa got lucky. So I ran the identical process on a completely different business — a car detailing company. Different customers, different price point, different search behavior, zero overlap with the first test.
Same sequence: the tedious consistency work, quotable service pages with honest pricing ranges, structured data, deliberate review-building. Same quiet period. Same outcome — the phone started producing customers who’d asked an AI assistant who does the best detailing work in the area and been handed the company’s name.
Two-for-two told me the mechanism was real. But both were local service businesses, so one objection remained.
Case 3: The marketing agency — the acid test
If AI visibility only worked for “best X near me” local queries, it would still be valuable — but limited. So the third test was the least local, most skeptical-friendly candidate I could pick: a marketing agency. An agency’s whole trade is generating demand; if anyone should be able to out-market a free channel, it’s them.
The process worked anyway — and then some. Today that agency’s number one source of new business isn’t its ad spend. It’s free leads from AI assistants recommending it to people who asked what agency they should hire. Sit with the irony: a business that sells lead generation gets its own best leads from being the answer, not from buying attention.
What generalizes
- The mechanism is market-agnostic. Assistants answer “who should I use?” the same way in every category: read, reconcile, recommend the legible few. Details here.
- The quiet period is real. Weeks passed before each phone started ringing. Signals compound; nothing here is overnight. Distrust anyone who promises otherwise.
- Pre-sold beats nurtured. Leads that arrive on an assistant’s recommendation close differently. The trust was borrowed from the assistant, and it transfers.
- The work is unglamorous and that’s the moat. Nothing we did was clever. It was consistent, machine-legible honesty, maintained over months — which is exactly why competitors don’t copy it.
This repeating pattern is what convinced us to walk away from the funnel-first playbook entirely.