Ask an AI for the best med spa in Austin and it mostly names directories, not clinics. When I ran that query in July 2026, the answer leaned on Yelp, DiscoverMedSpa, MedicalSpaLocator and, at one point, Hotels.com. A handful of practices got named directly. Everyone else was represented by a third party that sells their competitors advertising.
That is a strange thing to discover about your own business, and most owners have never checked. It takes about ninety seconds.
Run the query yourself first
Open ChatGPT, Perplexity or Google's AI mode. Log out, or use a private window, so your own history does not flatter the result. Ask the question a patient would actually ask:
- "Best med spa in Austin for Botox and fillers"
- "Where should I go for lip filler in Austin"
- "Most trusted medical spa near downtown Austin"
Write down two things: whether you were named, and what got named instead. The second is the more useful answer.
Why the aggregators win
Answer engines do not rank pages the way a search results list does. They assemble an answer from sources they consider safe to repeat, and a directory page listing twelve clinics with ratings is a much easier thing to summarise than twelve competing clinic homepages that each claim to be the best.
Directories are built for that job. They publish structured, comparable, frequently-updated data: name, address, hours, services, price band, review count, rating. A clinic site usually publishes a hero image, a booking button, and a paragraph about the founder's philosophy. One of those is machine-readable. The other is not.
The clinics that do surface tend to surface on review volume and directory presence, not on anything they control on their own site.
That is the part worth sitting with. If you are visible in AI answers today, it is probably because a third-party platform decided you were, and that platform can change its mind, reorder its list, or sell the slot above you tomorrow.
Ratings are not the differentiator anyone thinks they are
Austin has roughly sixty-eight medical spas. Their average Google rating sits near 4.7 stars across about fourteen hundred reviews in total. When the whole category is clustered between 4.6 and 4.9, a rating stops being information. It cannot separate you, because it does not separate anyone.
I spent four years running brand and PR at Curaleaf and now run marketing for a five-location cannabis operator in Connecticut. Regulated categories hit this wall early, because paid advertising is mostly closed and everyone ends up competing on the same handful of trust signals. The lesson transfers cleanly: when the obvious signal saturates, the advantage moves to the signals nobody is maintaining.
What actually moves the needle
1. Publish the answer, not the brochure
Answer engines cite pages that answer a question completely in one place. Most clinic sites have a services page listing "Botox, fillers, microneedling" as a menu. Almost none have a page that answers "how much does Botox cost in Austin and what changes the price," in plain language, with real ranges and the honest caveats.
Write the page you would be slightly embarrassed to publish because it gives away the consultation. That is the one that gets cited.
2. Make your facts machine-readable
Add LocalBusiness and MedicalBusiness schema with your real address, hours, practitioners and services. Add FAQPage schema to the answer pages above. This is a one-afternoon job for a developer and it is the single cheapest thing on this list.
3. Fix the listings you are not looking at
Google Business Profile is the one everyone maintains. Apple Business Connect, Bing Places and Foursquare are the ones that quietly go stale, and Foursquare in particular feeds a surprising amount of downstream data. Inconsistent hours or a wrong suite number across those sources is a reason for a model to trust a directory over you.
4. Get named somewhere a model already trusts
Being quoted in a local publication, an industry roundup, or a genuinely useful thread does more for citation than another page on your own site. You are not buying links. You are giving the model a second, independent source that says you exist and are good at a specific thing.
A reasonable ninety days
- Weeks 1 to 2. Run the queries, log what came back, screenshot it. Audit your listings across all four platforms.
- Weeks 3 to 6. Write three answer pages for your highest-intent treatments. Add the schema.
- Weeks 7 to 12. Earn two or three independent mentions. Re-run the same queries and compare against the screenshots.
Re-running the identical query on a set date is the measurement. There is no dashboard for this yet, which is exactly why so few practices are doing it.
The honest caveat
Answers vary between models, between sessions, and between phrasings of the same question. A single query is an anecdote. Five queries, logged on the same date each month, is a trend line. Treat it that way and you will avoid both the panic and the false comfort.
Check what the machines say about you before you spend another dollar on ads that point at a website they are not reading.