The Person Who Saves Your October Is the One Who Leaves in November

Today is September 15. Your extended partnership and S-corp returns are out the door, or they aren't, and either way the room is quiet in the particular way it gets when everyone is tired and nobody wants to say so.

There are thirty days left until October 15. Most of what gets written about accounting firm retention is written in November — after the season, when the firm sits down to ask why someone gave notice. This is not that article. This is about the thirty days you still have, because the decision you're worried about is being made right now, in the middle of the crunch, by someone who will be perfectly pleasant about it until the day they aren't.

The good news first, because it changes the problem

Turnover across accounting firms fell to 11.8% in 2025 — the lowest level in more than 20 years of Inside Public Accounting's benchmarking data, down from a 15.9% peak in 2022. And it skews hard by size: firms under $5 million in revenue reported 8.1% turnover, against 14.6% for firms above $75 million (Inside Public Accounting).

If you run a mid-market firm, the headline "talent crisis" narrative is not your narrative. You are not facing an exodus.

You are facing something more specific and, in a firm of 22 people, considerably worse: the loss of one or two individuals. And it is almost never a random one or two.

It's the same profile every time, and that's a design flaw

Ask any managing partner who left after the last crunch and you'll get a name, not a demographic. Then ask what that person was doing in the last six weeks of the season, and the answer is usually some version of: whatever needed doing.

That isn't a coincidence or a culture problem. It's an allocation defect. As one workforce-management analysis of the profession puts it, the most dependable people absorb the pressure because leaders know they'll get the work done — strong performance is effectively rewarded with more work, while other capacity sits unused. And it gets worse specifically in firms where allocation runs on spreadsheets, email requests, or a manager's informal sense of who's free: without firm-wide visibility, leaders overload their best people unintentionally (HiBob).

That last word is the important one. Nobody decided to do this. The overflow went to the person most likely to catch it, forty separate times, and nobody was holding the running total.

Firms manage the average. People quit over the distribution.

Here is the gap that keeps costing firms good seniors.

Leadership thinks in aggregates — total hours, realization, whether the team is "slammed." The AICPA's 2025 Trends report found work-life balance was the number one reason accounting professionals left their firms (AICPA & CIMA), and firms reasonably conclude the lever is hours. So they try to reduce hours, or apologize for them, or promise next year will be different.

But two people working the same 55-hour week can have completely different experiences of it. One took on a stretch engagement she asked for. The other watched three files land on his desk on a Thursday afternoon because he was the one who picks up the phone. Same hours. Only one of them is updating a résumé.

What burns people out is not volume. It's unchosen volume, arriving without acknowledgment, in a pattern they can see and you apparently can't. The hours are what they'll say in the exit interview, because "the workload was unfair" is a harder sentence to say to someone you like.

The timing problem nobody fixes

Now the part that makes all of this preventable, and makes most firms miss it anyway.

Post-busy-season departures cluster in the months after the crunch (Karbon) — which is also exactly when firms schedule their retention response. The stay interview is in November. The comp conversation is in November. The "how was your season?" lunch is in November.

The person decided in early October.

By November you are not preventing a decision. You are negotiating with one that's already made, usually with money, which is the least effective tool available for a problem that was never about money. The window where the experience is still being written is the window you're standing in right now.

What you can actually do in thirty days

Not a program. You don't have time for a program, and a program in week four of extension season would land as theater anyway. Three things, none of which take longer than a morning.

1. Count the open items per person. Just count them.

You are probably not measuring workload distribution at all. Most firms aren't — that's the finding above. You don't need a utilization system to start; you need a count of open, assigned work by person, today.

Do it crudely if you have to. A list per name, open items only, sorted oldest first. The number itself matters less than the spread — and you don't need a threshold to read it, because the outlier is obvious the moment the counts sit side by side. Whoever is well clear of everyone else is your November resignation, and you found them with thirty days left instead of thirty days late.

2. Rebalance the last stretch, visibly

Not everything — the last thirty days is the wrong time for a wholesale re-assignment, and moving a complex file mid-review costs more than it saves. Move the things with the lowest transfer cost: the waiting-on-client follow-ups, the routine 1040s with clean documents, the administrative chase work.

Then do the part firms skip: say out loud that you moved it, and why. "I looked at who's carrying what, you're carrying too much, I've moved four things." That sentence is the entire intervention. It converts an invisible imbalance into a visible correction, and it tells the person the thing they most need to know — that someone is watching the distribution, not just the deadline.

3. Book the debrief now, for October 16

One recurring task, one calendar slot, set today. Not "sometime after the season." A date. Because the other reason November conversations fail is that by then everyone has had two weeks to reconstruct the season into something more tolerable than it was, and the specific, fixable details are gone.

Why one departure costs more than you've priced it at

If you need the business case rather than the human one, it's brutal, and it's mostly about timing rather than dollars.

Finance roles requiring a CPA now take an average of 73 days to fill — 41% longer than comparable non-CPA positions (Talentfoot). One 2026 estimate puts the all-in cost of replacing an accountant at $58,000 or more once recruiting, lost productivity and compliance risk are counted (Lauber Business Partners).

Now put 73 days on a calendar starting October 16. That's an average, not a promise — but on the average you are interviewing through the holidays and onboarding an unfamiliar person around the turn of the year.

That's the real cost. Not the fee. You don't lose a salary — you lose a trained senior and receive, in exchange, a stranger who needs supervision during the six weeks when supervision is the scarcest thing in the building. A resignation on October 16 is a tax on next busy season, paid in the only currency that's actually short.

Where a task system fits — and where it doesn't

Be clear about what a shared task system does here, because overselling it is how firms end up with another tool nobody opens.

Toodledo will not tell you how many hours anyone worked. It is not a time-tracking or utilization platform, and it won't compute a realization rate. What it will do is hold every piece of client work in a shared folder — the folder is the sharing boundary, so the whole team sees the same board — with an owner attached, so that "who is carrying what" becomes a view you can pull in a minute instead of a reconstruction you do in an exit interview.

In practice, for a firm of 10–50:

• One shared folder per team or engagement group, so the work is visible to everyone who assigns it, not just to the person holding it.
• Contexts or tags to mark ownership and stage. Toodledo has no custom fields, and it doesn't need them for this — a tag per person and a tag per stage gives you the two cuts that matter.
• A saved search per person, sorted oldest first. That's your distribution report. Run it Monday morning; the spread is visible in ten seconds.
• A recurring weekly task called "Check the spread." Fifteen minutes. The whole point is that it survives October, when nobody has the attention to remember it voluntarily.
• Subtasks on the heavy files, so when you do move something, the next person inherits the checklist rather than a title and a shrug.

The system isn't the intervention. Looking is the intervention. The system just means looking costs you a minute instead of an afternoon you don't have.

See Toodledo plans for teams →

Frequently asked questions

Isn't it too late to change anything thirty days out?

Too late to change the workload; not too late to change the experience of it. The resignation decision is driven far more by whether the imbalance was seen and acknowledged than by whether it was fully corrected. Moving four low-transfer-cost items and naming why you moved them is a thirty-minute action with disproportionate effect.

How do I know who's at risk without asking a leading question?

Look at the distribution before you look at people. The person carrying materially more open work than the median is your answer, and you got it without a conversation that signals you're worried. Then ask that person a specific question — "you've got eleven open files, who can take two?" — rather than a general one about how they're doing.

Our turnover has been fine. Is this really a priority?

Probably your rate is fine — IPA's lowest band (firms under $5 million in revenue) came in at 8.1%, and even the largest firms only reached 14.6%. The risk isn't the rate. It's that in a 22-person firm the departure of one senior is a 4.5% turnover rate, and a genuinely bad quarter. Rates describe populations; firms this size live on individuals.

Should we just hire ahead of it?

At an average 73 days to fill, hiring ahead of an October departure means starting in July, which means predicting it in July. Watching the distribution is considerably cheaper than forecasting resignations, and it's the input you already have.

What about the people who like the crunch?

They exist, and they're not the risk — chosen intensity reads completely differently from assigned intensity. This is exactly why the distribution matters more than the hours: it helps you tell the two apart, which an hours report never will.

Sources

• Inside Public Accounting, "IPA Data Dive: Turnover Trends Across the Profession" — 11.8% turnover in 2025, lowest in 20+ years; 15.9% peak in 2022; 8.1% (under $5M) vs 14.6% (over $75M).
• Ohio Society of CPAs, "Accounting firm turnover eases after post-pandemic spike".
• AICPA & CIMA, 2025 Trends Report — work-life balance as the leading reason professionals left firms.
• HiBob, "Why accounting firms struggle with workload management" — top performers unintentionally overloaded where allocation runs on spreadsheets, email and informal manager knowledge.
• Talentfoot, "How the CPA Shortage Is Extending Time-to-Fill for Accounting Roles" — 73-day average time-to-fill, 41% longer than non-CPA roles.
• Lauber Business Partners, "The Hidden Costs of Accounting Turnover in 2026" — all-in replacement cost of $58,000 or more.
• Karbon Magazine, "Why your best accountants quit after busy season" — post-season departure clustering.

Turnover benchmarks, replacement-cost estimates and time-to-fill figures vary by source, region and firm size. Figures above are cited as published; confirm against your own data before using them in a partner-level decision.