AEO audit for an accounting and advisory firm

The firm

Little is known about how businesses use AI to choose an accountant. A marketing firm that serves professional-services firms wrote in July 2026 that it is "very difficult to even know what proportion of professional services buyers are using AI to research and select firms", and AEO HQ's industry research found no survey of how small businesses choose an accountant. Broader surveys point the same way. Among 519 U.S. B2B professionals who use AI, agencies and service providers were the category most often researched with AI, at 51% (the publisher sells search-marketing software). For local businesses of all kinds, use of ChatGPT and other generative AI tools for recommendations rose from 6% to 45% in a year (1,002 U.S. adults; the publisher sells local SEO software).

The questions clients ask assistants

This example uses the smaller panel in AEO HQ's measurement design: 20 client intents, each written three ways (60 unbranded prompts), plus 20 branded prompts about the firm and its people. Six intents are local, three services in each office city, and they name the city, because ChatGPT may use an approximate location based on the user's IP address. Wording comes from the questions partners hear in first meetings, the firm's Search Console queries, and contact-form messages, with client details removed.

Each prompt runs three times a week on each of the four assistants the audit measures (ChatGPT, Perplexity, Gemini, and Google AI Overviews), in a new chat with a clean session; the ten most important run eight times a week. Results are reported as rates, because the same prompt returned the same list of brands less than once in 100 runs (2,961 runs; industry study; a co-investigator works for a tracking vendor). The table shows 13 of the 80 prompts. The wording is illustrative.

How the assistants reach the firm

A firm's pages reach assistants through search indexes, and its local details also through Business Profiles and listings; how AI assistants find and cite sources covers each route. One route is specific to business services. Clutch, a B2B marketplace that lists accounting firms, launched an app inside ChatGPT on April 23, 2026 that brings "verified profiles, review highlights, pricing signals, and real client feedback" into a conversation when a user opens it.

• Google. AI Overviews and AI Mode link only to pages that are indexed and eligible to be shown with a snippet, and Google lists keeping Business Profile information up to date among the basics for its AI features. Google's examples of individual practitioners who may have their own profiles are doctors, dentists, lawyers, financial planners, and insurance or real estate agents. Accountants are not named, so whether a partner qualifies turns on Google's general test of a public-facing professional with their own customer base (our reading). See how to rank in Google AI Overviews and AI Mode.
• Local answers differ from Google's. For local tax preparation queries, the domains cited by a web-enabled AI engine and by Google overlapped by 15.4% (preprint), so a strong Maps position does not carry over on its own (our reading).
• ChatGPT. Sites that block OAI-SearchBot are not shown in ChatGPT search answers, and ChatGPT may share a user's location with providers of local information "such as nearby business listings". See how to rank in ChatGPT search.
• Copilot, Claude, and Perplexity. Copilot sends its queries to the Bing search service, Claude draws on Brave Search and Anthropic's own crawler, and Perplexity on its own index. See the guides for Claude and Perplexity.

What the audit checks

Access, indexing, and local profiles

• Crawlers. robots.txt and firewall rules are read for OAI-SearchBot, Claude-SearchBot, PerplexityBot, Googlebot, and Bingbot, since a crawler follows the group that names it, and the * group only when none does.
• Page text. Service descriptions, bios, and office details must be in the HTML the server sends: in Vercel's December 2024 data, none of the major AI crawlers rendered JavaScript.
• Index coverage. Service pages, bios, and office pages are checked in Search Console's Page indexing report and in Bing Webmaster Tools.
• Business Profiles. Names must reflect the business's real-world name, and an office in a co-working space qualifies only if it maintains clear signage, receives customers during business hours, and is staffed during business hours.
• Fees and titles on third-party profiles. Marketplace and directory profiles are compared with the firm's fact sheet, including old "starting at" prices.

Fees and prices

Buyers ask what firms charge, and assistants answer even when a firm publishes nothing. In one published example, ChatGPT said an accounting firm "does not publish a standard, one-size-fits-all price sheet" and still gave specific price ranges and hourly rates (one example, July 2026; the author's firm sells professional-services marketing). More broadly, 27% of AI-using B2B buyers say AI answers do not reflect real pricing or contract structures. The rules allow fee information but set conditions, listed in the claims table below. We recommend publishing either fees or the factors that set them, within those conditions.

Claims and the rules that limit them

Each claim is listed with its URL and the rule it touches, for the partners and counsel to decide. The rows summarize the AICPA Code of Professional Conduct, which applies to AICPA members; Treasury Circular 230, the regulations on practice before the IRS; and 22 Tex. Admin. Code § 501.82, the advertising rule of the Texas State Board of Public Accountancy.

CPAs and enrolled agents are both among those who may practice before the IRS, so representation pages should say which people do that work. GA4 key events and a source question on the contact form are checked against how to track AI referral traffic in GA4.

Sample findings

These are sample findings for the hypothetical firm. They show the form a finding takes and the evidence it needs; they describe no real firm, and nothing in them was measured. Impact is our judgment of how directly a finding affects what assistants can find and state correctly, or how much rule risk it carries; it is not a forecast.

The deliverable

The report is a PDF with a 60-minute readout call:

1. Summary. Scope, dates, and the fixes to make first, with rule items marked.
2. Method. Panel, assistants, dates, location, run counts, and matching rules, so the baseline can be repeated.
3. Baseline rates. Per assistant, and per city for local prompts: mention rate, citation rate, share of voice against eight named local firms, and branded accuracy, each with counts and 95% intervals. No pooled score and no rank.
4. Source map. Cited domains grouped as the firm's site, marketplaces and directories, review sites, publishers, and government sources.
5. Fee and credential accuracy. Each branded answer about fees, licenses, and services graded against the fact sheet and license records.
6. Access, indexing, and profiles. Crawler access, rendering, index status, and every profile against the fact sheet.
7. Claims review. Each item with its URL and rule, for the partners and counsel.
8. Findings and fix plan. Owners, effort, and how each fix will be verified.
9. Measurement plan and run log. The plan below, and every answer and citation from the baseline.

Measurement plan

Every measurement uses the same frozen panel, following the methodology page and how to measure AI visibility. Two weeks give 460 unbranded and 120 branded runs per assistant, 2,320 in all.

City rates rest on few runs: 5 of 54 runs (9.3%) has a 95% Wilson interval of 4.0% to 19.9% (AEO HQ calculation), before correlated runs widen it. They order fixes; they cannot show small differences. A change between windows counts only when the interval for the difference excludes zero, as the statistics section describes.

Engagement timeline

The pricing page gives the audit's current delivery time; this example assumes about two weeks from completed intake.

What this example does not show

• Results. No rates, inquiries, or clients, because the firm does not exist.
• How long changes take. A 2026 review of 45 studies found no technique with "a stable, longitudinal, cross-platform causal effect on organic discoverability" (preprint).
• A compliance opinion. AEO HQ flags items against the cited rules; the partners and counsel decide. Only the Texas board's rule was checked, and other states differ.
• Attest and licensing matters. Independence, firm licensing, and peer review are outside the audit.
• Consulting-only firms. AEO HQ's research found no profession-specific advertising code for U.S. management consultants; the FTC's general rule that advertisers must have evidence to back up their claims still applies.
• Evidence specific to accounting. The surveys cover business services or local businesses in general, and several come from companies that sell to firms.

Next step

The audit's current scope, price, and delivery time are on the pricing page. SEO, AEO, and GEO for professional services firms covers the industry in more depth.