Revenue growth

What should trial to paid be, and what moves it?

Before the benchmark, fix the two denominators, because almost every reported trial-to-paid number is wrong in one of these ways.

The first is what counts as a trial. If your denominator includes people who signed up and never logged in again, you are measuring your signup form, not your trial. Count a trial from the first session, and keep a second line for signups that never started, because that gap is an activation problem and it has a different fix.

The second is the window. If you count a payment whenever it arrives, your rate improves every month you look at it, and no two reports ever agree. Pick 30 days after the trial ends, and keep that window even when it flatters you less.

With those settled, the published ranges are useful as a sanity check and nothing more. In a survey of more than 1,000 products run by Kyle Poyar, Lenny Rachitsky and the Pendo team, published in August 2023, a free trial converting 8 to 12 percent was called good and 15 to 25 percent great; self-serve freemium was good at 3 to 5 percent and great at 6 to 8 percent; sales-assisted freemium sat higher, at 5 to 7 percent good and 10 to 15 percent great. Developer-focused products came in at a median of 5 percent, about half the rate of everything else. In the SaaS Conversion Report that Kyle Poyar published with ChartMogul and ProductLed in March 2026, the median free-to-paid rate across 200 B2B products was 8 percent.

The same report found that trials asking for a credit card up front converted at 30 percent, more than 5 times the ones that didn't, and that only 20 percent of products ask.

That is the number most often misread, so it is worth being plain about. A card-gated trial does not convert better because the card persuades anyone. It converts better because it removes the people who were never going to buy before they reach the denominator. Add a card field on Monday and your conversion rate will rise on Friday. Your count of new paying customers may fall at the same time. Only one of those two numbers pays salaries.

What actually moves the rate is the distance between the signup and the first time the product does the thing the page promised. Shorten that and the rate moves. Make the reminder emails louder and it doesn't. The method for finding that moment, and the five decisions a trial has to carry a person through, are in Trial to paid.

Is the price right, and for whom?

Price is the cheapest test available to a growth team and usually the last one it runs.

Michael Marn and Robert Rosiello, writing in Harvard Business Review in September 1992, worked out the average economics of 2,463 companies in Compustat and found that a 1 percent improvement in price, assuming no loss of volume, increases operating profit by 11.1 percent. That is a paper about industrial pricing from more than three decades ago, and it is not a software benchmark. The shape of it still holds, which is the point worth taking: price runs straight to the bottom line, and changing it costs nothing but nerve.

The "for whom" half of the question is the half that gets skipped. A single price is a bet that willingness to pay doesn't vary across your customers, and that bet is almost always wrong. Three places tell you so without any research budget:

• The discount log. Every approved discount is a price your sales team believes the market will bear. If the same 20 percent comes off every mid-market deal, you don't have a discount policy, you have a mid-market price you haven't written down.

• Where deals stall. If the pause comes before the price, you have a value problem. If it comes after, and always at the same tier, you have a packaging problem: the tier above is asking for money without offering the thing those buyers came for.

• Usage per unit of spend. Sort your accounts by what they pay, then look at what they use. Where the heaviest users pay the least, there is an entire price tier sitting unbilled.

Test a new price on new customers only, hold it for a full sales cycle before reading it, and read it on revenue rather than on win rate. Win rate falls when you raise price. That is the mechanism working, not a failure.

How long is payback, and what shortens it?

CAC payback is the number of months of gross profit a new customer produces before it has repaid what it cost to win them. The 2019 SaaS Private Survey that KBCM Technology Group ran with forEntrepreneurs put the definition well: the number of months of subscription gross profit required to recover the fully loaded cost of acquiring a customer. Both of those qualifiers do work. Gross profit, not revenue. Fully loaded, not media spend.

On the ranges, the honest answer is that the famous one is old. David Skok's guideline, written in 2011 and still the one most founders have heard, was to keep payback under 12 months. Skok has since said plainly on his own site that the guideline came from a period when capital was scarce, that land-and-expand businesses can work at around 20 months, and that sub-12-month recovery is now very rare to find.

The survey data makes the same point from the other direction. Benchmarkit's 2024 B2B SaaS Benchmark Report, run with Pavilion across roughly 936 companies on calendar 2023 numbers, put the median new-customer CAC ratio at $1.76: $1.76 of sales and marketing spend for every $1 of new annual recurring revenue. Work that through at an 80 percent gross margin and the median company needs about 26 months of gross profit to get its money back. At 75 percent it is about 28. That arithmetic is mine, not theirs, and you should redo it with your own margin, but it tells you where the middle of the market actually sits.

Three things shorten it, and only three: the price and the billing terms, the mix of segments and channels you spend into, and the share of that spend that turns into paying customers. Cutting the budget is not on the list. It shortens nothing and reduces growth. The formula, the levers worked through one by one, and the two ways the number gets flattered are in CAC payback: the one number for a Series A.

Which number does the board actually watch?

Net revenue retention, and it is usually the number the growth team has least to do with.

The same Benchmarkit report put median net revenue retention at 101 percent for 2023, down 4 points since 2021, with gross revenue retention in the bottom quartile at 79 percent and companies selling contracts above $100,000 holding 90 percent or better. A company at 101 percent is standing still without new logos. A company at 120 percent has a growth rate it didn't have to buy.

The reason this belongs to growth and not only to customer success is in the cost. In that same report, the median expansion CAC ratio was $1.00 against $1.76 for new customers. Expanding an existing account cost a little over half what winning a new one did. It is the cheapest revenue in the company, and in most companies nobody owns it: sales is paid on new logos, success is paid on renewals, and the upsell sits between them.

If you want one thing to take from this stage into next quarter, give expansion revenue an owner, a target and a weekly number. That is a decision, not a project, and it can be made on a Monday.

What good looks like

• One written definition of a trial, a paying customer and the window between them, agreed with finance, used in every report.

• A price that has changed at least once since the product did.

• CAC payback computed on gross profit with every acquisition cost loaded in, for new customers only, refreshed monthly.

• Expansion revenue with a named owner and a number of its own.

• Net revenue retention on the weekly dashboard rather than only in the board deck.

• Nobody in the company arguing about which of two conversion rates is correct, because there is only one.