GEO for Accounting Firms: How CPA Firms Get Recommended by ChatGPT, Gemini, Claude and Perplexity

GEO for accounting firms is the work of making sure ChatGPT, Gemini, Claude and Perplexity can find, verify and name your firm when a business owner asks who should handle their taxes, their books or their finance function. In our testing, the firms that get named tend to be the ones whose licenses, specialties and reviews are stated plainly on their own site and match what public registries already say about them.

Key takeaway: AI engines are already sending new business to some CPA firms. The firms that benefit are the ones whose credentials and specialties an engine can read, cross-check and repeat with confidence. Measure where you are named today, fix the proof layer, then measure again on the same questions.

Why GEO for accounting firms matters now

Accounting Today, October 2025 reported that CPA firms across the country were seeing more new business from people who found them through conversations with ChatGPT, Gemini, Claude and other models. The figures in that piece come from interviews with firm leaders, not from a survey. The head of one firm said AI-sourced leads made up 15 to 20 percent of his leads, mostly from people with digital asset questions. The marketing chief at another firm said the conversion rate from AI traffic was three times the site average, with ChatGPT the leading referrer by a wide margin. The firms also described these prospects as arriving with specific needs rather than a general request for an accountant.

The audience doing the asking is large. Pew Research Center, June 2025 found that 34 percent of U.S. adults had used ChatGPT, about double the share in 2023, and that 28 percent of employed adults said they use it for work.

Specific needs are the point. When a buyer describes an industry, a state and a tax problem, the engine answers with a short list of firms, and a firm that is not on it never gets the visit.

How buyers ask AI engines for an accounting firm

The AI Prompts below are examples, written the way business buyers tend to describe their situation.

1. Example: "Which CPA firms in Denver specialize in R&D tax credits for software companies?"
2. Example: "I own a licensed cannabis dispensary in Michigan. Which accounting firms understand Section 280E?"
3. Example: "Find an enrolled agent who can represent my small business in an IRS audit."
4. Example: "Recommend a bookkeeping firm for a construction contractor that needs job costing."
5. Example: "Who provides fractional CFO services for a venture-backed SaaS company preparing to raise a Series A?"

Each prompt carries a credential, a specialty, an industry or a location: the facts an engine has to find about your firm before it can name you. A homepage that only says "full-service accounting" gives it nothing to match.

What is GEO for accounting firms?

GEO for accounting firms, short for Generative Engine Optimization, is the practice of making a CPA, tax, bookkeeping or fractional CFO firm visible and accurately described inside the answers that AI engines write. When a business owner asks ChatGPT, Gemini, Claude or Perplexity which firm should handle a Section 280E return, an R&D credit study or a monthly close, the engine composes a short answer and often names a few providers. GEO is the work of earning a place in that answer. It has three parts. The first is measuring which firms the engines name today for the questions your buyers ask. The second is strengthening the public proof an engine can read about your firm, such as licenses, peer review status, enrolled agents and specialty pages. The third is measuring again on a fixed schedule, so each change can be tied to a result.

What AI engines can read about your firm: the proof layer

An AI engine cannot call your references. It works from what is written down: your website, public registries, directories, reviews and association listings. For an accounting firm, the strongest proof is what a regulator or professional body already publishes, because your claims can be checked against it.

CPA licensure and state board verification

CPA licenses are issued by state boards of accountancy, and NASBA, the national association of those boards, hosts CPAverify. NASBA describes CPAverify as a license search tool for CPAs and accounting firms, populated with licensing data sent directly from the boards, with 53 jurisdictions participating, and notes that it also carries markers of enforcement and disciplinary actions issued by state boards (NASBA, March 2026).

What to do: list every CPA on your team page by the full name that appears on the license, with the licensing state, and check that the firm name on your site matches the firm record where your state issues one.

AICPA peer review

The AICPA describes peer review as either a review of selected engagements for compliance with professional standards, or an evaluation of a firm's system of quality control. Its Peer Review Public File lists public accounting firms and their enrollment status in the AICPA Peer Review Program, and shows peer review documents for some firms (AICPA & CIMA, November 2023).

What to do: if your firm performs work covered by peer review, state your enrollment status on your site in the same terms the Public File uses, and keep it current.

Enrolled Agent status

The IRS defines an enrolled agent as a person who has earned the privilege of representing taxpayers before the IRS, either by passing a three-part comprehensive IRS test or through experience as a former IRS employee. Enrolled agents have unlimited practice rights and must complete 72 hours of continuing education every three years (IRS, January 2026).

The IRS also publishes a Directory of Federal Tax Return Preparers with Credentials and Select Qualifications, which it says can help taxpayers find preparers who hold IRS-recognized professional credentials or an annual filing season program record of completion (IRS, August 2026).

What to do: name each enrolled agent on your site, describe the representation work your firm actually takes on, and check that your credentialed people appear in the IRS directory.

Specialties you can prove

In our testing, narrow AI Prompts are answered with firms that have narrow pages. Each specialty deserves its own page that names the rule, form or problem and explains how the firm handles it. Three examples:

• R&D tax credit work. The research credit is claimed on Form 6765, Credit for Increasing Research Activities (IRS, June 2026). A page that explains how your firm documents qualifying research and prepares Form 6765 uses the same language buyers and engines use.
• Cannabis and Section 280E. The IRS states that businesses that traffic marijuana in contravention of federal or state law are subject to the limitations of Internal Revenue Code Section 280E (IRS, February 2026). A firm that serves dispensaries and cultivators should say so plainly, and name the states where it works.
• Crypto and digital asset accounting. For U.S. tax purposes the IRS treats digital assets as property, not currency, and broker reporting on Form 1099-DA began for transactions on or after January 1, 2025 (IRS, September 2026). A firm with real depth here should describe the reporting work it handles, such as reconciling broker forms against wallet and exchange records.

Claim only the specialties you can prove.

Reviews, directories and association profiles

Engines also read pages your firm does not control: your Google Business Profile, review sites, state CPA society directories, industry association member lists and software partner directories. Keep the firm name, address, phone number and specialties identical across all of them. Reviews that mention the specific work, such as a 280E return or an R&D credit study, give an engine detail it can repeat. Ask satisfied customers to describe what you did for them, within the advertising rules that apply to your firm.

How do AI engines decide which accounting firm to name?

AI engines do not publish how they choose which professional firms to name, so any honest answer comes from observation. In our testing across ChatGPT, Gemini, Claude and Perplexity, the firms that get named for a specific question tend to share three traits. First, the firm states the exact service in the buyer's own words on a dedicated page, for example Section 280E tax preparation for dispensaries rather than general tax services. Second, the firm's claims can be checked somewhere else: a license on CPAverify, a listing in the IRS preparer directory, enrollment in the AICPA Peer Review Public File, or a profile on a state society or industry association site. Third, reviews and articles on other sites describe the firm the same way the firm describes itself. When those three line up, engines tend to describe the firm with more detail and fewer qualifications.

A 10-prompt self-check you can run today

Replace the bracketed text with your own details. Run each prompt in ChatGPT, Gemini, Claude and Perplexity, in a fresh session where you can, and record for every answer whether your firm was named, which other firms were named, which sources were cited, and whether your firm was described accurately.

1. "Best CPA firm in [city] for small businesses"
2. "Which accounting firms in [city] specialize in [your top specialty]?"
3. "Who is the best accountant for [industry] companies in [state]?"
4. "I received an IRS notice. Which enrolled agents or CPAs in [city] handle representation?"
5. "Recommend a bookkeeping firm for a [industry] company with [number] employees"
6. "Fractional CFO firms for [stage] companies in [region]"
7. "Which [city] CPA firms have experience with [a specific rule or form, such as Form 6765 or Section 280E]?"
8. "Is [your firm name] a good accounting firm?"
9. "What does [your firm name] specialize in?"
10. "Compare [your firm name] and [main competitor]"

Prompts 1 to 7 test discovery, when the buyer does not know your name. Prompts 8 to 10 test accuracy, when the buyer does.

As an example of the arithmetic, using your own counts: ten prompts across four engines gives forty answers. If your firm were named in 6 of those 40, your AI Mention Rate for this set would be 15 percent. What matters is a baseline you can repeat next month.

How to measure ChatGPT visibility for accounting firms

Use the same names every month so the results can be compared:

• AI Prompt: one fixed question you run across engines. Add prompts rather than swapping them out.
• AI Mention Count: the total number of times your firm is named across all answers in a period.
• AI Mention Rate: the share of answers in which your firm is named.
• AI Mention Share: your mentions as a share of all mentions of the firms in your defined competitive set.
• AI Engine Coverage: the per-engine breakdown of which engines name you, and for which prompts.
• AI-Referred Traffic: visits to your site that arrive from AI engines, which most web analytics tools can show by referring domain. Pair it with form fills and booked consultations.
• AI Visibility Score: a composite of how often and how prominently AI engines name your firm, counting only AI Prompts that do not include your firm's name.

How should an accounting firm measure its AI visibility?

An accounting firm should measure AI visibility the way it would test any engagement process: with a fixed sample, a written method and a repeat schedule. Start with a list of AI Prompts drawn from real buyer questions, covering your specialties, your locations and your firm's name. Run every prompt in ChatGPT, Gemini, Claude and Perplexity on the same day each month and record whether the firm was named, which other firms were named and which sources were cited. From that log, calculate AI Mention Count, AI Mention Rate and AI Mention Share against a defined set of competitors, and break the results down by engine for AI Engine Coverage. Then check AI-Referred Traffic in your web analytics and tie it to consultations booked. Changing the prompt list between months breaks the comparison, so add prompts rather than replace them.

GEO services for accounting firms: what a good agency does

Whether you do this in house or hire an accounting firm GEO agency, the order is the same: a measured baseline across all four engines, then the proof layer (credential, specialty and team pages and directory profiles that agree with the public registries), then monthly reporting in the metric names above.

Analytical Insider measures and improves how AI engines mention a company; see our Generative Engine Optimization page and, for plan details, our pricing page. Be wary of any promise of a fixed placement in AI answers. In our experience, the answers change as engines update, and no outside party controls them.

Where verified pipeline fits

GEO improves the odds that a buyer who is already asking will hear your name. Many firms also want to reach businesses that are not asking yet, such as software companies that have never looked at the research credit. That is outbound work, and it needs a clear definition of who to reach.

Analytical Insider also runs AI sales agent teams that deliver verified leads. The same specialty definitions that shape your GEO pages, such as industry, size, state and tax problem, become the profile the sales agents work from.

What to do this month

1. Run the 10-prompt self-check across all four engines and save every answer.
2. Look up each CPA on CPAverify and each enrolled agent in the IRS preparer directory, and fix any name mismatches on your site.
3. If your firm is enrolled in the AICPA Peer Review Program, state your status on the site in the Public File's terms.
4. Pick your two strongest specialties and give each one a dedicated page that names the rules, forms and industries involved.
5. Align your Google Business Profile, state society listing and key directory profiles with your website.
6. Ask recent satisfied customers for reviews that describe the specific work, within the advertising rules that apply to your firm.
7. Set up AI-Referred Traffic reporting and schedule the same ten prompts for next month.

Related reading

• How to get your brand cited in ChatGPT responses
• GEO vs SEO: what is the difference and which one do you need?