You Can't Hire Your Way Out of This One The accountant shortage is structural, not a dip you can out-recruit. A practitioner's case that capacity is a keystroke problem — and why 2026 is the first year you can fix it without adding to payroll.

Every firm owner I talk to has the same plan for the capacity problem, and it's the same plan they had five years ago: hire. Post the opening, screen the resumes, get another preparer in the seats before the wall hits. It's a reasonable plan. It's also quietly stopped working, and the reason it stopped working is the thing worth writing down — because once you see it, the whole way you think about a September like this one changes.

Here's the uncomfortable version, up front: the accountant shortage is not a temporary dip you can wait out or out-recruit. It's structural. And a structural problem does not have a hiring solution.

The math that no job posting fixes

Start with the supply side, because it's the part everyone underestimates. The Bureau of Labor Statistics projects roughly 124,200 openings for accountants and auditors every year through 2034 — a steady, enormous demand for people. Now hold that next to where the people come from. The number of candidates sitting for the CPA exam has been in a sustained decline for most of a decade, and the active CPA workforce is aging, with more of it concentrated in the later-career stretch every year. Demand is flat-to-up. The pipeline that's supposed to feed it has been narrowing since before most firms noticed.

There is a hopeful signal in the data, and I'll give it its due: the AICPA reported that accounting enrollment at four-year colleges rose 8.9% in spring 2026, the third straight year of growth. That's real, and it matters for 2030. It does nothing for your 2026. A student who enrolled this spring is not reviewing a partnership return this September, and won't be for years. The people you need for the wall in front of you already exist or they don't.

So picture what actually happens when four thousand firms all run the same hiring plan into the same shrinking pool at the same time. You don't get four thousand new hires. You get a bidding war. Robert Half's most recent research on the profession found that 61% of finance and accounting leaders say skilled professionals are harder to find than they were a year ago, and 87% now routinely pay a premium for candidates with the skills they need. That's the sound of a lot of firms discovering the same thing at once: the lever they've always pulled has gone stiff, and pulling harder just costs more.

You can't hire your way out of a shortage. By definition, there isn't enough to hire.

Where your September actually goes

If more people isn't the answer, the only place left to look is the work itself — specifically, at how much of it needs a person at all.

So look. Pull one extended 1120-S off your own stack this week and watch what your preparer does with it, minute by minute. Somewhere in there is the part clients pay for: the read of the situation, the judgment call on an aggressive position, the thing that only a licensed brain can do. Time it honestly and it's a surprisingly thin slice of the afternoon.

The rest — the bulk of it — is assembly. It's chasing the document that never came. It's a K-1 that arrived as a photograph. It's a number that was typed once by a payroll company, printed onto a form, and is now being typed a second time by someone at your firm into a field on a screen. Multiply that by every box, every form, every return stacked against the September 15 deadline, and you have found your capacity problem. It was never a shortage of judgment. It's a surplus of transcription.

That distinction is the whole game, so let me put it as plainly as I can: your firm is not short on accountants. It's short on accountant-hours being spent on things only an accountant can do. Those are different problems, and only one of them has a solution you can actually reach this year.

The lever you actually control

Here's the reframe. Capacity is not a number of people. Capacity is returns-per-person, and returns-per-person is decided almost entirely by how much non-judgment work sits on top of each preparer.

You cannot change the size of the labor pool. It's bigger than your firm and it's moving the wrong way. But you have total control over how many keystrokes stand between a document arriving and a return being ready. That number is a choice. It's been an expensive choice for a long time, because the only way to lower it used to be to hire someone to absorb it — which brings us right back to the pool that isn't there.

That's the specific thing that has changed, and it's why 2026 is a genuinely different year than 2021 even though the shortage headlines read the same. For the first time, the transcription layer — the reading, the keying, the moving of numbers from a PDF into the software — can be lifted off a preparer without adding a person to the payroll. The work still gets done. It just stops being done by someone whose actual value is their judgment.

Take the keystrokes, keep the judgment. That single sentence is, to me, the entire opportunity of this moment, and it is worth being precise about both halves of it.

The honest version of "let AI do it"

Because there's a wrong way to hear that sentence, and the profession is about to spend a few years learning it the hard way.

The wrong way is "let the AI do the return." A machine can read a W-2 beautifully and hand you a clean, well-formatted number that is off by a digit — and it will never once tell you it wasn't sure. Confidence is not accuracy, and on a return that carries a human's signature, a confidently wrong number is worse than a slow right one. Anyone selling you full automation of a tax return is selling you the liability along with it, and pretending the signature line doesn't exist.

The right way is narrower and far more useful: automate the part that was never judgment, and make the part that is judgment faster to do than it was to type. Let the machine read and place the data. Then show every field to a licensed human against its source, and trust nothing until that person confirms it. The preparer stops being a typist and goes back to being a preparer — checking, catching, deciding — which is both a better use of a scarce person and, not incidentally, a better job. People did not spend years earning a credential to spend September re-keying box 1.

This is the design principle we built Sophicor on, and I'll be straight about why it's built the way it is rather than the flashier way. We deliberately did not build the "approve all" button. A screen of numbers with no document beside them doesn't get verified; it gets rubber-stamped. So the review puts the source in front of the person, field by field, so that confirming actually means looking. It's a hair slower per field than a blind approve. That's the point. The speed lives in the reading and the keying; the two seconds of human certainty is the part we refuse to automate away.

What this means for a firm in front of the September wall

There's a practical version of all this, and the calendar makes it urgent. September 15 is the deadline for calendar-year partnerships and S-corporations that timely extended — Forms 1065 and 1120-S, with no second extension behind them — and the third-quarter estimated tax installment for individuals lands the same day. Both are on irs.gov if you want to check me.

You will not hire anyone between now and then. That door is closed for this cycle. But you can still change the mix of what your existing people are spending their remaining hours on. Pull your open list this week and sort it — not by client, not by size, but by what is actually blocking each return. My bet, having done this exercise in more than one firm, is that most of what's left isn't waiting on judgment. It's waiting on assembly: documents to arrive, numbers to be moved, moved numbers to be checked. That pile is the part a machine can take. The pile underneath it — the reason clients pay you — is the part it can't, and shouldn't.

Every profession that automated lost the transcription first and kept the judgment. Medicine didn't lose the physician to software; it lost the dictation backlog to speech recognition. Law didn't lose the lawyer; it lost the room of associates reading boxes for a keyword. Tax has one wrinkle the others didn't — because we sign the return, we've always billed the freight and the judgment as a single thing, and it's made both feel un-automatable. They aren't the same thing. Pulling them apart is the opportunity, not the threat.

The shortage isn't a storm you wait out. It's the new weather. Firms that keep trying to hire their way to capacity will spend the next decade paying more for fewer people. Firms that get the transcription off their preparers will find they had more capacity than they thought — it was just buried under keystrokes. The pool of accountants is not going to grow in time to save your September. The number of things you're asking each accountant to do by hand still can.

Sources referenced: U.S. Bureau of Labor Statistics Occupational Outlook (accountants and auditors); 2025 AICPA/NASBA Trends Report; AICPA spring 2026 enrollment data; Robert Half, Demand for Skilled Talent; IRS tax calendar and estimated-tax guidance at irs.gov.