Generative engine optimization · Citation ranking
When ChatGPT names one business first, it's a veterinarian's own site 75% of the time. For a dentist, it's 35%.
We've covered before how directory citations shift depending on what someone asks. This time we looked at something narrower: when ChatGPT answers with a single source up front, before any list or comparison, who gets that spot: the business itself, or a directory standing in for it.
We pulled every query from our September 28 run: ten service categories, five cities, twenty questions per category split across comparison, booking, symptom, and procedure phrasing. For each one we isolated whichever source the model cited first and checked whether it belonged to an actual business or to a directory, review site, or ranking blog.
Before counting anything, we did the check we always do before publishing these numbers: pulled up the homepage of every first-position source we hadn't already classified. Thirteen of them turned out to be directories we'd missed, including a Denver listings site, a "ratings and reviews" aggregator for Tulsa, and HomeAdvisor, which somehow wasn't in our list despite being one of the largest contractor directories in the country. Two others looked like directories at a glance, one of them even calls itself "Best Pain Management" in the title, but turned out to be real clinics with search-friendly names, so we left those alone. We fixed the list and recounted before writing a word of this.
After the correction: veterinarian topped the list at 75%, with personal injury lawyer close behind at 70%. Chiropractor came in at 60%, and auto repair, optometrist, and plumber all landed at 55%. Med spa landed right at the midpoint, with half of its first citations going to the business itself. Gym dropped to 45%, hair salon to 40%, and dentist finished last at 35%. Across all ten categories combined, a business's own site held the first position 54% of the time.
What fills the remaining 46% tells its own story. For dentist queries, the first-cited source was just as often yably.com, bestinhood.com, or mybestdenver.com as it was an actual dental practice. For hair salon, directories like chicagoka.com and besthairguider.com, along with scheduling aggregators like Fresha, kept showing up ahead of real salons. None of those are obscure names either, they're exactly the kind of SEO-built listing sites that rank well on Google and apparently get pulled into AI answers just as easily.
Veterinarian sat at the opposite end for a reason that became clear once we looked at what actually got cited. The questions that pushed veterinarian this high weren't the generic "best vet in Chicago" type. They were things like "my cat has been vomiting for two days, where do I take her" and "emergency vet open now." Those got answered with MedVet, Humane Colorado, CatsOnly.vet, and similar clinics, each cited from its own page describing exactly that situation. A directory can't answer something that specific without already knowing which clinics handle emergencies, so the model skipped it in favor of the clinic itself.
Personal injury lawyer followed the same logic. Most of the queries in that category ask about a specific kind of accident or a specific legal situation rather than a general ranking, and law firm sites tend to build out pages for exactly those situations, because that's how their own SEO works. That gave the model something concrete to cite instead of a directory's generic listing.
Dentist and hair salon sit lower for what looks like the mirror-image reason. A lot of the questions in both categories are closer to the comparison type, best dentist in a city, best hair salon for a certain service, and that's the exact shape of question a ranking site is built to answer. A different cut of this same data showed the same pattern before: comparison-style questions favor directories, specific-situation questions favor the business. What's new is that the pattern holds even when four question types get collapsed into one first-citation score per industry, instead of comparing two question types head to head inside a single category.
We should be clear about what this does and doesn't show. This is one day of testing across five cities, twenty queries per category. Enough to see a pattern but not enough to treat 75% or 35% as a fixed number for any of these industries going forward. A different week, a different model update, or a different set of cities would likely move these exact percentages, even if the overall shape held. We're also only measuring which source got cited first, and we can't say whether the person asking actually clicked through or called anyone.
If your category sits closer to the dentist end of this list, the fix isn't complicated. It just takes building the specific page instead of leaning on the general one. A directory can answer "best dentist in Denver" because that's a ranking question. It can't answer "same-day appointment for a cracked tooth" nearly as well, because that requires a page that actually says so. The businesses that won the first-citation spot in categories that otherwise lean toward directories almost all had a page built around the specific situation someone was asking about, rather than a homepage listing services in general terms.
Source: askchat.studio's GEO probe database, first-citation analysis across ten service categories (veterinarian, personal injury lawyer, chiropractor, auto repair, optometrist, plumber, med spa, gym, hair salon, dentist), five U.S. cities, twenty queries per category spanning comparison, booking, symptom, and procedure intent, collected September 28, 2026, through the OpenAI Responses API with web search grounding.