Pricing for Product Led Growth

Most broken free plans fail in the same place. The limit sits before the value moment, so the user hits a wall while still deciding whether the product works, and they leave with no reason to pay. Free plan design is pricing work, not product work, and the threshold you pick is the single most consequential number in a product led business.

Why the position of the limit matters more than its size

The free plan has one job: get a user to the point where the product has visibly done something useful, then let them run into a ceiling shortly after. Everything else is detail.

Think about the sequence. A user signs up, does some setup, and at some point crosses the line where they’d be annoyed to lose the account. In Slack that’s the first real conversation thread with colleagues. In Figma it’s the first file someone else commented on. In Calendly it’s the first meeting booked without an email exchange. If your limit binds before that moment, you’ve built a demo with a timer. If it binds two months after, you’ve built a free product with an optional donation page.

The practical version: instrument your activation event, look at the usage distribution of accounts that reached it, and set the limit so that the engaged cohort hits it inside 30 to 60 days. If you’re not sure what your activation event is, that’s the prior problem, and the PLG metric set is where to start rather than here.

Choosing the limiting dimension

Five dimensions are available, and they are not equally good. The right one grows automatically as the customer gets more value, so willingness to pay rises alongside the constraint.

Dimension	Grows with value	Risk	Works well for
Objects created (projects, boards, forms)	Yes	Users delete old objects to stay free	Workspace and document tools
Usage volume (API calls, minutes, events)	Strongly	Spiky usage causes bill shock anxiety	Infrastructure, video, analytics
History and retention window	Yes	Feels punitive if too short	Messaging, logs, analytics
Integrations and automations	Partly	Blocks the stickiness you want to create	Workflow and ops tools
Seats	Only sometimes	Caps the collaboration loop directly	Single player tools with no sharing

Seats deserve their own warning. If your product gets better when more people are in it, a seat cap on the free plan is a tax on your own distribution. Figma let anyone view and comment for free and charged for editors, which kept the file circulating through organisations that hadn’t bought anything yet. Slack’s old free plan capped message history rather than people, for the same reason. Both companies chose a dimension that let the account grow while the constraint tightened.

Usage limits are the cleanest when your cost of goods scales with consumption. If every free user costs you real compute, as with Vercel builds or a video product’s rendering minutes, the limit should track that cost so free never becomes a liability. Set it at the point where a generous hobbyist is fine and a small business is not.

Setting the threshold with percentiles rather than intuition

Here is the method. Take every free account created in a 90 day window that reached your activation event. Measure their consumption of the candidate dimension at day 30. Plot the distribution. Then set the limit between the 60th and 75th percentile of that distribution.

Why that band? Below the 50th percentile you frustrate the median engaged user before they trust the product. Above the 85th percentile only power users ever feel the ceiling, and power users are a small slice. The 60 to 75 band means a clear majority of the people who actually adopted the product will meet the limit while they still remember why they signed up.

One tradeoff nobody writes down: tightening a free plan almost always improves conversion rate and reduces total signups. If your growth model depends on free user volume feeding word of mouth, a tighter plan can lift conversion percentage and shrink absolute revenue at the same time. Watch new paid customers per month in absolute numbers, not the conversion ratio.

What you must never gate

Three things. Gate any of them and you break the mechanism that makes product led growth cheaper than sales led growth.

Collaboration. Inviting a colleague is your distribution channel. Charging for the invite is charging for your own marketing. Let free users bring people in, and charge when the group becomes an organisation that needs admin control.

Data export. Holding data hostage generates churn-flavoured rage and a review problem. It also fails: someone writes the scraper. Export freely, and earn the retention with the workflow instead.

The core workflow. If the product is a form builder, free users must be able to build and publish a form. If it’s a video tool, they must be able to record and share. Cripple the core and the user never learns what they’d be paying for.

The first paid price point, and why 12 dollars a seat is a trap

Low entry prices feel friendly and quietly cap the business. At 12 dollars per seat per month, a three seat account is 432 dollars a year. Subtract payment fees, support contact cost, and any human touch at all, and the account cannot absorb more than about 150 dollars of acquisition cost at a reasonable payback.

That’s fine if your acquisition is genuinely zero-touch. It’s fatal the moment you add a salesperson, a paid channel, or a customer success motion. Plenty of teams set 12 dollars in year one, then spend year three trying to raise it against a base of 9,000 accounts who anchored on the old price.

The number to protect is annual contract value per account, not per seat. A 15 dollar seat price with an average of 12 seats is a healthier business than a 49 dollar seat price with an average of two. Look at your actual seat distribution before you argue about the headline number, and if you’re due a reset, run it as a structured project using the 90 day pricing and packaging refresh rather than editing the pricing page on a Friday.

Instrumenting the upgrade trigger

A limit that binds silently is a churn event. A limit that binds with a clear, specific, in-context upgrade path is revenue. The difference is engineering work that usually gets deprioritised because it isn’t a feature.

Build three things. A pre-limit warning at roughly 80 percent consumption, shown in the product rather than by email. A hard-stop screen that names the exact limit hit, shows what upgrading unlocks in the user’s own terms, and offers a one click path to checkout. And an event fired to your analytics and your email tool at both moments, so lifecycle can follow up with something relevant.

The copy matters more than people expect. “You’ve reached your plan limit” is a wall. “You’ve used all 1,000 events this month. Your team sent 340 last week, so the 10,000 event plan covers you for about six months at this rate” is a sales conversation with arithmetic in it. Write the second kind, and pair it with the sequences in freemium to paid upgrade emails.

Freemium, free trial, or reverse trial

Three models, different conversion mathematics, and the choice depends on how long your product takes to demonstrate value.

Model	Conversion to paid	Signup volume	Best when
Open freemium	2% to 5%	Highest	Product has a network or content loop and low marginal cost
Opt-in free trial, no card	8% to 15%	Medium	Value is obvious within 14 days and setup is quick
Card required trial	30% to 50%	Lowest	High ACV, qualified traffic, strong brand
Reverse trial	5% to 12%	High	Premium features need habit formation before they are missed

Reverse trials are underused. Everybody gets the paid product for 14 to 30 days, then lands on the permanent free plan instead of hitting a paywall. You keep the signup, you keep the account, and the user now knows exactly what they lost. Notion and several developer tools run variants of this. It’s the right default for products where the premium features only feel necessary after someone has built something real.

If you’re choosing between models at all, the wider decision is really about motion rather than pricing mechanics, and product led growth for SaaS covers the operating consequences.

Adding a sales assisted tier without cannibalising self serve

Put a genuine gate in front of it. SSO and SCIM, audit logs, custom data residency, a signed MSA, invoicing and procurement support, volume commitments. These are things a five person team doesn’t want and a 400 person company can’t buy without.

The failure mode is a sales tier defined by feature envy rather than buyer type. If the sales tier’s differentiator is advanced reporting, your best self serve accounts will call sales for it, and you’ll have converted a 3,000 dollar credit card customer into a 6,000 dollar customer who costs 4,000 to serve. Define the tier by organisational requirements, not by capability, and the two motions stay in separate lanes. The tradeoffs run deeper than pricing, which is why self serve versus sales assisted is worth reading before you staff a sales team.

One more structural point: the free plan, the self serve tiers and the sales tier have to read as one ladder on a single page. Buyers compare vertically. A pricing page spec that keeps the value metric consistent across all four columns does more for conversion than any individual price change, and the way Ahrefs built its free tools into the same ladder is a useful example of the product led SEO teardown in practice.

Where to start on Monday

Pull the day 30 usage distribution for your activated free accounts on whichever dimension you already meter. If you can’t produce that query, that’s the week’s work. If you can, calculate the 60th and 75th percentiles and compare them to your current limit. Most teams find their limit sits at the 90th percentile or higher, which means almost nobody ever feels it, and the fix is a single number change plus an upgrade screen. When you’re ready to restructure the tiers around it, the packaging logic in good better best and the wider SaaS pricing strategy material will keep the ladder coherent.