Written by Miriad with AI assistance and checked against the sources at the end. A practical guide for owners of small accounting and bookkeeping firms, not legal or ethics advice: your professional body and your regulator decide.
TL;DR: Most accounting firm marketing strategies need no client data at all. Referrals, reviews, a clear website, local search, a tax-change newsletter and social posts can be drafted with AI from public rules and your own know-how. Client files never go into a public AI tool. Plan the work in the quiet months, use the client list only as the rules allow, and check every AI draft the way you would check a trainee's work.
It is the second week of February. The partner of a six-person practice has forty returns waiting for documents, a bookkeeper off sick and a sticky note on the monitor that says "fix the website". Marketing happens in the gaps, if it happens at all. When someone suggests that AI could help, the answer comes fast: "We can't put client data into ChatGPT."
That answer is right, and it is also where most firms stop thinking. Look at what a small firm's marketing is made of: a services page, a fee explanation, a review request, a newsletter about a deadline, a post about a new reporting rule. None of these needs one client's figures. The confidentiality rule decides which inputs are allowed. It does not rule out the work.
This guide takes the main marketing strategies for an accounting firm and checks each against one question: which client data does it touch? Where the answer is "none", AI can write the first draft. Where it is "the client list" or "the client file", a specific rule applies, and we name it with its source. Each section links to a deeper article; the full set is on our accounting firm articles page.
Why do accounting firms hold back on AI in marketing?
Accounting firms hold back because confidentiality is the core of the job, and the risk is real: ICAEW says that putting client data into AI inputs is very likely a breach. The worry is justified for client files. It is not a reason to keep AI away from marketing, which runs mostly on public facts and your own expertise.
The rules behind the worry are clear, and they are worth knowing in their exact wording:
- ICAEW (UK): if AI inputs include client data, it is "very likely that this will breach client confidentiality", and AI output should be reviewed as you would a trainee's work (ICAEW, 2024).
- AICPA Code 1.700.001 (US): a member does not disclose confidential client information "without the specific consent of the client" (AICPA Code, 2024).
- IRS Publication 5708 (US): under the Gramm-Leach-Bliley Act and the FTC Safeguards Rule, tax and accounting professionals count as financial institutions "regardless of size" and must keep a Written Information Security Plan (IRS, 2024).
So the professional rules do not ban AI. They ban careless inputs. The gap is elsewhere: in a 2026 vendor survey of about 600 accounting professionals, 98% said their firms use AI, but only 21% had an AI policy or strategy (Karbon, 2026). People are already using the tools. What is missing in many offices is the written line between what may go in and what may not.
Small firms feel this pressure on two fronts at once. The AICPA's 2026 Top Issues survey, as reported by CPA Practice Advisor, puts cybersecurity and data privacy in the top five current issues for sole practitioners and firms of 2 to 10 professionals, and names managing change from technology and AI as the leading five-year issue across firm sizes (CPA Practice Advisor, 2026). The way through is not to avoid AI, but to sort the work by the data it needs.
What single test should every marketing strategy pass?
Ask of every strategy: which client data does it touch? There are three answers. None, which covers most marketing. The client list, which covers newsletters, review requests and referral asks. Or the client file, which covers case studies, testimonials and anything with figures. Each level has its own rule, and the rule decides the tool.
| Data level | Typical marketing work | Rule to check | How AI may help |
|---|---|---|---|
| None | Website pages, articles on tax changes, social posts, fee explanations, local search profile | Advertising rules: honest, no exaggerated claims (ICAEW Code, 2024) | A public tool may draft from public sources; a qualified person checks every figure and date |
| Client list (name, email, return type) | Newsletter, review request, referral ask | US: 26 CFR 301.7216-2(n) limits what the list holds and what it may be used for (Cornell LII) | AI drafts the message text; the list stays in your email or practice management software |
| Client file (figures, documents, outcomes) | Case studies, testimonials, "before and after" stories | Specific client consent (AICPA 1.700.001); no client data in public AI (ICAEW) | Written consent first; draft by hand or in a tool the firm has approved for client data |
Example (illustrative, not a real case): A four-person bookkeeping firm lists its marketing jobs for the coming year: 14 items. Sorted by the test, 10 touch no client data (a website rewrite, a monthly batch of social posts, the body text of four newsletters, a fee page, the local search profile and five more), 3 touch the client list (sending the newsletters, review requests after year-end, one referral ask) and 1 touches client files (a case study with a construction client). The firm lets AI draft the first ten from public sources, keeps the list inside its email tool, and books the case study for July, after a signed consent form.
The test also tells you where to start. The "none" row is safe ground for learning how AI drafting works in your office, and it covers most of the visible marketing. The other two rows wait until the rules in the later sections are written down.
How do clients actually choose an accounting firm?
Mostly through people they trust. In a 2022 Chorus Insight study of 159 professional-services buyers, cited by ACCA's Accountancy Business, word of mouth was how 48% first chose their current adviser, against 20% for search engines and 8% for online advertising (ACCA AB, 2025). Referrals first, search second.
That study is not accounting-only, and it is not recent, so treat it as a direction rather than a forecast. It still matches how a small practice usually grows. The typical client journey, in our own ordering from several sources, runs like this: a referral or a search (ACCA AB), then an intake form or a first call (Journal of Accountancy), then a written notice of the outside tools that will hold the client's data (AICPA Code), then onboarding and document collection, and then recurring, deadline-driven work.
Search and reviews do their job even inside a referral. A friend gives a name, and the prospect looks it up. A 2026 survey of 1,002 US consumers, across all industries, found that 97% read reviews for local businesses, 89% expect owners to respond, and 74% only consider reviews from the last three months (BrightLocal, 2026). For an accounting firm, that argues for a steady review routine after each busy period, not one push a year.
The review rules are strict, and they match the confidentiality instinct. Google bans incentives for reviews, selective requests aimed only at happy clients and review quotas for staff, and it treats reviews by current or former employees as a conflict of interest (Google Business Profile Help). Ask every client the same way, and reply without confirming any detail of their affairs.
Two articles go deeper here. Accounting firm leads starts from the clients you want and works backwards to the channels. Accounting firm marketing ideas lists ideas sized for a firm where the partners also do the client work.
When in the year should an accounting firm do its marketing?
Do the building work in the quiet months, and only light, pre-written work in the peaks. US firms face the April deadline and the extension season; UK practices peak around 31 January; Hungarian offices around May. A plan that ignores the filing calendar collides with client work and is quietly dropped by March.
The deadlines themselves are public and fixed early, which makes the calendar easy to build:
- US: the individual deadline for tax year 2025 was April 15, 2026. Form 4868 gives an automatic six-month extension, but "an extension of time to file is not an extension of time to pay" (IRS, 2026). That sentence alone is a ready client message for March.
- UK: Self Assessment paper returns are due 31 October 2026, online returns and payment 31 January 2027, and a second payment on account on 31 July (GOV.UK, 2026).
- Hungary: the 2026 personal income tax return and payment were due 20 May (NAV, 2026), and annual accounts for a calendar-year company are filed by the last day of the fifth month after the balance sheet date, so by 31 May (Számv. tv. 153. §).
| Period (US individual practice) | Office workload | Marketing work that fits |
|---|---|---|
| May to August | Lower | Website and service pages, articles, case-study consent, the year's plan |
| September to October | Extended returns due | Short, pre-written posts only |
| November to December | Year-end planning | Year-end newsletter, review request, referral ask |
| January to April | Peak | Pre-scheduled deadline reminders, review replies |
What should those quiet months produce? The AICPA survey above names managing tax law complexity as the top current issue for small firms (CPA Practice Advisor, 2026). If complexity is your pain, it is also your clients' pain, and explaining one change clearly is useful marketing. The accounting firm marketing plan template builds the year around this calendar, and the tax-change newsletter template turns one rule change into a client email.
Which marketing strategies need no client data at all?
Website pages, local search, articles on tax changes, social posts and fee explanations need no client data. They run on public rules and on your own expertise, which makes them the safest place to start with AI: the tool drafts from a published source, and a qualified person checks every figure and date before anything goes out.
The website. Prospects who arrive by referral check three things: what you do, for whom, and roughly what it costs. Many firm sites bury all three under a generic welcome. Accounting firm website examples judges real patterns by how clearly they show services, fees and the client types served.
Search. A small firm will not rank for "accountant" in a large city, and it does not need to. It can rank for specialist and local questions: a service plus a client type plus a place. Your Google Business Profile carries part of this work, and its rules are specific: the business name may not carry taglines or service descriptions, and virtual offices are not eligible (Google Business Profile Help). Accounting firm SEO covers the local and AI-search basics.
Social media. Confidential work makes accountants think they have nothing to post. In fact, deadlines, rule changes, common mistakes and how the office works are all public material. Accounting firm social media lists post types that show expertise with no client detail.
How much time does this take? There is no reliable benchmark for small firms, so use a frame instead of a guess:
Monthly marketing hours = (posts × minutes per post + articles × hours per article × 60 + newsletters × hours per newsletter × 60) ÷ 60
Example (illustrative, not a real case): A firm plans 8 posts at 20 minutes each (160 minutes), 1 article at 3 hours (180 minutes) and 1 newsletter at 2 hours (120 minutes). That is 460 minutes, a little under 8 hours a month. With AI drafting from public sources, the firm assumes the article drops to 2 hours, but the review time stays: a qualified person still checks every rule and date. Fill in your own minutes after the first month, not before.
How can you use the client list without breaking the rules?
Use the client list for tax information and tax-related offers, and keep the list itself out of AI tools. In the US, the IRC 7216 regulations set out what a tax preparer's client list may hold, and allow it to be used to send educational tax information or to offer more tax return preparation. Other uses need a closer look.
The rule in 26 CFR 301.7216-2(n) lets a preparer keep a list with client names, mailing and email addresses, phone numbers, entity classification and return form number, and use it for educational tax information or to solicit additional tax return preparation (Cornell LII). Note what is not on the list: income, refund amounts, deductions, anything from inside the return.
That shape fits a clean AI workflow. The body of the email is drafted from the public source, in any tool, with no names. The merge with names happens inside the email tool, which already holds the list. Nothing from the list travels into a prompt, and nothing from a return travels into the list.
Example (illustrative, not a real case): A two-partner CPA firm with 380 individual tax clients wants to remind them in March about the filing extension. The partner pastes the public IRS wording into an AI tool and asks for a 150-word plain-English email with one clear action line. She checks the dates against the IRS page. The email tool adds first names from the client list, which holds name, email and return form number only. The partner also wants to mention the firm's new payroll service in the same email; before doing that, she checks what the list may be used for, because that is a separate question.
The accounting firm leads article covers that second question and the solicitation rules around it. Outside the US, the same discipline holds for a different reason: the client list is still client information under your professional code, so it stays in the system it lives in.
What may an accounting firm say in its marketing?
Say what is true, specific and checkable, and nothing that promises outcomes. ICAEW's code bans exaggerated claims, disparaging references and unsubstantiated comparisons. The AICPA Code calls promotion misleading if it creates unjustified expectations of favourable results or quotes fees you expect to raise substantially. AI drafts tend to overstate, so this check matters.
The sources in full: the ICAEW Code R115.2 requires accountants to be honest and truthful in marketing, with no exaggerated claims, disparaging references or unsubstantiated comparisons, and R115.3 bans promotion that amounts to harassment (ICAEW Code, 2024). AICPA 1.600.010 treats as misleading any promotion that creates "false or unjustified expectations of favorable results", implies influence over a regulator, or lowballs fees the member expects to increase substantially (AICPA Code, 2024).
A short rewrite table for the first AI draft:
| AI-style draft line | Problem | Line that passes |
|---|---|---|
| "We'll cut your tax bill." | Expectation of a result | "We check every return for the deductions you are entitled to." |
| "Smarter than the big firms." | Unsubstantiated comparison | "A two-partner firm: you speak to the person preparing your return." |
| "We know people at the tax office." | Implied influence | "We represent clients in correspondence with the tax authority." |
| "Bookkeeping from a low monthly fee" (when most clients pay more) | Lowballed fee | The real fee range, with what drives it |
Put these rules into the prompt, so the draft starts closer to the line. Then read the draft against the table anyway, because the duty stays with the firm, not with the tool.
How do you set up AI in the office before marketing grows?
Write down three things before AI touches more than a draft: which tools are approved, which data may go into each, and who checks the output. In the US, a Written Information Security Plan is already required for tax and accounting firms of every size, so the AI rules belong inside it, not in a separate drawer.
The IRS guidance on the plan is practical: designate a qualified individual, assess risks, test safeguards, and choose service providers whose contracts require them to keep safeguards (IRS Publication 5708, 2024). Each of those maps onto an AI tool. Who approves a new tool? What could go wrong with it? Did anyone test what it stores? Do its terms commit it to protecting data?
Five office rules that cover most marketing use:
- Public AI tools get public material only: rules, published guidance, your own notes, your own website text.
- No client names, figures, documents or recognisable situations in a public tool, even "anonymised".
- Client data goes only into tools the firm has approved in writing, inside the security plan.
- Every AI draft is reviewed by a qualified person before it leaves the office, as ICAEW says, like a trainee's work.
- A mistake (client data pasted by accident) is reported to the responsible partner the same day.
Using AI in an accounting firm without exposing client data turns these into concrete office rules, including what to do about public share links. The procedures manual article shows where they sit among the firm's other written procedures, and the client onboarding checklist covers how to tell a new client, in writing, which outside tools will hold their data.
What does a realistic 90-day start look like?
Start with the strategies that touch no client data, in a quiet month, with a fixed weekly slot. A 90-day plan for a small firm can cover the review routine, the services and fees pages and one client newsletter, while the partners keep doing client work. Everything else waits until you can see what brings enquiries.
| Weeks | Work | Data level |
|---|---|---|
| 1–2 | Write the five AI office rules; add them to the security plan | None |
| 3–4 | Rewrite the services page and the fee explanation; check against the rewrite table | None |
| 5–6 | Fix the Google Business Profile name, categories and hours; set up one review request email | None / client list |
| 7–8 | Draft one tax-change newsletter from a public source; send through the email tool | Client list |
| 9–10 | Add "How did you hear about us?" to the intake form | None |
| 11–12 | Review: enquiries by source, reviews received, time spent | None |
Weekly time is the constraint, so put a number on it before you start:
Weekly slot = (total plan hours ÷ number of weeks), rounded up to a fixed calendar block
Example (illustrative, not a real case): A six-person practice estimates 30 hours for the 12-week plan: 4 for the rules, 8 for the two pages, 4 for the profile and review email, 6 for the newsletter, 2 for the intake form and 6 for reviews along the way. That is 2.5 hours a week; the managing partner books a 3-hour block every Thursday morning in May, June and July, and a bookkeeper owns the review email.
The articles in this set, in the order most firms need them:
- Accounting firm marketing plan template
- Using AI without exposing client data
- Accounting firm website examples
- Accounting firm SEO
- Accounting firm social media
- The tax-change newsletter template
- Accounting firm leads
- Accounting firm marketing ideas
- Client onboarding checklist
- Procedures manual for a small accounting firm
If you want your AI tools to start from your firm's own written facts, the services, the fee rules and the data rules above, instead of whatever was pasted last, the Business Brain Starter Kit is built for that. Get notified when it launches.
FAQ
Can an accounting firm use AI for marketing without breaching confidentiality?
Yes, if client data stays out of public tools. Most marketing work, such as website pages, articles on tax changes, social posts and fee explanations, needs no client data at all. ICAEW warns that client data in AI inputs is very likely a breach, so client files go only into tools the firm has approved, and every draft is reviewed by a qualified person.
Which marketing strategies should a small accounting firm start with?
Start with the ones that touch no client data and support referrals: a clear services and fees page, a correct Google Business Profile, a steady review request routine and one newsletter on a tax change. These fit into a fixed weekly slot during the quiet months, and they help the prospect who has been given your name and looks you up.
When should an accounting firm do its marketing?
Do the building work in the quiet months after the main filing deadline, for a US practice roughly May to August, and keep the peak season for pre-written reminders and review replies. UK practices plan around the 31 January Self Assessment deadline. A plan that ignores the filing calendar tends to be dropped as soon as the busy season starts.
Can I use my tax client list to send a newsletter?
In the US, the IRC 7216 regulations allow a tax preparer to keep a client list with names, contact details, entity classification and return form number, and to use it for educational tax information or to offer more tax return preparation. Using it to sell other services is a separate question. Keep the list inside your email tool and out of any AI prompt.
Can we publish client testimonials or case studies?
Only with the client's specific consent. The AICPA Code does not allow disclosure of confidential client information without the specific consent of the client, and a case study usually contains exactly that. Get the consent in writing, show the client the final text, and draft it by hand or in a tool the firm has approved for client data, not in a public AI tool.
Does a small accounting firm need a written AI policy?
It needs written rules, and in the US they fit inside the Written Information Security Plan that tax and accounting firms of every size must already keep. Five rules cover most marketing use: public material only in public tools, no client details, approved tools for client data, a qualified reviewer for every draft, and same-day reporting of mistakes.
Sources
- ICAEW Regulatory News, "Don't let AI tools land you in hot water", 2024-02-12. https://www.icaew.com/regulation/regulatory-news/regulatory-news-2024-02/dont-let-ai-tools-land-you-in-hot-water
- AICPA, "AICPA Code of Professional Conduct" (copy hosted by the Minnesota Society of CPAs), 2024. https://www.mncpa.org/mncpa/media/Documents-Shared/about-mncpa-info/AICPA-code-of-conduct.pdf
- IRS / Security Summit, "Publication 5708: Creating a Written Information Security Plan for your Tax & Accounting Practice", 2024. https://www.irs.gov/pub/irs-pdf/p5708.pdf
- Karbon, "State of AI in Accounting 2026 Report" (vendor survey), 2026-01-20. https://karbonhq.com/resources/state-of-ai-accounting-2026/
- CPA Practice Advisor, "AICPA Survey Cites Change Management for Technology and AI as Top Long-Term Issue Facing Accounting Firms", 2026-06-23. https://www.cpapracticeadvisor.com/?p=185547
- ICAEW, "Professional behaviour in the Code of Ethics", 2024. https://icaew.com/-/media/corporate/files/about-icaew/who-we-are/governance-and-committees/professional-standards-committee/professional-behaviour-in-the-code-of-ethics-mar-2024.ashx
- Cornell LII, "26 CFR § 301.7216-2 — Permissible disclosures or uses without consent of the taxpayer", accessed 2026. https://www.law.cornell.edu/cfr/text/26/301.7216-2
- ACCA Accountancy Business (Emily Garnham), "Encourage client recommendations", 2025-05. https://abmagazine.accaglobal.com/global/articles/2025/may/practice/encourage-client-recommendations.html
- Journal of Accountancy (Anita Dennis), "Real-life ways small firms use AI", 2026-08-01. https://www.journalofaccountancy.com/issues/2026/aug/real-life-ways-small-firms-use-ai.html
- BrightLocal, "Local Consumer Review Survey 2026", 2026-02-11. https://www.brightlocal.com/research/local-consumer-review-survey/
- Google Business Profile Help, "Prohibited & restricted content", accessed 2026. https://support.google.com/business/answer/7400114
- Google Business Profile Help, "Guidelines for representing your business on Google", accessed 2026. https://support.google.com/business/answer/3038177
- IRS, "When to file", 2026. https://www.irs.gov/filing/individuals/when-to-file
- GOV.UK (HMRC), "Self Assessment tax returns: deadlines", 2026. https://www.gov.uk/self-assessment-tax-returns/deadlines
- NAV Sajtószoba, "Indul a 2026-os szja-szezon", 2026-01-19. https://nav.gov.hu/sajtoszoba/hirek/Indul_a_2026-os_szja-szezon
- Jogtár, "2000. évi C. törvény a számvitelről", 2000, consolidated text accessed 2026. https://net.jogtar.hu/jogszabaly?docid=a0000100.tv