Who Does AI Recommend for Financial Advisors and Lawyers in Texas?

We asked ChatGPT and Google AI Mode 20 questions about financial advisors, wealth managers, and lawyers in Houston, Dallas, San Antonio, and Austin, then asked every question again. The lists moved. Asked the same question a few minutes later, ChatGPT named fewer than half of the same firms (38 of 82, or 46%) and kept the same first firm in 7 of 20 questions. The two engines mostly named different firms. Both cited Google more than any other source, and past Google they split: ChatGPT leaned on NerdWallet, Avvo, and Expertise.com, and Google AI Mode on Super Lawyers and Thumbtack.

Key findings

• Asked twice, the lists changed. ChatGPT, asked again 7 to 19 minutes later, named 38 of the 82 firms from its first answers again (46%) and kept the same first firm in 7 of 20 questions. Google AI Mode, asked again 26 to 27 hours later, named 43 of 75 again (57%) and kept the same first firm in 4 of 16.
• The two engines rarely agreed. Counting each firm once per question, ChatGPT named 127 firms across the 20 questions, and 28 of them (22%) were also named by Google AI Mode for the same question. In 5 questions the engines had no firm in common, though in 3 of those Google AI Mode answered only once.
• Only 5 firms were named in all four answers to their question, both asks on both engines. That happened in 5 of the 16 questions where all four answers came back.
• Google came first on both engines, and past Google each had its own sources. ChatGPT cited a Google page in 19 of its 40 answers, NerdWallet in 11 of its 16 advisor and wealth answers, and Avvo in 14 of its 24 lawyer answers. Google AI Mode cited Google Business Profiles in 22 of its 36 answers, Super Lawyers in 9 of its 21 lawyer answers, and NerdWallet in none.
• A firm's own website rarely carried the answer. Of the 339 times a firm was named, its website was shown 221 times (65%), and its own site was a cited source 17 times (5%).
• Advisors were named with few reviews. The 28 advisory and wealth firms Google AI Mode showed with a review count had a median of 24 Google reviews, and 11 had fewer than 20. The 40 law firms it showed had a median of 238.
• Every one of the 76 answers was recorded with a caveat, such as that the right choice depends on your goals or your case.

Ask twice, get a different list

We asked each question twice on each engine. ChatGPT's second ask came 7 to 19 minutes after the first, in a fresh chat. Google AI Mode's came the next day, 26 to 27 hours later.

ChatGPT kept fewer than half its firms. Of the 82 firms in its first answers, 38 came back in the second (46%). It named the same first firm both times in 7 of 20 questions, and it gave the same list twice once.

Google AI Mode held a little more. Of the 75 firms in its first answers, 43 came back the next day (57%). It named the same first firm in 4 of 16 questions and never gave the same list twice. 4 of its 40 asks returned an error instead of an answer, which left 16 questions with two AI Mode answers to compare.

So one answer is a snapshot. Whether a firm gets named is a rate across many asks, not a yes or no from a single one.

ChatGPT and Google AI Mode named different firms

Across both asks, and counting each firm once per question, ChatGPT named 127 firms for the 20 questions. Google AI Mode named 28 of the same firms for the same question (22%). In 5 questions the two engines had no firm in common: estate planning and family law in Houston, family law and personal injury in San Antonio, and financial advisors in Dallas. In 3 of those (both Houston questions and San Antonio personal injury), Google AI Mode's first ask failed, so only one of its answers was compared.

Their first picks lined up in 4 of 20 questions, where a firm one engine named first was also named first by the other in one of its asks.

Five firms were named in all four answers to their question:

• Financial Synergies Wealth Advisors, for financial advisors in Houston
• Financial Life Advisors, for financial advisors in San Antonio
• 1900 Wealth, for wealth management firms in San Antonio
• Abraham, Watkins, Nichols, Agosto, Aziz & Stogner, for personal injury lawyers in Houston
• Goranson Bain Ausley, for family law attorneys in Dallas

ChatGPT named 4.1 firms per answer on average (3 to 5). Google AI Mode named 4.8 (2 to 8).

What each engine cited

ChatGPT showed its evidence as citation pills, and Google AI Mode as a source panel and inline chips. Google was the most-cited source on both: ChatGPT cited a Google page in 19 of its 40 answers, and Google AI Mode cited Google Business Profiles in 22 of its 36. Past Google, the two leaned on different sites, and the sites changed with the profession.

For advisors and wealth managers, ChatGPT cited NerdWallet in 11 of 16 answers and a Google page in 9. Google AI Mode cited Google Business Profiles in 13 of 15.

For lawyers, ChatGPT cited Avvo in 14 of 24 answers, Expertise.com in 11, a Google page in 10, bestlawyers.com in 6, and Yelp and Doyle's Guide in 5 each. Google AI Mode cited Google Business Profiles and Super Lawyers in 9 of 21 each, and Thumbtack in 5.

What the named firms had in common

Most had a website the answer showed. Of the 339 times a firm was named, its website was shown 221 times (65%). The website was rarely the evidence: the firm's own site was a cited source 17 times (5%), 13 in ChatGPT and 4 in Google AI Mode.

In Google AI Mode, the Business Profile was the evidence. It cited Google Business Profiles in 22 of its 36 answers, and in 13 of its 15 answers about advisors and wealth managers.

Advisory firms did not need many reviews. Google AI Mode showed a review count for 28 different advisory and wealth firms. The median was 24 reviews, 11 had fewer than 20, and the smallest count shown was 1. For the 40 law firms it showed with a count, the median was 238.

Ratings were high but not perfect. Of the 133 ratings the answers displayed, 4 were below 4.0, and the lowest was 3.0. We did not verify ratings or review counts.

ChatGPT often named lawyers instead of firms. Of the 99 names it gave in its lawyer answers, 44 were individual attorneys. In Google AI Mode it was 20 of 101.

Most of the firms named were local. In advisor and wealth answers, 20 of 147 firm mentions (14%) went to national firms, such as Fisher Investments, Creative Planning, Savant Wealth Management, and Morgan Stanley.

Who was named, city by city

These are the firms each engine named first, in the order each answer gave. This is not our ranking, and two asks are not a stable ranking.

One firm in these tables, Post Oak Private Wealth Advisors in Houston, is a 6AI client. Its results were recorded and counted the same way as every other firm's.

Other things we noticed

Google AI Mode failed 4 times in 40. Instead of an answer, it showed “Something went wrong and an AI response wasn't generated.” All 4 were first asks: family law and estate planning in Houston, personal injury in San Antonio, and financial advisors in Austin. Every second ask, the next day, returned an answer.

A few firms were named in more than one city. Arnold & Itkin, a Houston personal injury firm, was named for Houston, Dallas, and San Antonio. Goranson Bain Ausley and Walters Gilbreath were each named for family law in three of the four cities.

Google AI Mode used an old name. Both of its Austin wealth management answers named Durbin Bennett Private Wealth Management, which has done business as Maslow Wealth Advisors since February 2024, according to SmartAsset. ChatGPT used the current name.

One Google AI Mode card mixed up two firms. In its second Austin personal injury answer, a card named “FVF Law” showed 41 reviews, an address on South I-35, and a Website button to feltoonlaw.com, a different firm's website. The first answer's FVF Law card showed 343 reviews, a Burnet Road address, and fvflawfirm.com. We did not count the second card as FVF Law Firm.

Some of ChatGPT's Google citations pointed to Google sites for other countries. In 9 of its 40 answers, a citation pill showed a domain such as google.co.in (India), google.com.sg (Singapore), or google.com.br (Brazil).

How we ran it

We asked one question per category and city, worded exactly: “Who is the best [category] in [city], Texas?” The five categories were financial advisor, wealth management firm, personal injury lawyer, family law attorney, and estate planning attorney. The four cities were Houston, Dallas, San Antonio, and Austin, which makes 20 questions.

We asked two engines, ChatGPT (chatgpt.com) and Google AI Mode, both logged out in a clean browser with no account and no history. ChatGPT reported the model gpt-6-mini for all 40 of its answers and ran its web search on its own. Google AI Mode was set to each question's city with Google's location setting. ChatGPT has no location setting, so its location came from the browser's network, in Austin.

Each question ran twice on each engine, 80 asks in all, and every ChatGPT ask started a fresh chat. ChatGPT's asks ran on October 23, 2026, between 11:13 and 11:37 a.m. Central. Google AI Mode's first asks ran on October 23 between 11:12 and 11:40 a.m., and its second asks on October 24 between 1:16 and 2:43 p.m. Central. Google asks ran one at a time. 76 asks returned an answer: all 40 in ChatGPT and 36 of 40 in Google AI Mode. The 4 errors were not retried.

For each answer we recorded the firms it put forward as options, in order, with any website, rating, and review count it showed, and the sources it cited. Directories and publications given as sources, such as NerdWallet, Avvo, and Super Lawyers, were recorded as sources, not as firms. “Cited” means the evidence the answer showed: ChatGPT's citation pills, and Google AI Mode's source panel and inline chips.

We matched names that refer to the same firm, such as an attorney and the attorney's firm, or “Abraham Watkins” and “Abraham, Watkins, Nichols, Agosto, Aziz & Stogner,” using each firm's website, and counted each firm once per answer. Two entries were corrected by hand: the mixed-up FVF Law card described above, and one garbled name in a Google AI Mode answer, which we left out.

We planned a fifth city, Fort Worth. Google blocked our checks before its Google AI Mode questions ran, so Fort Worth is left out of every count here.

Limitations:

• Two asks per engine per question. More asks would give steadier rates, and any single result here can change.
• Uneven spacing. ChatGPT's second asks came minutes after the first, and Google AI Mode's about a day later, so the Google AI Mode comparison mixes repeat-ask change with a day of change.
• Two engines, logged out. Signed-in users and other assistants, such as Gemini, Perplexity, Copilot, and Claude, may get different answers.
• One location for ChatGPT. It ran from Austin for every city, while Google AI Mode was set to each city.
• What the page showed. Ratings and review counts are what the answers displayed, and we did not verify them. Google's source panel can hide sources behind “Show all,” which we did not open.
• Judgment calls. Whether a name is a firm or a source, a person or a firm, and local or national was judged from the answer and the firm's website, and a few calls may be wrong.

What it means for your firm

These are patterns from 76 answers, not rules. They point the same way as our Austin study: AI names the firms that third-party profiles vouch for, and a firm's own website rarely does the convincing alone.

• Your Google Business Profile is the evidence in Google AI Mode. It was cited in 13 of 15 advisor and wealth answers.
• For lawyers, ChatGPT cited Avvo most (14 of 24 answers), then Expertise.com and Google. Google AI Mode cited Google Business Profiles and Super Lawyers equally (9 of 21 each), then Thumbtack.
• Advisors don't need hundreds of reviews to be named: the median advisory firm Google AI Mode showed had 24. For an SEC-registered adviser, reviews and ratings fall under the Marketing Rule, so they go through compliance first.
• Being named on one engine says little about the other. Only 22% of ChatGPT's firms were also named by Google AI Mode for the same question.
• One answer is a snapshot. Whether you are named is a rate over many asks, and it changes between them.

Getting onto those sources, and keeping them consistent, is the work. About every 72 hours, 6AI Analytics asks Google, ChatGPT, Google AI Mode, and the other assistants the questions your clients ask, and records who is named and which sources each answer cites. The SEO, GEO, and AEO pros who built it make the changes. For advisors, every change goes through your CCO first (how that works), and for law firms every page is written to the Texas advertising rules (how that works). To see where your firm stands, get a free AI visibility check.

The data file, with every firm named and the sources for each answer, is available on request. Book a demo and we'll ask your clients' questions live.

Questions about the study