How Accountants and CPA Firms Get Recommended by AI

By the alphaa team — we run AI-visibility scans across thousands of small businesses, including accounting and tax practices. Last updated 29 July 2026.

Short answer: AI assistants recommend the accounting firms they can describe confidently. That means three things in practice: your site says specifically who you serve and what you charge to do what, your firm details and licence status are consistent and verifiable across Google, state board and IRS directories, and other people — reviewers, forum threads, local press — say the same things about you. Firms that describe themselves as a "full-service accounting firm serving businesses of all sizes" give a model nothing to hold on to, so it names someone else.

What people actually ask AI when they need an accountant

The queries that matter come in two waves, and most firms only think about the second one.

• Wave one — the problem question. "Do I need a CPA if I just formed an S-corp?" "What's the difference between a bookkeeper and a CPA?" "How much should a small business pay for tax prep?" "Can I deduct my home office if I'm a 1099 contractor?"
• Wave two — the who question. "Find me a CPA in Boise who works with restaurants." "Best accountant for freelancers near me." "Who handles multi-state payroll tax for small employers in Ohio?"

Being the source the model used to answer wave one is a large part of how you become the name in wave two. This is the same dynamic we described for law firms: professional services get recommended through explanation, not through slogans.

Notice too how loaded those wave-two queries are with qualifiers — a city, an entity type, an industry, a specific tax problem. A model can only match those qualifiers to firms whose pages actually contain them.

Why accounting firms are unusually invisible to AI

Four patterns show up again and again in scans of accounting practices:

• Deliberate vagueness. Many firms avoid naming a niche because they fear turning work away. The result is a homepage that could belong to any of 40,000 US firms. Vagueness is the single biggest AEO problem in this profession.
• No pricing signal at all. "Contact us for a quote" is a reasonable business decision and a retrieval dead end, because "how much does a CPA cost" is one of the highest-volume questions in the category.
• Content locked in PDFs and portals. Tax guides published as downloadable PDFs, or behind a client login, are far weaker signals than the same content as a normal web page.
• Seasonal silence. A site that updates in February and goes quiet until the following January reads as stale for ten months of the year.

Step 1 — Fix the entity layer

Before content, make your firm resolvable as one clear entity. AI engines cross-check what they read, and accounting is a category with unusually good public verification sources. Use them.

• One exact firm name everywhere. "Harding & Co. CPAs PLLC" on the website, the Google Business Profile, the state board record, LinkedIn, and every directory. Not "Harding CPA" in one place and "Harding and Company" in another.
• Correct Google Business Profile category. "Certified Public Accountant", "Accountant", "Tax Preparation Service" and "Bookkeeping Service" are distinct categories that surface for distinct queries. Pick the primary one that matches your actual licence and revenue mix, then add the others as secondary.
• Be findable in the verification sources. Your state board of accountancy licence lookup, the IRS Directory of Federal Tax Return Preparers, your state CPA society directory, and the AICPA if you are a member. These are exactly the kind of authoritative, structured records that make a model comfortable naming you.
• State the credential precisely and truthfully. "CPA licensed in Texas (licence #…)", "Enrolled Agent", "QuickBooks ProAdvisor". Never imply a credential you do not hold — this is both a regulatory problem and a trust problem, and a contradiction between your site and the board record is the fastest way to make a model hedge.

The mechanism behind all of this is entity resolution, covered in depth in how AI identifies your business as an entity.

Step 2 — Trade vagueness for specificity

Here is the rewrite that moves the needle. Before:

We are a full-service accounting firm providing accounting, tax and advisory services to businesses and individuals. Our experienced team is committed to excellence and personalised service.

After:

Harding & Co. is a two-partner CPA firm in Boise, Idaho. We do tax preparation, monthly bookkeeping and S-corp payroll for independent restaurants, food trucks and craft breweries across Idaho and eastern Oregon. Typical engagements: a single-location restaurant on monthly books and an 1120-S return, from $450/month. We do not take individual-only 1040 clients.

The second version is 60 words and contains a city, a state pair, an industry, three named services, two form numbers, a starting price, and an explicit exclusion. Every one of those is a hook a retrieval system can match against a real question. The first version contains nothing retrievable at all. Note that the exclusion helps you: telling a model who you are not for makes it more confident recommending you to the people you are for.

Step 3 — Publish the answers, not brochures

Pages that get quoted in this category tend to be:

• Cost pages with real ranges. "What we charge for a 1065 partnership return" — with the ranges and the factors that move them. Caveat them honestly; a range beats silence.
• Entity and situation guides. Sole proprietor vs LLC vs S-corp, when the S-corp election stops saving money, what changes when you hire your first employee in your state.
• Jurisdiction-specific pages. State-level filing requirements, local business licence and franchise tax obligations. National content is a commodity; your state and city are not.
• Deadline and process pages kept current for the filing year, with a visible last-updated date. Tax content decays fast, and both readers and models discount undated advice.
• A clear FAQ in plain HTML — short question headings and two-to-four-sentence answers, which are easy for an engine to lift verbatim. Pair it with appropriate schema markup so the structure is machine-readable too.

Step 4 — Earn the third-party evidence

Accountants are chosen on trust, and models weight independent corroboration accordingly. Ask every client you finish a good year with for a Google review, and encourage specificity — "handled our multi-state payroll registration" is worth far more to retrieval than "great service." The mechanism is explained in why your Google reviews shape your AI visibility. Beyond reviews: be genuinely useful in the places small-business owners ask tax questions, get listed with your local chamber and industry associations, and answer reporters' questions when tax season stories come around.

What not to do

• Do not promise outcomes. "Guaranteed maximum refund" and "we can cut your tax bill in half" are both a compliance risk under state board and IRS advertising rules and a credibility problem with AI engines, which increasingly discount pages that read as promotional rather than informational.
• Do not invent credentials, awards or client logos. These get cross-checked.
• Do not mass-produce thin city pages. Fifty near-identical "CPA in [City]" pages for cities you do not serve is the oldest trick in local SEO and it degrades trust in everything else on your domain.
• Do not buy "guaranteed AI placement." No vendor, including us, can guarantee a mention in ChatGPT. What you can do is improve the evidence a model reads.

A realistic 30-day plan

1. Days 1–3: Ask ChatGPT, Claude, Gemini and Perplexity the ten questions a real prospect would ask in your city and niche. Record what they say and who they name.
2. Days 4–7: Reconcile your firm name, address, phone and credentials across the website, Google Business Profile, state board record, IRS directory and LinkedIn.
3. Days 8–14: Rewrite the homepage and services pages for specificity — niche, geography, named services, price signal.
4. Days 15–24: Publish two genuinely useful pages: one cost page, one state-specific situation guide. Add an FAQ block to each.
5. Days 25–30: Request reviews from your ten happiest clients, asking them to mention the specific work. Re-run the ten questions from day one and note any change.

Be patient about the timeline. Entity and review signals propagate over weeks to months, not days — see how long AEO actually takes.

Quick answers

Will AI recommend a solo practitioner over a big firm?

Regularly, yes — when the query has qualifiers the solo firm matches and the big firm does not. A two-partner firm that clearly serves restaurants in one state will beat a regional firm with a generic site for "CPA for a restaurant in Idaho." Specificity is the great equaliser here.

Do I need to publish prices?

You do not have to, but a range with context materially improves your chances on the highest-intent cost queries. If you truly cannot, publish the factors that determine price instead — that is still retrievable content.

Is this different from SEO?

It overlaps, but the target differs: SEO optimises for a ranked list of links, AEO optimises for being the firm a model is confident enough to name. Consistency and verifiability matter more; keyword density matters less.

Can any of this be guaranteed?

No. AEO shapes the public signals AI systems read. It shifts probability, it does not buy placement, and answers vary between engines and over time.

The bottom line

Accounting is a category where the winning move is uncomfortable: say precisely who you serve, what you do, roughly what it costs, and who you turn away. Make that story identical everywhere a machine can check it — your site, Google, your state board, the IRS directory, your reviews. Do that and you stop being one of forty thousand indistinguishable firms and start being the specific answer to a specific question.

Run a free AI visibility scan →