A model where the product itself does the acquiring and the converting. The honest test: what share of your customers pay without talking to anyone?
In practice
Product-led growth is a distribution model, not a quality bar: it works where a product can deliver value before a conversation, which is a property of the product and of who buys it rather than of how much the team wants it to be true. Hence the value of a measurable test instead of a claim — the share of customers who reached paid without talking to anyone. And of computing it twice, by customer count and by revenue, because the gap between the two is almost always the finding: a company can be at 25% by customers and 3% by revenue, in which case the motion exists but is not yet paying for anything. Neither figure makes the company good or bad; they only settle which of the two it is, which has to be known before deciding where the next year of engineering goes.
The formula
Self-serve share = (customers who reached paid without talking to anyone) ÷ (all new customers in the period)
- Without talking to anyone
No demo, no discovery call, no negotiated quote. A support ticket does not count as a conversation — helping someone who is already buying is not selling to them.
- Customers, and revenue
Compute the share twice, once by customer count and once by revenue. The two answers are usually far apart, and the gap between them is the finding.
- Period
A quarter is usually the right window: long enough to absorb a single large deal, short enough to notice the motion drifting from one quarter to the next.
Product-led growth is a distribution model, not a quality bar. It works where a product can deliver value before a conversation — which is a property of the product and of who buys it, not of how much the team wants it to be true.
Worked example
A company that calls itself product-led
- 200 new customers in a quarter. 150 booked a demo from the pricing page; 50 signed up and paid without ever talking to anyone.
- Self-serve share by customer count: 50 ÷ 200 = 25%. Already less product-led than the company describes itself, but defensible as a motion being built.
- Now by revenue. The 50 self-serve customers average €40 a month (€2,000 MRR); the 150 sales-led average €400 (€60,000 MRR). Self-serve share of revenue: 2,000 ÷ 62,000 ≈ 3%.
- 25% and 3% describe the same quarter, and only one of them should drive the decision. If the question is "should we invest in the product as a sales channel", the answer has to be weighed against the 97% of revenue the other motion currently brings.
Neither number makes the company good or bad — a sales-led business with a free trial is a perfectly sound business. What the two numbers settle is which one it is, which is a question that has to be answered before deciding where the next year of engineering goes.
Orders of magnitude
- The pattern most often cited is that product-led companies show both a lower acquisition cost and a lower average contract value. Both, which is why neither figure judges the model on its own: the number that does is CAC payback, where the two effects meet.
- The model needs a product whose value is visible before a conversation and a buyer who is allowed to buy alone. Sell something that requires a security review and a procurement process and no amount of onboarding polish makes the motion self-serve — that is a constraint of the market, not a failure of execution.
- Running both motions on the same segment without deciding which one owns it is the common failure, and it shows up as sales people being paid on deals the product had already closed.
How to improve it
- Measure the self-serve share both ways before claiming the model, and publish both. One number is a slogan; two are a diagnosis.
- Make value land before the conversation would have happened. That is the time-to-value question, and in a product-led motion it is not a comfort feature — it is the sales call.
- Put the expansion trigger inside the product rather than in a calendar: a usage threshold that fires an offer is the product doing the selling, which is what the model means in practice.
- Decide which motion owns which segment and write it down. Self-serve under a size, sales-led above it, and no overlap — the overlap is where both teams claim the same customer and neither owns the experience.
In the Tour
Do you have a primary acquisition channel that's identified and measured?
- Yes, clearly identified and tracked20 pts
- We have one, but don't track it closely7 pts
- Not really / it's scattered0 pts
The Tour asks whether you have a primary acquisition channel that is identified and measured, for 20 points. It is a pointed question for a product-led team, because the honest answer is often "the product" — and a channel you cannot point at is a channel you cannot measure. Teams claiming the model and answering 7 here usually have a free tier and no idea what share of revenue comes through it, which is exactly the figure this page asks for.
Questions people ask
How do I know if I am actually product-led?
Compute the self-serve share by revenue for last quarter. Not by sign-ups, not by customer count — by revenue. Most companies that describe themselves this way find a number well under a third, which does not make them wrong to aim for it, only wrong to plan as though they had arrived.
Do I need a free tier?
You need a way for someone to reach value before paying, which a free tier, a trial, or a genuinely useful demo environment can all provide. The free tier is the most common form and the most expensive one — it commits you to serving users who may never pay, and that cost belongs in your gross margin.
Is product-led growth cheaper?
Per customer, usually yes; per euro of revenue, not necessarily, because the contracts are smaller. The comparison that settles it is CAC payback, which folds both effects into one number. A motion with a €200 CAC and a €20 monthly margin pays back in ten months; one with a €5,000 CAC and a €900 margin pays back in six.
Can both motions coexist?
Routinely, and most companies that grow past a certain size end up running both. What does not work is running them on the same customers without a rule: the rule can be company size, plan, or geography, but there has to be one, written down, or the two motions spend their time re-selling each other's customers.