Ask ten accounting-firm owners what their margin on payroll is and you will get eight shrugs and two guesses. Payroll is the service most firms cannot price clearly. That is also why it is the service most firms quietly lose money on.
The problem is rarely the underlying cost of payroll. The problem is structural: a flat per-client fee asked to cover wildly different amounts of work, exception work absorbed without billing, and a labor pattern nobody is tracking. None of that is a payroll problem. It is a pricing-and-operations problem with payroll on the bill.
The two prices you have to know
There are two numbers in payroll, and most firms can only quote one. The first is what you pay your platform per client — for Payrollix, $12 plus $2 per employee beyond five; for legacy platforms, a per-company base plus a per-employee fee. The second is what you charge the client. The difference between those two numbers, minus your labor, is your margin per client.
If you do not know that number for your top ten clients, you do not know your payroll P&L. You have an average and a hope. Most firms with this gap have a small handful of clients quietly subsidizing the rest and a few clients quietly bleeding the firm — but nobody knows which is which.
What firms actually charge
Industry pricing for monthly payroll service runs roughly $50 to $150 per client per month, with the spread driven by complexity. Single-state clients with five employees and a stable schedule sit at the low end. Multi-state clients with thirty employees, quarterly bonuses, and PTO accruals sit at the high end. A few firms still charge a single flat fee for everyone.
Flat pricing per client is what kills margin on the complex ones. The 5-employee single-state client priced at $75 is fine; the 30-employee multi-state client priced at the same $75 is a charity case. The fix is not necessarily a complex tiered schedule — it can be a base plus per-employee, mirroring what platforms do. The principle is that complexity should be priced, not absorbed.
Where the margin quietly goes
The leaks are predictable, and they are almost always unbilled work that should be billed. Tax-notice resolution is the classic one — a state issues a notice, the firm spends three hours fixing it, no invoice is sent. W-2c corrections after year-end are another. Onboarding a new client takes real hours, and most firms include it free. High-touch clients who change pay rates twice a month and submit hours three days late get charged the same as the low-touch client who submits the same approved file at the same time every cycle.
Each leak feels small in isolation. Stacked across a 100-client book over a year, the lost time runs into hundreds of hours — at any reasonable hourly rate, that is the firm's payroll margin walking out the door.
The math changes at scale
At 50 clients, fixed costs dominate — your time learning the platform, building templates, setting up integrations. Per-client margin looks thin. At 200, the fixed costs amortize and process maturity starts to drop the labor per client. At 500, software cost is linear but labor per client can be a fraction of what it was at 50 — if you have built process. If you have been running 50 clients manually and your plan for 500 is more hands, scale will make you less profitable, not more.
This is the bend in the curve most firms get wrong. They expect linear improvement and get linear pain. The improvement only shows up after process and automation are in place.
Hire for exceptions, automate processing
The instinctive move when payroll volume climbs is to hire someone to process. It is also the move that crushes margins, because processing is the part that scales with automation. The right order is the reverse: automate processing first, hire for exceptions when the exception load outruns one person.
The trigger to hire is not run count climbing. It is response time slipping — when a tax-notice email or a payroll-blocking question takes two days to answer instead of two hours. That is the signal. Hiring earlier than that is hiring against your margin; hiring later than that is hiring against your clients.
The clients you should not take
Some clients are margin traps and no amount of pricing fixes them. The one or two-employee client where fixed per-client effort cannot be covered by a tiny fee. The client launching in seven states their first year, before your process handles multi-state cleanly. The client who refuses to use the self-service portal and generates a steady stream of email and phone calls.
Saying no to those clients early is one of the most underrated ways firms protect their payroll margin. It is also one of the hardest, because every client looks billable individually. The discipline is to look at the book as a portfolio: which clients pay for your time, and which clients spend your time on someone else's behalf.
See what flat per-client pricing looks like
Start your 14-day free trial. No credit card required.
Start Free Trial