Best PTO policy for small business: accrual or unlimited, and what each one actually costs

For most businesses under 200 employees the best PTO policy is accrual per pay period with a cap at 1.5 to 2 times the annual grant, a starting grant of 10 to 15 days that scales with tenure, and a written waiting period phrased as not earning rather than accruing and forfeiting. Unlimited PTO is the right answer for a narrow case: a small, fully exempt, salaried team with high trust and no California employees. Everywhere else, accrual with a cap gives you the same cost control with none of the payout exposure.

The debate is usually framed as a culture question, which is why so many small companies get it wrong. Unlimited PTO reads as modern and generous, accrual reads as bureaucratic, and the decision gets made on vibes. But the two policies differ in three concrete, measurable ways: what liability sits on your balance sheet, what you owe when someone leaves, and how much time people actually take. Only the first of those favors unlimited without qualification.

The liability is the real difference, and it is bigger than people expect

Under an accrual policy, unused PTO is a balance you owe. Take a 25 person company paying an average of $75,000, offering 15 days a year. The grant is 120 hours per person. If the average employee carries half a year of unused accrual, that is 60 hours each, or 1,500 hours across the company. At an average hourly equivalent of about $36, the accrued balance is roughly $54,000 sitting on the books.

In a state that treats accrued vacation as wages, that $54,000 is not a soft number. It is payable in cash when people leave, in whatever order they happen to leave. Under an unlimited policy that number is zero, because nothing accrues, and that single fact is the honest case for unlimited PTO. Everything else usually cited in its favor, recruiting appeal and reduced administration, is real but small next to this.

The catch is that the zero is not guaranteed. In McPherson v. EF Intercultural Foundation, Inc., a California appellate court in 2020 applied the state's vacation payout statute to a policy the employer described as unlimited, because in practice requests needed approval, nobody was told they could take a large amount, and time off during the busy season was strongly discouraged. The trial court called it an "undefined" policy rather than an unlimited one, and the employees were awarded vacation wages anyway. You can end up with an unlimited policy that carries the liability of an accrual policy and none of the records to defend it.

The four options, and what each one costs

Notice that the accrual with a cap row is bounded rather than zero, and that is the trade being made. You accept a known, capped liability in exchange for records, predictability and a policy that behaves the same in all fifty states. For a company that plans to keep hiring across state lines, that consistency is worth more than the accounting entry.

Cap the accrual, never forfeit it

The instinct behind use it or lose it is sound: nobody wants a decade of unused vacation crystallizing into a cash payment. The execution is the problem. In states where earned vacation is wages, a forfeiture clause is void no matter how clearly it is written or how explicitly the employee agreed. The Colorado Supreme Court said so directly in Nieto v. Clark's Market, Inc. in 2021, striking down a handbook clause providing that an employee discharged for any reason would forfeit all earned vacation pay. Once an employer chooses to provide vacation pay, the court held, it is no less protected than any other wages and cannot be forfeited once earned.

A cap achieves the same objective and survives. Once an employee's balance reaches the ceiling, accrual pauses until they take time off. Nobody loses anything they earned, the incentive to actually use vacation is stronger than under forfeiture (because the clock stops immediately rather than at year end), and your maximum exposure per person is a number you chose. Set it at 1.5 to 2 times the annual grant. Below 1.5 people hit the cap during normal life events; above 2 the cap stops doing its job.

If you go unlimited, meet the four conditions on purpose

The McPherson court did something more useful than deciding one case: it set out, expressly, when an unlimited policy would not trigger the California payout statute. In writing, the policy should provide that the time off is not a form of additional wages for services performed but part of a promise of a flexible schedule; it should spell out the rights and obligations of both sides and what happens if time off is not scheduled; in practice it must allow genuine opportunity to take the time; and it must be administered fairly, so that it neither becomes a de facto use it or lose it policy nor produces inequities between someone who works long hours and takes little time and someone who does the reverse.

Two of those four are about behavior, not drafting, and they are the ones small companies fail. If you adopt unlimited PTO and your people take an average of nine days a year, condition three is not met and you have evidence against yourself. The fix is unglamorous: measure usage, publish a minimum expectation (two weeks is common), and have managers chase the people who are under it. An unlimited policy nobody uses is an accrual policy with no records.

Keep sick leave out of the bank if you operate in more than one state

Merging vacation and sick time into a single PTO bank is genuinely simpler, and it is the right call for a single state employer with no local sick leave ordinance. The moment you hire across state lines it gets expensive. State and city paid sick leave laws carry their own accrual rates, carryover minimums, permitted uses and anti-retaliation rules, and when sick leave lives inside a combined bank, the whole bank tends to inherit the strictest of those rules. You wanted one simple policy and you bought the union of every state's requirements.

The same logic that applies to break laws across multiple states applies here: the law follows the state where the employee physically works, so a company with people in five states has five sets of leave rules running at once whether or not the handbook acknowledges it.

Price the grant before you set the salary

The step almost everyone skips is putting a dollar figure on the policy. Fifteen days is 120 hours out of 2,080, or 5.8 percent of paid time producing no output. On a $75,000 salary that is about $4,350 a year per person. Across 25 people it is roughly $109,000 of paid time, before a single hour of it is paid out at separation.

That is not an argument for offering less. It is an argument for deciding it deliberately. Two companies paying identical base salaries are not paying the same if one offers 10 days and the other 20, and market salary data almost always describes base pay only, so leave sits outside it as a separate layer. If you are going to be generous on PTO, you can legitimately sit slightly below market on base and say so in the offer conversation. That only works if you know both numbers, which means building defensible salary bands from real market data first and treating the leave grant as an explicit adjustment rather than an accident. Work the accrual arithmetic through the PTO accrual calculator and you will have the hours figure the pricing depends on.

There is one more place the number shows up, and founders are usually surprised by it. An accrued leave balance is a liability like any other, so if you ever raise on a priced round or put the company in front of a buyer, it comes out of the purchase price. Anyone who has watched what a buyer's valuation actually prices in knows that unrecorded employee obligations are one of the first things diligence surfaces. A capped accrual policy makes that number small, known and easy to explain, which is worth something on its own.

What to do this quarter

Write down the current accrued balance in hours and in dollars. If you have never calculated it, that number alone usually settles the cap question. Then check the wording of your waiting period: it must say employees do not earn paid time off until they complete the period, not that they accrue it and lose it if they leave early. Those two sentences feel the same to a new hire and are opposite in court.

Then set the cap, publish it with at least a quarter of notice so nobody hits it by surprise, and price the grant into the loaded cost of every open role before you post a range. If you are in a state with posted salary range obligations, that loaded cost is what makes the posted number defensible in the first place, which is the same discipline behind posting salary ranges in a job ad.

Sources: US Department of Labor, Vacation Leave; California Labor Code 227.3; Suastez v. Plastic Dress-Up Co., 31 Cal.3d 774 (1982); McPherson v. EF Intercultural Foundation, Inc. (Cal. Ct. App. 2020); Minnick v. Automotive Creations, Inc., 13 Cal.App.5th 1000 (2017); Nieto v. Clark's Market, Inc., 488 P.3d 1140 (Colo. 2021); 26 M.R.S. 626. This is general information about published law, not legal advice. Leave obligations turn on your own facts and your own states, so confirm the policy with employment counsel before rolling it out.