Texas Dining AI Index 2026: which restaurants ChatGPT, Gemini and Google AI name on their own

Last updated: 5 September 2026

There is public data on how invisible American restaurants are inside AI answers, but almost none of it says which restaurants actually get named, and none of it covers Texas at the level a working operator could use. This index closes that gap. On 5 September 2026 we put 200 real diner questions to ChatGPT, Gemini and Google AI Overview across Austin, Dallas, Houston and San Antonio, and recorded 1,622 answers. We tracked 67 Texas restaurant brands, and 62 of them came up at least once. Every brand was scored against that same 1,622-answer denominator. The finding worth remembering is simple: the most-named restaurant in Texas still fails to appear in about sixteen of every seventeen answers.

The numbers in brief

• Terry Black's Barbecue is named most often, in 5.98 percent of the 1,622 answers. That is roughly one answer in seventeen.
• The median across the 62 restaurants that appeared at least once is 1.17 percent.
• Outside the published top fifteen, no restaurant clears 1.91 percent.
• The next names in order after the leader are Franklin Barbecue on 4.69 percent, Ninfa's on Navigation on 4.56 percent and Pappas Bros. Steakhouse on 3.64 percent.
• Eater is opened in 20.04 percent of answers, more than any booking platform, review site or city tourism board.

What we measured, and how

We track 200 questions phrased the way a diner actually types them. They cover eight themes: everyday occasions, barbecue, Tex-Mex and Mexican food, vegetarian and dietary needs, visitors and tourists, groups and celebration bookings, business dinners, and special occasions such as anniversaries. Each question names a Texas city or a landmark inside one. That matters more than it sounds. Without a place anchor, a model will happily answer a restaurant question for a completely different part of the country, and the measurement then measures nothing.

The figures in this report are drawn from 1,622 answers recorded on 5 September 2026 across ChatGPT, Gemini and Google AI Overview. We record what the engine returns, not what it was asked to return.

Three measures, in plain words:

• How often AI names them: the share of the 1,622 answers in which the restaurant is named.
• Share of the answers: the restaurant's portion of all 2,464 brand mentions across every answer.
• Average position: where the restaurant tends to land in a list when it is named. Lower is better.

Every restaurant was measured against the same 1,622-answer denominator. No restaurant paid to be included, and the names in this report identify research subjects, not customers and not endorsements.

Who the engines actually name

The table above is the visible top of a very flat field. Across all 62 restaurants that appeared at least once, the median is 1.17 percent. Outside the published fifteen, no restaurant reaches even 2 percent: the ceiling for everyone below the cut is 1.91 percent, held by The Pit Room, followed by El Tiempo Cantina on 1.85 percent and LeRoy and Lewis on 1.73 percent. Five of the 67 brands we track did not appear in a single answer.

The finding that matters more than the ranking

The stronger conclusion is not that Terry Black's leads. It is that leading means almost nothing here. The most-named restaurant in the whole state is absent from 94 percent of the answers we recorded. Second place trails first by roughly one and a quarter percentage points, and by the eleventh name the gap between one position and the next is a rounding error. In a category this flat, no restaurant owns the answer, which means the recommendation is still there to be won rather than defended.

There is a second surprise sitting inside the table. Texas is known worldwide for barbecue, yet barbecue takes only three of the top ten places. Steakhouses take three, and long-standing Tex-Mex institutions take three. The engines are not simply repeating the state's most famous export. They are answering the question that was asked, and most diner questions are about an occasion rather than a cuisine.

Being named often and being named first are different things

Two restaurants can sit next to each other in the table and be doing completely different jobs inside the answer.

Terry Black's is named in more answers than anyone else, but when it is named it lands at an average position of 2.3 and is mentioned 1.32 times per answer it appears in. Franklin Barbecue is named in fewer answers, 76 against 97, yet it holds the best average position of any restaurant near the top of the table at 1.7, and Truth BBQ matches it. In plain terms, Terry Black's makes the most shortlists and Franklin is the name the engines put at the top of the shortlist it makes.

The mention counts expose a second mechanic that matters a great deal to any operator running more than one room. Pappas Bros. Steakhouse is mentioned 2.08 times in every answer it appears in, Bohanan's 2.14 times, and El Fenix 2.40 times. Terry Black's manages 1.32. The difference is not reputation. It is rooms. A brand with several locations gets named once for the downtown answer and again for the neighbourhood answer, sometimes inside the same reply, so it takes up more space in an answer it appears in even when it appears in fewer answers overall. You make more shortlists, they take up more room on the ones they are on.

That distinction changes what you do about it. If you appear often but low, your problem is the quality and recency of the evidence that ranks you. If you appear rarely but high, your problem is coverage, and the fix is getting into more of the source pages the engines already read.

Where the answers come from

Three things stand out.

First, editorial beats everything. Eater and The Infatuation together are opened far more often than any booking platform. The shortlist a Texas diner receives from an AI engine is, to a large degree, a shortlist that food editors wrote first.

Second, community sources are not a fringe input. Reddit alone is opened in 12.89 percent of answers, ahead of Yelp and ahead of Tripadvisor. What people say about a restaurant in a public thread is part of the evidence the engine reads.

Third, the city tourism boards matter more than most operators assume. Visit San Antonio, Visit Dallas and Austin Texas are each opened in more than 4 percent of answers, which puts a free, claimable listing on a par with paid platforms. If your restaurant is missing or out of date on your own city's tourism site, fix that this week before you spend a cent anywhere else.

What this report can and cannot tell you

These results are directional, not definitive. They describe one question set, on three engines, on one day.

Read this way, the data can answer a bounded set of questions:

• Which Texas restaurants the engines name without being prompted with a brand name.
• How concentrated, or in this case how flat, the field is across a large question set.
• Which source types the engines reach for when they build a Texas dining shortlist.
• How the four largest Texas metros compare in how settled their answers are.

It cannot do several other things, and it is worth being blunt about them. It cannot prove that appearing more often produces more covers or more revenue. It cannot establish a causal link between any single change a restaurant made and its position here. It cannot tell any individual operator that they ought to appear in every one of the 200 questions, because most restaurants have no business appearing in most of them. A steakhouse has no claim on a vegan tasting menu question.

The operating instruction is therefore not to chase the leaderboard as one blended score. Split the questions by the ones you could genuinely win, read the exact answers the engines give for those, and repair the weakest layer of evidence underneath them.

What this means if you run a restaurant in Texas

If you are a single-site operator, the flatness of this field is the opportunity. A category where the leader sits under 6 percent has no incumbent to displace. The restaurants at the top of this table are not there because of a clever content strategy. They are there because decades of editorial coverage, community conversation and city listings all point at the same name. That evidence base is what an engine reads, and it is buildable. Start by making sure the sources already in the top of the retrieval table can actually see you: your city tourism listing, your booking platform profile, and the editorial outlets that cover your metro. Our guide on how to get a restaurant recommended by ChatGPT covers the order to do that in.

If you run several sites, the more useful read is per venue rather than per brand. Engines recommend places, not companies. A group with five rooms has five chances to be named in a neighbourhood answer, and in practice the newest or least documented room is usually the one dropping out. Check each location as its own entity, with its own page, its own structured data and its own claimed listings. We wrote up that pattern in how multi-site restaurant groups appear in AI search.

If you are comparing markets, the source table is a research queue, not a mass-pitch list. The point is not to email every publication on it. The point is to find the one or two where you have a genuine reason to be covered, a new room, a real change of chef, a dish with a story an editor can check, and to approach those. A pitch that says "please add us to your list" is ignored. A pitch that hands an editor a verifiable fact they did not have gets read. The same pattern held in our UK restaurant index, where editorial retrieval outweighed every other source type by a wide margin.

Sources

• Schmitdy category tracking, Texas dining question set, 5 September 2026. The 1,622 answers, 200 questions and 67 tracked brands behind every figure in this report.
• Terry Black's Barbecue, terryblacksbbq.com. Official site of the most-named brand in the index.
• Franklin Barbecue, franklinbbq.com. Official site, second in the published table.
• Ninfa's on Navigation, ninfas.com. Official site, third in the published table.
• Pappas Bros. Steakhouse, pappasbros.com. Official site, fourth in the published table.
• Al Biernat's, albiernats.com. Official site, fifth in the published table.
• Bohanan's, bohanans.com. Official site, sixth in the published table.
• Mi Tierra Café y Panadería, mitierracafe.com. Official site, seventh in the published table.
• Truth BBQ, truthbbq.com. Official site, eighth in the published table.
• Uchi, uchirestaurants.com. Official site, ninth in the published table.
• Matt's El Rancho, mattselrancho.com. Official site, tenth in the published table.
• La Fonda on Main, lafondaonmain.com. Official site, eleventh in the published table.
• Perry's Steakhouse & Grille, perryssteakhouse.com. Official site, twelfth in the published table.
• 2M Smokehouse, 2msmokehouse.com. Official site, thirteenth in the published table.
• Pecan Lodge, pecanlodge.com. Official site, fourteenth in the published table.
• InterStellar BBQ, theinterstellarbbq.com. Official site, fifteenth in the published table.
• Eater, eater.com. The most-retrieved source across the tracked answers.
• The Infatuation, theinfatuation.com. Second most-retrieved source.
• OpenTable, opentable.com. Most-retrieved booking source.
• Reddit, reddit.com. Most-retrieved community source.

Every brand named in this report is a research subject. Inclusion is not an endorsement, and no restaurant paid to appear.

If you want to know which of these questions your own restaurant already shows up in, we can run the same measurement for your market.