Blue J and CPA.com surveyed more than 1,000 tax professionals and published the results on 8 June 2026. The finding that travelled furthest was the adoption curve: 60% of respondents now use AI for tax research at least weekly, up from 33% a year earlier. Eighty-four per cent agree it saves time, and 69% anticipate shifting towards value-based, hybrid or fixed-fee billing as a result.
The survey also recorded where the time goes. Advisory projects at 44%, tax planning at 40%, compliance research at 39%, document analysis at 36%, drafting at 35%. Every one of those is production work. Not one of them is the work of getting the next client.
That is the gap worth sitting with. Firms have spent a year automating the delivery of the service and almost no time automating the thing that decides whether there is a service to deliver. Meanwhile 69% expect to change how they bill, which only works if there is a pipeline of clients who want the thing being billed differently.
What is marketing automation for an accounting firm?
It is the set of triggered, repeatable client-facing sequences a firm would otherwise do by hand or not at all: a follow-up when an enquiry comes in, a deadline reminder tied to a client's actual filing dates, a nurture sequence for a prospect who is not ready, and an internal alert when a client does something that signals an advisory need.
It is not a newsletter, and the distinction matters. A newsletter goes to everybody on the same day regardless of what they need. Automation sends one thing to one client because something happened. The second reliably produces conversations and the first mostly produces unsubscribes.
Why does the delivery side get automated first?
Because the pain is immediate and the owner of the problem is the person making the decision. A partner who spends four hours on tax research feels those four hours. A partner who never learns that a prospect visited the pricing page three times and then went elsewhere feels nothing, because the loss is invisible.
Invisible losses do not get fixed. That is the whole mechanism, and it explains why firms with sophisticated research tooling still have an enquiry sitting unanswered in a shared inbox for two days.
What should a firm automate first?
Enquiry response, and nothing else until that is solid. An acknowledgement within minutes rather than hours, a routing rule so it reaches the right person, and one follow-up if nobody replies. This is the cheapest automation a firm can build and it recovers more revenue than anything else on the list.
Speed is doing most of the work here. A prospect who contacted three firms is disproportionately likely to engage the one that replied first, and that has almost nothing to do with the quality of the reply. Firms consistently underrate this because the lost prospects never tell them.
What comes second?
The deadline calendar, driven by each client's real dates rather than a generic schedule. Extended returns, estimated payments, information requests, year-end planning prompts. The data to do this already exists inside the practice management system, which is why it is usually the fastest sequence to build once the first one works.
The reason it pays is that it converts an administrative touch into an advisory opening. A reminder that arrives with a specific, relevant observation about that client's position is a conversation. A generic deadline reminder is noise, and clients treat the two very differently.
Where does AI actually help, as opposed to automation?
Automation decides when something is sent. AI helps with what is in it. Drafting a first version of a client update, summarising what changed in a piece of guidance into client language, or turning a partner's three-sentence explanation into a page are all real time savings, and they are the same drafting and document work the survey already found 35% and 36% of firms doing.
The firms getting value from this are applying capability they already have to a function they had not considered. They are not buying anything new. They are pointing the same tool at the marketing problem, which is why the barrier is habit rather than budget.
What should never be automated?
The judgement about what a specific client should do, and any message that implies a partner has personally reviewed something when nobody has. Both are reputational risks that outweigh any efficiency gain, and clients detect the second faster than firms expect.
There is also a quieter failure. A firm that automates the warmth out of a relationship ends up with a technically efficient process that clients like less than the inefficient one it replaced. The test is whether the automation removes a delay or removes a person, and only the first is an improvement.
How does this connect to being found in the first place?
Automation improves what happens after somebody arrives. It does nothing about whether anybody arrives, and a firm with an excellent follow-up sequence and no visibility has built a very efficient way of processing three enquiries a month.
The two have to be built together. A steady publishing habit around the guidance clients care about creates the arrivals, and the automation makes sure none of them are dropped. We set out how the first half is built in our guide to digital marketing for accounting firms, and the AICPA survey we covered in AI topping CPA firms' five-year issue list shows how firms themselves rank the problem.
What does the billing shift have to do with any of this?
Everything. The 69% expecting to move to value-based, hybrid or fixed-fee billing are describing a change that only works if clients understand what they are buying. Hourly billing explains itself. Value pricing has to be explained, repeatedly, before the invoice arrives.
That explanation is a marketing function whether a firm calls it one or not. Sequences that set expectations, describe the advisory work in client language, and surface the value before the bill are the infrastructure that makes the billing change survivable. Firms that change the pricing without building that explanation get a difficult year.
How much of this can a small firm realistically build?
More than the software vendors imply and less than the case studies suggest. A five-partner firm can build fast enquiry response, a deadline sequence and one nurture track inside a quarter using tools it probably already pays for. What it cannot do is run fifteen parallel campaigns, and it should not try.
Three sequences that run reliably beat fifteen that half work. The failure mode for small firms is not being too cautious, it is building an elaborate system nobody maintains, which decays into a source of embarrassing messages sent to the wrong clients.
What does a working nurture track look like?
Six to eight messages over three months aimed at a prospect who enquired, had a conversation, and did not proceed. Each one carries a single useful idea relevant to their situation rather than a description of the firm. The last one asks plainly whether the timing has changed.
Firms tend to abandon these after two messages because nobody replies to the first two. That is the normal shape. The replies cluster late, usually when something changes on the client's side, and a sequence that stops at message two is a sequence that stops immediately before it works.
The content problem is smaller than it appears. Most firms already have the material in answers partners have written to clients over the years. Collecting six of those and editing them into standalone notes is an afternoon, not a project.
How do you tell whether it is working?
Track replies and booked conversations, not opens. Open rates have been unreliable for years and they flatter sequences that are producing nothing. A track that generates two conversations a month from a list of two hundred lapsed enquiries is doing real work even if the open rate looks unremarkable.
Review it once a quarter and cut whatever produced nothing rather than adding to it. Automation decays quietly, and the firms that keep it useful are the ones that delete more than they add.
What does the national picture suggest about timing?
That the window where this is a differentiator is open but not indefinite. Adoption of AI in delivery nearly doubled in a single year, and client-facing use tends to follow delivery use with a lag. We looked at where US firms actually apply AI in our analysis of the census data on AI use in sales and marketing, and the direction is consistent.
A firm building this in late 2026 is early among small practices and ordinary among large ones. A firm building it in late 2027 is neither.
Where to start this week
Measure the response time on your last twenty enquiries. Not the average, the worst one. That single number tells you more about your pipeline than any report, and it is usually bad enough to make the first automation an easy decision.
Then pick the one sequence with the clearest owner and build only that. The most common reason these projects fail is scope, not capability, and a firm with one working sequence in six weeks is in a far better position than one with a comprehensive plan and nothing running.
Our AI automation work for US firms covers what we build and what we advise against, the US accounting and CPA firm marketing page sets out how it fits with the visibility side, and the AI automation service describes the sequences themselves. If you want a view on which one to build first, a short conversation is usually enough to tell.
FAQs
How many accounting firms use AI weekly?
A Blue J and CPA.com survey of more than 1,000 tax professionals, published on 8 June 2026, found 60% now use AI for tax research at least weekly, up from 33% a year earlier. Eighty-four per cent agree it saves time, and 69% anticipate shifting towards value-based, hybrid or fixed-fee billing.
What is marketing automation for an accounting firm?
Triggered, repeatable client-facing sequences: fast enquiry response, deadline reminders tied to each client's real filing dates, a nurture track for prospects who are not ready, and internal alerts when a client signals an advisory need. It differs from a newsletter because it is sent in response to something, not on a schedule.
What should a CPA firm automate first?
Enquiry response, before anything else. An acknowledgement within minutes, a routing rule so it reaches the right person, and one follow-up if nobody replies. Prospects who contacted several firms disproportionately engage whoever replied first, which makes this the cheapest automation with the largest recovery.
What should accounting firms never automate?
Judgement about what a specific client should do, and any message implying a partner has personally reviewed something when nobody has. Both carry reputational risk that outweighs the efficiency gain. The useful test is whether an automation removes a delay or removes a person, because only the first is an improvement.
Can a small accounting firm build this without a large budget?
A five-partner firm can build fast enquiry response, a deadline sequence and one nurture track inside a quarter, usually with tools it already pays for. What it cannot sustain is fifteen parallel campaigns. Three sequences that run reliably are worth more than an elaborate system nobody maintains.
How does marketing automation relate to value-based billing?
Closely. Hourly billing explains itself and value pricing does not, so it has to be explained repeatedly before the invoice arrives. Sequences that set expectations and describe advisory work in client language are the infrastructure that makes a billing change survivable rather than contentious.