That's Ulrik Lehrskov-Schmidt, who has spent a decade designing credit systems: if credits expire at the end of each period, the system is functionally identical to a license. You're just calling it credits.

Chargebee's guide to prepaid credits, written with him, covers the decisions in the order you hit them. Whether credits fit your product at all. How to price one without defaulting to cost-plus. What full, partial and no rollover each do to your sales cycle. And the two exchange rates most vendors only use one of — procurement locks dollar-to-credit, but credit-to-product stays with you.

For finance, the section to read twice is the deferred revenue tail. Allow one rollover on an annual contract and your recognition window stretches to 24 months. That's the real objection, and the fix is capping the window rather than killing rollover.

Consulting | Shop | Website | Newsletter | Speaking

Hey {{first_name}} 👋

Writing this one jetlagged.

Just got back from New York. First time. I've been building a business that sells almost entirely to US companies for years now, from a desk in Valencia, and I'd never actually set foot there.

A few things I wasn't ready for.

The pace. Everyone warns you and it still doesn't land until you're in it. Four conversations happen in the time one would take in Spain, and nobody thinks that's unusual.

And how much faster things move in person. I met dozens of people over the week. Operators, founders, a few existing clients I'd only ever seen on Zoom. Conversations that would have taken six weeks of LinkedIn DMs happened in four minutes standing next to a coffee station.

I came home with a notebook full of problems people are actually dealing with right now. Some of it is going straight into the next few issues.

Also, this is issue #100. Which I did not plan around the trip, but I'll take it.

In the last issue we covered PLG growth in SaaS. Today we're staying in Reporting.

Let's talk about ⬇️

Someone presents a 3.4% freemium conversion rate to their board.

A board member says they've seen 8% as the benchmark.

The room gets quiet. Product gets tasked with fixing the funnel. Engineering time goes to paywall experiments for the next quarter.

Here's the thing. That 8% almost certainly came from a free-trial dataset or a sales-assisted motion. Comparing it to self-serve freemium is comparing two structurally different funnels and calling the gap a performance problem.

For self-serve freemium, 3 to 5% is respectable. The 75th to 90th percentile sits at 6 to 8%.

So the 3.4% was fine. The anchor was wrong.

Correct the anchor before you correct the number.

Freemium conversion rate is the percentage of free users who upgrade within a defined period. That's the whole definition.

It works as a proxy for three separate things at once. Whether your free tier proves real value. Whether your paywall sits at the right threshold. And whether your acquisition channels bring in people who could ever plausibly pay.

A low number can mean any of the three.

Most teams assume it's the first one and start rewriting onboarding copy. The actual constraint is usually the third. They're acquiring free users who were never going to convert, through channels optimized for signup volume instead of intent.

The formula:

(Free users who upgraded ÷ Total free users in the cohort) × 100

10,000 signups in January. 340 upgrade by end of June. That cohort converted at 3.4%.

The denominator.

Every signup, or only activated users who completed the core action that proves your product works?

A rate against all signups will always look lower, because a meaningful share of signups never open the product a second time.

Neither is wrong. Pick one and hold it constant. Switching denominators mid-year is how a company accidentally reports a doubled conversion rate that's really a definition change. That doesn't survive diligence, and it damages everything else in the deck when someone catches it.

The window.

Freemium upgrades trickle in over months. There's no clock forcing a decision the way a trial does.

Measure at 30 days and you'll understate a product where most upgrades land closer to day 90.

Six-month cohort window is the standard, and it's how OpenView, ChartMogul, and ProductLed all define it in their published benchmarks. Track 30/60/90/180 breakouts inside it.

If your stack can't produce this without a manual spreadsheet reconciliation, that's worth noting on its own. It usually means the metric lives in three tools and nobody's definition matches.

The slide: conversion rate, 4.1%, up from 2.3% the prior quarter. Framed as the result of a paywall redesign.

What changed: the denominator. Previous quarter measured against all signups. This quarter measured against activated users.

Nobody did anything wrong on purpose. The analyst who built the new dashboard used activated users because that's the cleaner number. The person presenting didn't know the definition had moved.

The paywall redesign may well have worked. There's no way to tell from that slide, because the comparison isn't a comparison.

The fix: denominator definition lives in a footnote on every slide where the metric appears. When it changes, you restate the prior periods or you don't show the trend.

These come from the two largest published datasets. Lenny's Newsletter and OpenView established the 3-5% good and 6-8% great bands across 1,000+ products. The 2026 ChartMogul and Growth Unhinged report analyzed 200 B2B products and found a median of 8% across all models, with a 10x spread between top and bottom performers.

Now look at that last row again.

Card-required trials convert at roughly 30%. Card-free trials at around 6%. Five times the rate.

That gap is a filter, not a quality difference. Nobody enters a credit card unless they're seriously evaluating a purchase. The population entering that funnel is pre-qualified in a way freemium deliberately is not.

Freemium is built to cast the widest possible net. A low conversion rate on a wide net is the model working as designed.

Category matters too, though less than model choice. RegTech trial-to-paid converts around 23.6% while freemium-to-paid sits near 2.6% across all categories. Legal tech and RegTech convert freemium users at close to double the rate of consumer-adjacent categories, mostly because the paid trigger is sharper and easier to justify internally.

A shameless plug. It's issue 100, I'm taking the liberty.

I've been building something for the people who read this newsletter and want more than a weekly read.

SaaS Finance Lab. A founding cohort for SaaS finance leaders, founders or anyone else who wants the models, not just the theory.

Every month: one complete financial model built from scratch and fully documented. Plus a live session where I walk through it, take questions, and work through whatever people are actually stuck on.

The models are the real thing. Three-statement with the linkages built, cohort NRR decomposition, driver-based revenue builds, runway scenario models. What I'd hand a client, not a stripped-down sample.

Founding cohort: $490 for the year. That price is locked for founding members permanently. Seats are capped because the live sessions only work small.

For the first 20 people who join from this issue, I'm including everything currently in the shop at no extra cost.

If you've been reading for a while and found any of this useful, this is the version where I can actually go deep with you.

Segment before you optimize. A 3% blended rate might be 8% from organic search and 0.5% from a paid campaign dragging the average down. That's a channel-mix problem. The fix is reallocating spend, and no amount of upgrade-flow work touches it.

Check activation before the paywall. If activated users convert at 12% but only 35% of signups ever activate, your constraint is activation. Fix the first session before you touch pricing.

Put a dollar figure on free-user cost. Every free user carries infrastructure and support cost. Set that against your blended CAC for paid conversions. If free-tier carrying cost is climbing faster than paid conversions, that's a board-level trade-off rather than a product tweak.

Feed the rate into runway as a range. Base case, downside 1-2 points lower, upside 1-2 points higher. If your runway estimate swings more than a quarter across that range, your capital plan is more fragile than your deck suggests.

Report the trend with the channel mix that produced it. A single monthly number invites a single reaction. A six-month cohort trend alongside channel breakout gives the board something to decide.

The pattern across all six is the same. The problem is rarely the metric. It's fragmented ownership of the definition, assumptions nobody wrote down, and no cadence turning the number into a decision.

A dashboard tool will show you the same 3.4%. What changes the outcome is whether someone is accountable for what that 3.4% should trigger next quarter.

Five elements. Replaces the single percentage.

Cohort table. Signup month, total signups, activated users, paid conversions at 30/60/90/180.

Channel breakout. Same table split by acquisition channel, so the board sees which channels produce paying customers.

Cost line. Monthly infrastructure and support cost per free user, next to blended CAC for the same period.

Range-based runway impact. Current runway at base, downside, and upside conversion assumptions.

The ask. One recommendation. Reallocate spend toward the highest-converting channel, or put engineering time into activation instead of paywall experiments.

That structure turns a metrics update into a trade-off conversation. If your reporting stops at the first element and never reaches the last one, the deck is describing the past.

Freemium conversion rate is useful when it has a defined denominator, a defined window, and a decision attached.

Three things to hold onto.

Compare against your model. Self-serve freemium against self-serve freemium data. The card-required trial number in your head came from a funnel that filters for intent before anyone signs up.

Treat it as a cohort range. It feeds your cash forecast, your headcount plan, and your blended CAC. A point estimate hides the sensitivity.

Check upstream before you touch the paywall. Activation gaps and channel mix explain more weak conversion rates than pricing does.

If your team is debating a freemium number instead of using it, the metric isn't the issue. The gap is a model that translates the number into cash, runway, and a trade-off the board can actually act on.

Reply "FREEMIUM" and I'll send the cohort tracker. Signup month table with 30/60/90/180 breakouts, channel segmentation, free-user carrying cost line, and the three-case runway sensitivity built in.

One hundred issues. If you were here for the early ones, thank you for staying. If you joined recently, you picked a reasonable time.

Back on Valencia time by Thursday, probably.

{{first_name}} You can get free stuff for referring friends & family to my newsletter 👇

50 referrals - Cash Flow Models Bundle 💰

10 referrals - SaaS Financial Model 📊

{{rp_personalized_text}}

Copy & Paste this link: {{rp_refer_url}}