AEO for accountants and CPAs: be the name AI gives.

Accounting demand runs on a printed calendar — January 15, March 15, April 15, the extension dates in the fall — and each deadline now sends its questions to AI assistants first. A business owner asks ChatGPT who should untangle an S corp's books; a stressed filer asks who can still take a return in March. AEO for accountantsmeans being in those answers before each wave crests — and knowing, day by day, whether the engines name your firm or a competitor's. This page maps the question cycle to the filing calendar and lays out what a firm builds ahead of it.

Two clients, two calendars

An accounting firm sells to two buyers who behave nothing alike in an AI chat, and lumping them together is the first mistake a visibility plan can make. The engines never lump them: they compose a fresh answer for every question, so the firm named for S-corp questions and the firm named for last-minute 1040 questions can be different firms in the same city.

The small-business client: year-round, relationship-shaped

The owner of an eight-person company is not shopping for a return; they're shopping for the firm that closes their books every month. The questions carry entity types and industries — “best CPA for an S corp in Boston”, “who should handle my startup's R&D credit?” — and they arrive all year, clustering around quarterly estimated-tax deadlines and year end. One answered question can decide years of monthly fees, which is why owners ask an assistant before committing to a CPA relationship, not after.

The individual filer: a compressed season of urgency

The 1040 buyer asks in a narrow window and in a hurry, usually with the complication stated up front — “a rental, two states, and equity that vested”. Volume is high, loyalty is lower, and the season decides the year: the firm the engines keep naming in February and March fills its book; the firm they skip waits for the next wave.

The filing calendar is the demand curve

Accounting is the rare profession whose demand spikes are printed on a government calendar. That makes the AI question cycle unusually predictable — and makes timing a planning input instead of a guess:

• Mid-January.W-2s and 1099s land, filing season opens, and the fourth-quarter estimated payment falls due January 15. The 1040 wave starts building; business owners resurface with “my books are a mess from last year” cleanup asks.
• March 15. S-corp and partnership returns come due a month before most owners expect — the deadline people learn about from a penalty notice. The questions turn specific and urgent: an extension, a new firm, or both.
• April 15.Peak. Individual returns and first-quarter estimates land the same day, and “who can still take me” asks crowd every engine.
• Summer. The 1040 tide goes out and the switching window opens: owners who felt underserved in April interview replacements in June — calmly, one private AI conversation at a time.
• September 15 and October 15. Extended business and individual deadlines pull a second, smaller wave.
• November–December.Year-end planning, entity-election questions, next year's bookkeeping decisions — the quarter when relationships form, and the building window for any firm that wants to be visible by January.

The questions, filed by deadline

Below is the calendar again, this time as the questions themselves. Fifteen shapes, filed by the window they arrive in, written the way clients write them — with the entity, the dollar figure, the state, and the panic all in one sentence. A partner reading this list should be able to point at three questions their firm answers better than anyone in the metro and then check whether a page on their site says so.

The lesson in the arrangement: the questions a firm most wants — entity elections, niche industries, multi-state complexity — are asked in the two quietest quarters, months before the season the firm actually markets in. Demand and visibility work run on different clocks, and only one of them is printed on the IRS calendar.

That gap matters because citations have lead time. In published testing, a page takes a median of about 6.8 days to earn its first ChatGPT citation once indexed, and roughly 42% of pages get cited within 30 days (Semrush / practitioner testing, 2026) — observed medians, not a promise. A firm that ships its specialty pages and directory cleanup in October is accumulating evidence while competitors wait for the season; a firm that starts in February is publishing into the peak. The step-by-step of that shipping work is in how to get your business mentioned by AI.

A trust hire, decided in a private chat

An accountant reads every number a business has, including the embarrassing ones. That's why the hire has always run on referrals — you borrowed someone else's trust. What's replacing the referral isn't a results page; it's a private conversation. People tell an assistant things they'd never open with at a networking event — “we haven't reconciled in two years”, “I think my last preparer got the depreciation wrong” — and then ask who can fix it. The question your next client asks about your market is being asked where you can't hear it.

Direction is enough, because the second fact is measured and blunt: strength on Google does not carry into AI answers.

How a firm loses the answer it should own

For accounting the loss has four recognizable causes, and none of them is about the quality of the work:

• Software is the default answer, and it is a good one. For anything simple, engines point at consumer tax products and national storefronts, both of which have published thousands of indexed explainer pages. A local firm never enters that contest, and should not try to.
• Directory rosters answer the local question.The IRS directory of federal tax return preparers, review-site roundups, and “best accountants in [city]” listicles are structurally ideal for a shortlist question: a filterable list of names on one quotable page. A firm not on those rosters is absent from the pool; a firm listed under a name variant is effectively a different firm.
• Firm websites say nothing an engine can use. “Tax. Accounting. Payroll. Advisory.” Four nouns answer none of the fifteen questions above. The partner's real expertise — the construction WIP schedules, the dental practice buy-ins, the multi-state nexus work — usually exists only in the partner's head and in client conversations no crawler will ever see.
• The license record is a single line of text. A state board listing confirms the firm exists. It carries no specialty, no industry, no context — so the strongest verification an accounting firm has contributes almost nothing to a niche question.

The engines assemble a recommendation from their own evidence — what they learned in training plus what they retrieve live — and for a trust profession that evidence is third-party by design: license records, certifications, reviews, and the consistency of your identity across all of them. The mechanism is worth ten minutes of any partner's time: how ChatGPT actually picks businesses explains the pipelines, and the six reasons a business stays invisible is the diagnostic. Law firms are living the same shift a chapter ahead — AEO for law firms shows where trust-bought professions end up.

Specialization wins answers a generalist can't

Here is the accountant-specific twist most visibility advice misses: AI engines answer the question as asked, and buyers ask with their industry in the sentence. “Best CPA in Chicago” is a contest against every generalist in the metro plus the national chains. “CPA for construction contractors in Chicago”, “accountant for Amazon sellers”, “bookkeeper who knows restaurants” — those are contests most firms never entered, because nobody published a page that says it plainly.

Specialty positioning pays twice. It matches how firms actually grow — niche accounting has been the profession's most reliable differentiation play for a decade — and it matches how engines compose: the firm whose site, advisor-directory profiles, and reviews all repeat “construction contractors” is the natural name for that question. Specialty answers also travel: for remote-friendly work, a firm known for one niche gets named outside its own metro. If your firm genuinely has a niche, this is the cheapest visibility contest available to it. If it doesn't, weigh the whole case with cold eyes first — is AEO worth it is written for exactly that skeptic.

Where the engines read about accountants

Recommendations get assembled from a recognizable public record. For accounting firms, that record concentrates in a few places:

• Advisor directories: the QuickBooks ProAdvisor and Xero advisor listings — verified, specific about software stack and industries served, and exactly the kind of third-party page an engine can quote.
• The license record: your state board of accountancy listing — the anchor for the detail firms get wrong most often, the exact registered firm name.
• Review surfaces:Google Business Profile and Yelp — where recency and specifics get read. “Cleaned up three years of books for our restaurant” carries more evidence than a bare five-star rating.
• Communities:Reddit's small-business and tax threads, where “anyone have a CPA who understands e-commerce?” gets quoted into answers — see why AI cites Reddit so much.
• Your own site, if the crawlers are permitted. OAI-SearchBot is the crawler tied to appearing in ChatGPT Search; GPTBot is primarily training-data collection, and blocking GPTBot does not remove a firm from ChatGPT Search. Anthropic splits the same way with Claude-SearchBot and ClaudeBot. Professional-services sites collected blanket blocks in 2024; the file is worth a deliberate second read.

Consistency across that record is the unglamorous part that decides close calls. If the state board lists one firm name, the ProAdvisor profile another, and the website footer a third, the engines can't tell that all three records vouch for the same firm — the evidence fragments, and the recommendation goes to a firm whose record adds up. Structured data helps here for a narrow, honest reason: Google states no special markup is required to appear in AI Overviews or AI Mode, but an AccountingService block is a clean way to state the registered name, the license, the specialties, and the service area once, unambiguously, matching the visible page. The machinery underneath the local questions — Google Business Profile, reviews, NAP — has its own playbook: how AI picks “near me” businesses is the companion page.

Trust rules a firm cannot bend — and what we will not do

Accounting sells confidentiality and judgment, which makes the content standard stricter than in most local marketing. State boards of accountancy govern who may call themselves a CPA and how a firm may advertise; client confidentiality duties govern what may ever be said about the work. The honest way to evaluate a content vendor here is the list of things it will not write. Ours:

• We will not claim a tax outcome.No “we cut your tax bill,” no refund estimates, no savings figures, no “IRS problem solved.” Pages state what the firm does and who it does it for.
• We will not tell a client's story.Not anonymized, not composited, not “a manufacturer we helped.” Confidentiality does not have a marketing exception, and an invented case study is worse than none.
• We will not describe a credential the firm does not hold. CPA, EA, and state-license language appear only as the firm states them. This is a state-board matter, not a tone preference.
• We will not write a superlative or a ranking.No “best CPA in the city,” including inside title tags and structured data, where superlatives usually survive a copy review.
• We will not list a service in schema the firm does not perform, or that the visible page does not mention.
• We will not promise a season.A firm that starts in February gets February's honest odds. The citation timelines on this page are medians observed in testing, and we say so every time they appear rather than converting them into a delivery date.

And nothing publishes without a partner's approval, which matters more here than in most industries: between February and April there is no attention to spare for reviewing marketing copy, so the queue simply waits. None of the above is legal or accounting advice; state rules differ and your firm decides.

What gets built for a firm — and when to start

Disclosure, so you can weigh what follows: this page is published by AEO Action, and this section describes our own product. Everything above it is true whoever does the work.

AEO Action runs the loop as software — an autonomous AEO department with an approval gate — and for an accounting firm the output maps onto the ledger above rather than onto a generic content plan:

• A niche page per specialty the firm genuinely has. Contractors and WIP, dental practice buy-ins, e-commerce sales tax, real estate depreciation — one page per specialty question, stating the work, the entity types, and who at the firm does it.
• A deadline page for the questions the calendar creates. What happens if a March 15 S-corp return is missed, what an extension actually buys, what a CP2000 notice means. These are the asks that arrive with urgency attached, and they are almost always answered by a national brand today.
• A capacity page for “who can still take me.” Whether the firm is accepting new clients, for which work, through which date. That single fact decides the whole April question set and appears on almost no firm website.
• An entity-and-planning FAQ set for the quiet quarters. S-corp election timing, quarterly estimates, bookkeeper versus CPA — the November questions that turn into January engagements.
• Structured data plus a Reddit research pack. One machine-readable statement of the registered firm name, license, specialties, and service area matching the visible pages; plus the small-business and tax threads where these questions get asked, prepared as research rather than as posts.

The cadence, read against the calendar. Each day's scan asks the engines your clients' questions — regenerated per scan from the firm's profile, specialties, and cities, across up to six engines: ChatGPT, Gemini, Perplexity, Google AI Overviews, Microsoft Copilot, and Claude (the last a labeled model mode, never counted as live) — and the orchestrator prepares the single highest-value artifact it finds. Nothing publishes without your Approve; up to four artifacts wait on day one of a trial, depending on the scan. Start in October and by Thanksgiving that cadence is a shelf of specialty pages the February starters will not have.

Pages that go live are submitted through IndexNow within minutes — Bing decides what to index, and no honest vendor claims otherwise. Worth stating precisely why the step is in the routine: independent testing found roughly 87% of ChatGPT's citations also rank in Bing's top results (Seer Interactive, 2026). That is a correlation measured by a third party. OpenAI documents OAI-SearchBot as the crawler for ChatGPT Search and does not publish which index or indexes sit behind it, so the honest sentence is that the overlap is large and worth working, not that a pipeline has been documented. The Bing and Copilot guide takes it apart properly.

Then the receipts, which is what a partner will actually judge this on. Every approved landing page is registered against its target question and re-checked on each following scan for up to 120 days: shipped on date A, first cited on date B. By April you know which of October's pages earned their keep. Every seven days, all year, a plain-English report a partner can read between client calls — what was asked, what changed, what shipped, what got cited. In the quiet quarters it reads like maintenance; in March it reads like a scoreboard.

Publishing works with whatever the firm's site runs on: Webflow and Squarespace publish directly, the WordPress plugin pulls approved schema hourly and approved drafts every six hours — we never push into your site — and everywhere else the work arrives as paste-ready HTML the system verifies once live.

The price sits against the retainer market, not the dashboard market: the Solo plan is $149 a month next to marketing retainers commonly at $2,000–6,000/month(2026 market research). An agency is still the right pick for some firms — if you want a human team running your ads, website, brand, and this, you're buying staff time, and staff time is priced as a retainer. The software case is the firm that wants this one loop run daily through the whole filing calendar without another hire; the services page lists what that loop covers, and the full price ladder, rung by rung, is in how much AEO costs in 2026. If an enterprise monitoring platform is also on the shortlist, the comparison with Profound is written for that decision.

The first three minutes are free either way: the audit puts ten questions clients ask about firms like yours to ChatGPT, Gemini, and Perplexity and scores the 30 live answers, quoting the three most revealing word for word with competitor names highlighted. No account, no email, no card. The PDF is permanent; the share link lasts seven days.