How to choose payroll software: the Hardest-Constraint Test

Payroll guides list features. Payroll failures come from a constraint nobody checked.

Here is the short version of how to choose payroll software: find your hardest constraint first, eliminate every tool that cannot clear it, and only then compare what is left on price and support. Most teams do the reverse, compare feature grids for a week, and discover the dealbreaker after the contract is signed.

This guide gives that process a name, the Hardest-Constraint Test, and walks it end to end. It draws on the payroll tools we have run real pay cycles on, the real prices we track with dates and sources, and the wider map of what HR work is worth automating in the first place. One scope note before we start: the tax specifics below are US specifics, IRS deposits, 1099s and W-2s. The constraint method travels to any country; the penalty numbers do not.

Payroll software is a liability decision first

A missed marketing email costs you nothing. A missed pay run costs you your team’s trust, and a missed tax deposit costs you real money: the IRS failure-to-deposit penalty runs from 2% of the deposit for a slip of a few days, through 5% and 10% tiers as the delay stretches, to 15% once the IRS has sent a notice and more than ten days pass.

That is why the first question about any payroll tool is not what it can do. It is what it executes on your behalf, because the legal liability for unpaid payroll taxes stays with you, the employer, whatever the vendor handles. Payroll software sits on a ladder: some tools only calculate what you owe, some also deposit the taxes, and some file the returns. We broke this down as the Calculate–Deposit–File ladder in our free payroll software round-up, and it applies to paid tools just as much: two products at the same price can sit on different rungs, and the gap between rungs is where penalties live.

So before any comparison: write down which rung you are buying. If the tool does not deposit and file, you are the payroll department. That can be a rational choice at three employees. It is rarely one at fifteen. The glossary keeps a plain-English definition of payroll software built on this same distinction, short enough to send to a colleague.

The Hardest-Constraint Test

Feature grids fail small teams because they weight everything equally, and payroll constraints are not equal. One unmatched constraint sinks the tool; forty matched features do not save it. The test has three steps:

Step 1 — name your hardest constraint. Read the six below and pick the one that is hardest for your team, not the one that appears first. Most teams have one that bites hardest.

Step 2 — eliminate, don’t compare. Every tool that cannot clear that constraint leaves the list, whatever its price or reviews. Do not keep a “great deal” that half-clears it. Half-cleared constraints become support tickets with deadlines attached. And if several of the six apply to you, the hardest one does the first cut, not the only cut: a survivor must still clear every constraint you actually carry.

Step 3 — compare the survivors on three numbers. Total monthly cost at your real headcount, the year-two price, and the hours it takes to reach a human on your plan’s support channel. Those three, compared at the same rung of the Calculate–Deposit–File ladder, decide between finalists; a price that files and a price that does not are prices for different products.

The six constraints that decide the choice

These six are the blockers that decide real choices most often. If your business has another non-negotiable, accounting-software integration, say, or built-in time tracking for an hourly crew, treat it the same way: it eliminates, it does not merely score points.

1. Contractors, or people abroad

If part of your team is 1099 contractors, check whether the tool prices them separately: some vendors charge a lower per-contractor rate, and on published 2026 prices Gusto runs a contractor-only plan at $35 a month, though only for teams with no W-2 employees at all; hire one employee and you are back on a standard base-plus-per-person plan. Paying people abroad splits into two different problems: overseas contractors run through a contractor-payments platform, while employing someone abroad runs through an employer of record, whose headline fee (from $599 per employee per month on Deel’s published pricing, captured July 2026, before salary and employer taxes) belongs to a different budget conversation entirely — a distinction we unpack in our Gusto vs Deel comparison. We have run pay cycles on both sides of that line; see Deel for the cross-border side.

2. Employees in more than one state

Every state where an employee works, and one remote hire is enough to add a state, means state registrations, unemployment insurance accounts, and sometimes local taxes. The constraint to check is not “supports all 50 states”, which mainstream vendors advertise, but who registers you in a new state, how long it takes, and what it costs per state. Ask that question exactly; the answer separates tools that automate multi-state payroll from tools that merely tolerate it.

3. Tipped teams

Restaurants and salons live under rules most payroll tools treat as an afterthought: tip reporting, tip pooling, and wage rules that interact with tips. If this is your world, make the tool demonstrate a tipped pay run before you sign, not describe one; the restaurant payroll picks that survive tips apply this constraint to priced products, tip credit maths included.

4. Garnishments and ordered deductions

Court-ordered wage withholdings (garnishments, for child support, tax debt or a court judgment) arrive with legal priority rules and no flexibility on timing. A tool that makes you compute garnishments by hand has quietly moved you down the responsibility ladder. If anyone on your payroll has an active order, this is your hardest constraint even if it feels like an edge case.

5. Union or certified payroll

Public-works contracts and union agreements bring certified payroll reporting and prevailing-wage rules. Mainstream small-business tools mostly do not carry this; if you need it, your shortlist is a different shortlist (certified payroll software is its own category to search), and finding that out in week one is the whole value of this test.

6. A true zero budget

Genuinely free payroll exists, with hard edges. The best-known free tool is free for up to nine employees, and a truly zero budget means self-filing: the calculations cost nothing, and the tax deposits and returns stay yours to make. The moment you pay to make compliance someone else’s job, count both add-ons: tax filing runs about $35 a month, and paying deposits through the vendor’s own bank about $35 more, roughly $70 in all, per FitSmallBusiness’s July 2026 reporting; the vendor publishes neither price. The two stacks compare differently against a cheap paid tool. The $35 filing-only route stays under Patriot’s published $37-plus-$5-per-worker price at every headcount, as long as deposits keep going out from your own bank; the full $70 stack costs more than Patriot below roughly seven employees and only wins near the free tool’s own cap. Free also comes in shapes with no US tax engine at all: one free tool produces payslips only, with no US withholding, and another prints W-2s and 941s from a spreadsheet without e-filing anything. The free payroll round-up maps which “free” claims survived checking against the vendors’ own pages, and which did not.

Compare the survivors on three numbers

Total monthly cost at your headcount. Payroll vendors price on different models: base-plus-per-employee, per-employee-per-month, flat, or quote-only. Stickers across models cannot be compared directly; multiply everything out at your real headcount, and ask two price-list questions vendors rarely volunteer: what an off-cycle pay run costs, and whether running two pay schedules (say, hourly weekly and salaried semi-monthly) costs extra. On published 2026 prices, a common mainstream structure is $49 a month plus $6 per person, which is about $109 a month at ten employees. A per-employee-per-month product at the low end of the $10–25 range starts cheaper for a tiny team and overtakes the base-plus-fee model as headcount grows; at the top of that range it costs more than $49-plus-$6 from as few as three employees, so do the multiplication before assuming per-head is the small-team deal. And several of the biggest payroll brands published no prices at all as of July 2026, quoting only after a sales call; we keep the current list of quote-only holdouts in the Paychex alternatives round-up. Our pricing index keeps the current figures with a source and date on every number.

The year-two price. The price you sign at is rarely the price you keep: we documented one mainstream payroll vendor raising its base fee from $40 to $49 between two dated captures a few months apart in 2026, in our Gusto alternatives round-up. Ask for the renewal number in writing, and treat a refusal as an answer.

Hours to a human. The third number is literal: how many hours it takes to reach a person on the support channel your plan includes, not the one the top plan advertises. No vendor publishes it, so you measure it yourself during the trial, which brings us to support.

Who answers the phone when a pay run fails

Payroll support is not like software support, because payroll problems have statutory deadlines. A ticket queue that answers in four days is fine for a broken dashboard and useless for a pay run that did not land.

When we test payroll tools on real accounts, support access is part of the score for exactly this reason. Before you sign, find out what the support channel is on your plan, not the top plan: chat with a bot, chat with a person, email, or a phone number. Then test it once during your trial with a real question. The response you get while they are courting you is the best response you will ever get.

Common mistakes when choosing payroll software

Choosing on the sticker instead of the model. A flat-priced tool and a per-person tool trade places depending on your headcount. Do the multiplication at your size and at your size next year.

Assuming every tool files. The calculate–deposit–file gap is the most expensive assumption in this category, because the penalty lands months later, with interest. The same word problem applies to “automated”: which tools file for you, and which will even start the run for you, is priced out in the payroll automation software picks.

Buying the suite when you needed payroll. Bundled HR modules you will not run are not free; they are weight. Buy the rung of the ladder you need. The broader automation map is the place to decide what else, if anything, deserves software this year.

Ignoring the exit. Year-to-date exports, W-2 history, and who owns your data mid-year are questions for the demo, not for the divorce. Clean break points are quarter ends; ask how onboarding mid-year works before you need it.

Trusting undated prices. Payroll prices moved this year, and pages quoting last year’s numbers rarely say so. Any price without a date next to it, including on review sites, deserves suspicion. It is why every number we publish carries one.

The one-page version

Choosing payroll software comes down to one honest sequence: name the hardest constraint, eliminate everything that cannot clear it or any other constraint you carry, then compare the survivors on cost at your headcount, the year-two price, and reachable support, and test the exit before you enter. The feature grid can wait; the paycheck cannot.