Product-Led Growth Playbook

Product-led growth means letting the product do the work a salesperson or a marketer would otherwise do: show the value, convince the buyer and bring in the next user. It fits when a new user can reach value alone and the person who signs up is close to the person who pays. Then the work is to define activation, choose a free model, put the paywall after value and let usage tell sales whom to call.

Founders ask me whether they should “go product-led” as if it were a channel you switch on. It isn’t. When the product can sell itself, it is the cheapest growth a startup can buy; when it can’t yet, a free plan just gives the product away. Most product-led companies that sell to businesses still have salespeople. The difference is where sales starts: from what an account does in the product, not from a cold lead.

Buyers are pulling the same way. In Gartner’s survey of 646 B2B buyers, fielded in August and September 2025, 67% said they prefer a rep-free buying experience, and 45% had used AI during a recent purchase^[1]. So even a sales-led company should let a buyer see enough of the product, or enough proof of it, to decide the next step without waiting for a call.

This playbook covers the commercial side of a product-led motion: the free model, the paywall, the loop and the handoff to sales. What happens between sign-up and a lasting habit, and how to measure churn and revenue retention, is in the Customer Onboarding and Retention Playbook.

Answer four questions before you build a free plan. If the answer to two or more is no, lead with sales and use the product as proof, not as the funnel.

• Can a new user reach real value alone, in one session? If setup needs a data migration, an integration project or a consultant, a free plan will mostly produce abandoned accounts.
• Is the person who signs up close to the person who pays? PLG works when a user can buy, or can easily show the buyer why they should.
• Does the product get more useful as more people use it? Collaboration, sharing and output that others see give you a loop. A single-player tool has to be marketed to every new user.
• Does the price fit a card or a small approval? Large contracts can still start with product usage, but they end in a sales conversation.

Define activation from your own data, not from what feels important. Compare the users who stayed with the users who left, and find the early actions that separate them: usually importing real data, inviting a teammate or producing a first output that someone else sees. The action that predicts retention is your activation moment. The Customer Onboarding and Retention Playbook shows how to run that comparison and test the result; in a product-led company it matters even more, because the free plan, the paywall and the sales handoff are all built around it.

Write it as one sentence with a time limit: “An account is activated when it imports a customer list and sets up one alert within seven days.” If the team can’t agree on the sentence, you are not ready to build a free plan around it. For a rough outside reference, a 2022 survey of more than 500 products by Lenny Rachitsky and Yuriy Timen found a median activation rate of 25%, and 30% for SaaS products alone; each company defined activation its own way, so watch your own trend rather than the level^[2].

Cut the time to first value

Every step between sign-up and the activation moment costs you users. Take each step and ask whether it has to happen before value, after it, or never.

• Ask for the minimum at sign-up. Anything you can learn later, learn later.
• Show the output first. Offer sample data or a template, so a user sees a result before they connect anything.
• Defer the hardest setup step. Make it optional, or move it after the first result.

Three models, each suited to a different product:

• Free trial. The full product for a fixed period, then pay or lose access. It suits products whose value shows quickly and wouldn’t show on a limited plan.
• Freemium. A limited plan, free for as long as the user wants it. It suits products that spread when free users share or collaborate, and whose free plan costs little to serve.
• Reverse trial. In Elena Verna’s definition, a reverse trial “gives customers access to paid features and puts them on a freemium plan when the trial ends”^[3]. Trial users see what the paid plan does, and the ones who aren’t ready to pay aren’t cut off.

Expect each model to convert differently. The latest benchmarks are from Kyle Poyar’s survey of 200 B2B software products in January 2026, with ChartMogul and ProductLed, which counts the sign-ups that pay within six months (Figure 3). A trial without a card converts 4–6% at the good end and 10–15% at the great end; freemium 3–5% and 8–12%; a trial that asks for a card upfront 25–35% and 50–60%^[4]. The median across all products was 8%.

A higher rate doesn’t make a better model, because the models draw different numbers of sign-ups. The same report works each one through 1,000 website visitors: freemium gets 90 sign-ups and 5 paying customers, a card-required trial 35 sign-ups and 10.5 paying customers (Figure 4). So compare the models on paying customers per 1,000 visitors and on the cost of serving free users, not on conversion rate alone.

The upgrade prompt belongs at the moment a user has already got value and wants more of it: a second project, another teammate, more usage, an export that someone else will see. Gate on a limit that grows with value. Never gate the feature a user needs to find out whether the product works.

Then check that the pricing page works for a buyer with no salesperson: plan names the buyer understands, one clearly recommended plan, and prices visible without a call. The Pricing Page Teardown scores a page, and the Pricing & ARPU Modeler shows what a change in the tier mix does to revenue.

A loop is how one user brings in the next: invites, shared links, public outputs, templates other people copy, or integrations that put your name inside someone else’s workflow. Pick the one that follows from what users already do, and measure every stage of it: how many users start the loop, and how many of the people they reach sign up.

A loop multiplies your other acquisition. It rarely replaces it. Write it as stages with a count and a rate at each, and work on the weakest stage first.

In a product-led company, sales starts from what accounts do in the product. Define a product-qualified account from signals you can see: it reached the activation moment, it has several active users from one company domain, it is close to a plan limit, or it has viewed the pricing page more than once. Route those accounts to a person, and leave everyone else on the self-serve path. The same idea applied to one person is a product-qualified lead (PQL); an account-level rule suits products that several people at one company use.

Write the rule down and give it an owner. Each quarter, check which routed accounts actually bought. A signal that doesn’t predict a purchase wastes a salesperson’s week. The ICP Fit Scorer adds the company-fit half of the picture.

Buyers who would rather buy alone still want a person at some point. In a second release from Gartner’s survey of the same period, 69% of 645 B2B buyers said they prefer to check AI-generated insights with a sales rep^[5]. And there is some evidence the rule pays: in OpenView and Pendo’s 2023 survey of about 1,000 software companies, tracking product-qualified leads or accounts “increased the likelihood of fast growth by 61%”, fast meaning at least 75% a year^[6]. That is a correlation from a survey, and the last edition: OpenView stopped making new investments in December 2023 and has published no benchmarks since^[7].

You can’t run PLG on page views. Before you change anything, track these events by weekly cohort:

• sign-up, and each step between sign-up and activation;
• the activation moment itself;
• upgrade prompts shown and clicked, and upgrades;
• invites sent and accepted.

Then test, and expect most tests to fail. Of the well-designed experiments at Microsoft that set out to improve a key metric, “only about one-third were successful at improving the key metric”, Ronny Kohavi, Thomas Crook and Roger Longbotham reported^[8]. Size each test before you run it, and read the result honestly afterwards. The A/B Test Significance Calculator does both.

Review the funnel every month, with one owner

Name one person who owns the path from sign-up to paid, usually whoever leads growth. Each month they review four things by cohort:

• Activation. Activation rate and time to activation.
• Conversion. Free-to-paid conversion, and which prompts the upgrades came from.
• The handoff. Which product-qualified accounts bought, and whether the rule needs to change.
• Revenue. The share of new revenue that started self-serve and, once self-serve customers have paid for a year, their net revenue retention, which the Retention and NRR Workbook tracks.

Then they pick one step to improve next month and write down why. Two patterns show where to look. If activation rises and conversion doesn’t, the paywall is in the wrong place. If conversion rises and retention falls, the upgrade prompt is selling something the product doesn’t deliver.

What changes by stage?

What mistakes should I avoid?

• A free plan before an activation moment. Without a defined moment of value, you can’t design the onboarding, the paywall or the sales handoff (Step 2).
• Gating what users need to judge the product. If the free plan hides the thing that proves the value, users don’t upgrade. They leave (Step 4).
• Treating the tour as onboarding. A tour shows features. Onboarding is whatever gets a user to the activation moment (Step 2).
• Copying another company’s free model. Freemium works for products that spread. A single-player product on freemium is a giveaway (Step 3).
• No sales motion at all. Larger accounts still want a person. Usage tells that person where to look (Step 6).
• Counting sign-ups. Sign-ups are the easiest number to raise and the least connected to revenue; review paid conversion and revenue instead (Step 7).