We measured every med spa in East Cobb.
Half are invisible to AI search.

Is blocking the AI crawlers why these med spas are invisible?

No. Every one of the 20 already allows the crawlers that actually fetch pages to cite in answers. The “add the AI bots to your robots.txt” advice is aimed at the wrong bots. GPTBot and ClaudeBot are training crawlers. OpenAI’s own docs say GPTBot crawls “content that may be used in training,” while the bot that surfaces you in ChatGPT search is a different one, OAI-SearchBot. Blocking the training crawler does not remove you from the answer.

The same holds on Google’s side. Google’s crawler documentation states plainly that Google-Extended does not affect inclusion or ranking in Search, AI Overviews included. That runs on Googlebot. So the robots.txt checklist everyone starts with was already complete across the entire market, and it explains none of the invisibility in Finding 01.

Exactly one business in the index, DESSNA, blocks anything at all, and it blocks the right things. Its robots.txt disallows the training and bulk-scraper crawlers (GPTBot, ClaudeBot, CCBot, Bytespider) while explicitly allowing every answer-time citation bot: OAI-SearchBot, ChatGPT-User, Claude-User, PerplexityBot, Google-Extended, Googlebot. That is the most deliberate configuration in the set, not a mistake. Robots.txt is not the lever here. It is already pulled. It was not the lever for Georgia’s Fortune 500 headquarters either: none of the 13 robots.txt answers we could read blocks an AI crawler.

What exactly counted as an East Cobb med spa?

East Cobb is not a city. It is unincorporated Cobb County, the roughly 200,000 residents who hold a Marietta mailing address but live outside the city limits, across zips 30062, 30066, 30067, 30068, and the East Cobb part of 30075. A business is in scope only if its physical street address sits in one of those zips. We included clinics marketing medical-aesthetic or injectable services (Botox, fillers, laser, microneedling), which brings in a few cosmetic-dermatology practices (Goodman, Dermatology Affiliates, DESSNA) alongside the standalone med spas. For multi-office brands the row reflects the in-scope office and the brand’s shared website.

What was measured, and what was deliberately not.

Two things, both reproducible. For each business we ran the four fixed queries through a web-answer surface and counted whether its own domain appeared in that query’s returned sources; directory pages that surfaced (Yelp, med-spa directories, the chamber listing) were not counted as a business’s own citation. And we fetched each site’s /robots.txt, treating an AI crawler as blocked only when a matching user-agent group carried Disallow: /. The ranking is on citations alone, because that is the signal we could measure the same way for all 20.

The limits, plainly. This is one answer surface on one day, and the search engine pool behind it ran degraded during the session, so the census is a defensible floor of 20 confirmed med spas rather than a proven ceiling, and a zero citation count means “not surfaced by these four queries on this surface today,” not “never cited.” We deliberately did not publish a per-site structured-data score this edition: several of these sites render their schema with JavaScript, which a plain fetch cannot see, so scoring markup reliably needs a rendering crawl. That audit, plus directory and review signals, is what the next edition adds. Everything here is reproducible as published: the same four queries, the same robots check, run against the same 20 domains.