Product-Led Growth Is Really Data-Led Growth: A Practical Playbook

Hila Qu says she has always rejected the view that PLG is the only future and sales is unnecessary. Working with clients, she sees many startups running both: a PLG motion that lowers the barrier and widens reach in a volume game, and a sales motion that targets big customers with large orders and a strong revenue base. Several clients try to get the benefits of both from an early stage.

She expects every company to need both eventually. A sales-dominated B2B vendor will see competitors add PLG and win end users who become champions inside their employers, while a PLG-only company will find others going after big customers. Lenny agrees, noting that PLG companies all added sales teams and sales-led enterprise products are now adding PLG, though Qu says it is harder to add PLG to a pure sales-led company.

For the definition, Qu points out that consumer products such as Facebook are product led by default, since there is no sales team. In B2B, sales traditionally had to close a contract before users could touch the product, but B2B users are the same people who use consumer software and expect to try before they buy. She cites Zoom as an example: you encounter it in someone else's webinar, sign up yourself and pay when you hit a limit.

The hallmark properties, she says, are a low barrier to entry through a free version or trial, no need for your boss's approval, a self-service checkout, and a free product that spreads on its own.

The first common pitfall, Qu says, is having no vehicle at all: many B2B sites show only a "book a demo" button, which cuts off the PLG entry point. Sales-led companies that add a free trial also face internal change, because sales, marketing and product must accept more people entering the funnel and new behavior data. A free experience must be built, not just advertised.

The second is underestimating the commitment. Some companies spend three months on a basic free trial and expect leads, conversion and self-service revenue to follow. Qu argues you can gather data and build conviction first, but then you should commit to at least a one- or two-year roadmap and be ready to change internal processes.

The third is lacking expertise and data. A common case is a company with no usage data at all, which Qu says defeats the purpose, since a free product should give you reach plus insight into which features correlate with conversion and retention. In such cases she suggests hiring someone experienced or bringing in an advisor.

Lenny asks for red flags of weak commitment. Qu lists equating PLG with launching a free version, assigning one person who would need to be a magician to coordinate sales, marketing and product, and adopting PLG because it is trendy. She says to check fit: relatively low product complexity, a short time to value, and many potential end users. A defense software vendor with three possible customers is an extreme case where PLG makes no sense, though most B2B software sits in the middle of a spectrum.

Qu lists what you need. The first is a vehicle: a free version, a free trial, an open source product, or, if none of those is possible, a realistic interactive demo, as Amplitude built when installing code was a barrier. The second is a short time to value, which means giving users a warm start with samples or templates and mini aha moments rather than a single big moment in five minutes.

The third is a self-service checkout so that people who are ready to pay can do it themselves. The fourth is a data foundation, since there is a big gap between activation and conversion that you can only understand through usage data, which lets you design journeys in product and email. Finally, pricing must be simple: if buyers must send information to sales to get a quote, the flow is broken.

Lenny summarizes it as something free and self-serve, quick time to value, self-serve checkout, a data foundation and simple pricing. Qu compares it to selling on an e-commerce site: all of these must be available to a B2B buyer, though having them does not guarantee success, which is why data and experimentation matter.

Qu recommends leaders start by understanding the PLG and sales-led funnels. In the traditional funnel, marketing turns visitors into leads and scores their interest from actions like opening emails or reading white papers; after enough points a lead becomes marketing qualified and goes to sales. In the product-led funnel, people sign up for a free version or trial and usage becomes the leading indicator, with marketing emails only a supplement.

She describes two conversion paths. If the price fits a company's credit-card budget, users can simply buy online, which is automated, low cost and runs every day. If usage is high and the account fits your ideal customer profile, such as a Fortune 500 firm, sales or customer success should reach out with white-glove service and try for a bigger deal. She calls this the PQL or PQA path, for product-qualified lead or account.

Using GitLab, a DevOps platform with a strong enterprise sales team from the beginning, she shows both funnels. A developer signs up for a free account for a side project, later recommends it when the employer wants to consolidate tools, starts a 30-day trial to test advanced features, and either buys five seats from the pricing page or draws outreach from sales because a big company is using it.

Mapping the funnel on a whiteboard is the easy part, Qu says. The details one layer down are where optimization opportunities live: what to say on the marketing site to drive signup, how to guide users to the three key features, and which payment options to offer.

Qu loves finding leverage, meaning small investments with the biggest results that build momentum. Her first step with advisory clients is a full PLG funnel audit in which she acts as a semi-interested user: is the value proposition clear, is signup smooth, do I reach an aha moment quickly or get lost, and can I even buy? She says there are usually many low-hanging fruits.

One client's checkout form asked a series of questions only relevant to UK customers, which confused her as a US user and caused her to drop off. A very common issue is that users are excited but do not know what to do on landing in the product. She also reviews the first emails a company sends, noting that one nudge email once brought her back and helped her reach the aha moment.

Lenny summarizes the audit as: is the landing page compelling, can you try it on your own, do you reach an aha moment, and can you buy. Qu adds that she asks for high-level data at each step, such as visitors, signups, aha moment and checkout. Combining her user experience with the numbers usually reveals the biggest opportunity, most often activation or conversion.

Qu defines the aha moment as the first time a user experiences the product's value. Facebook's famous version was adding 10 friends in seven days, which signaled social connection. For B2B SaaS, the value is typically a supported workflow, saved time or money, or solving a pain point. Lenny notes that aha moment and activation are often used interchangeably.

At GitLab, the growth team brainstormed about 10 potential high-value actions, such as merging a first PR or running a first pipeline. For each, they compared the 90-day conversion rate and 30-day retention against the average, looking at both because either alone gives an incomplete picture. Actions that lifted both most became candidates.

Because GitLab is a platform used by teams for different reasons, such as security or CI/CD, they ended up with a combined milestone: two users using two features within 14 days. The two-user part reflects a team product, since the first user is confident enough to invite a colleague, and the two features reflect platform use. Qu stresses that correlation does not prove causation, so the next step is experiments to get more people to take those actions and see whether conversion and retention actually rise.

Qu says you can run the audit yourself by imagining you are a consumer buying a product. If activation is weak, first find the right aha moment metric as your goalpost, then design a product experience where doing beats showing and showing beats telling: remove friction, give samples or templates, and supplement with email. If conversion is weak, the issues can include a buy button that is hard to find, or a checkout that is not localized.

She gives the example of a client whose data showed a low success rate in India because the chosen payment provider did not support that market well; adding another provider improved results at once. She advises running a hundred experiments on checkout because those users are so close to paying. A PQL or PQA motion is the next, bigger effort, and product-led acquisition suits collaborative tools like Airtable and Figma, where users invite their teams.

Lenny recaps the buckets as acquisition, activation, conversion and retention. Qu says activation is usually the best starting point for B2B software, because it historically was not designed for quick use. Her benchmark is Miro, which asks a few targeted questions about your use case and hands you a template so you are doing real work within about five minutes.

For conversion she says to aim for the ease of an e-commerce checkout such as Lululemon or Amazon, without confusing pricing. She would get activation and self-checkout working, with a reasonable user base, before investing in PQL, and would invest in acquisition if the product has built-in virality. Lenny adds that improving onboarding and activation is also one of the biggest retention levers.

Qu calls retention the messy middle because it plays out over a long period and users can silently leave at any moment, unlike the fast, testable stages of activation and conversion. She sees two steps. The first is building a habit, which requires enough frequency, and she warns that a product used once a month will struggle to form one.

She illustrates with Acorns, where she was head of growth. The product was set-and-forget passive investing, so users did not need to return to succeed, which made engagement hard to drive. The remedy is to build habit-forming or collaborative features into the workflow rather than rely on notifications.

The second step is expansion, which she treats as part of retention. Once usage is steady she sees three buckets: upgrading to a higher tier, buying more seats, and consuming more through add-ons. At GitLab that meant moving up tiers, buying seats and buying more CI/CD minutes, and she says activation and conversion tactics apply well here, guided by usage data.

Her biggest retention win at Acorns came from activation work. Data showed that a recurring investment feature correlated strongly with retention, so she ran experiments to get more people to set it up. The team then added higher-frequency use cases, such as an IRA retirement account and a debit-card spending account, which reframed retention as an adoption problem.

Qu describes two data buckets. The first is product usage data, which many B2B companies lack because sales-led selling never needed granular usage detail. The second is what she calls a customer 360 database, connecting usage with marketing campaigns, the CRM and account stages, so that you can see, for example, which users at a target account use which features and whether they responded to campaigns.

On infrastructure, she recommends a data hub such as Segment, a product analytics tool such as Amplitude, PostHog or Mixpanel, an experimentation tool such as Optimizely or Eppo, and a lifecycle marketing tool that triggers messages from product behavior, unlike lead-nurture email that scores opens. On top of that she mentions data enrichment tools like ZoomInfo and Clearbit, onboarding builders such as Appcues, and PQL tools such as Endgame and Pocus.

To get started, she says a product analytics tool and probably a data hub come first, and it is hard to get the choice completely wrong. More important is the data itself, since it is garbage in, garbage out. Many clients begin by auditing their instrumentation to see whether key actions are tracked in the right format, and by creating a data dictionary that lists each key action, its event name and properties so everyone shares the same definitions.

Lenny notes a previous guest, Crystal Widjaja, wrote about why analytics efforts fail. Qu adds that once you have real data users you should set up a data warehouse and ETL, since going without one is wild and fragile. She emphasizes that this investment also benefits the product and customer success teams, who can see what all customers use, not just what the loudest ones ask for.

Qu says the most common start is hiring a head of growth or lead growth PM with experience, then building a core growth squad with engineering, design and data. A less common option is a temporary cross-functional tiger team, which she suggests when the first focus is PQLs, since that work needs deep data and counterparts in sales and marketing. It is often a step before full commitment, and the two can be combined.

Once the team exists, she says to give it resources and an initial focus and look for early wins, which build confidence in PLG across the organization. Then think about a PLG org, not just a team, since the motion is cross-functional. She describes a head of growth product, a head of growth marketing and a head of product-led sales, usually reporting into product, marketing and sales respectively, with strong collaboration.

Metrics differ from sales-led. Top of funnel is high-quality signups rather than raw traffic, growth product owns usage and activation, such as activated teams and number of PQLs, and product-led sales owns converting PQLs into revenue through conversion rate, efficiency and revenue.

For the minimum team, a growth PM leads, with stronger analytics and experimentation skills than a typical PM and KPIs closer to sales and marketing. Qu would hire a data analyst very early, even first, because without insight experiments are directionless, then a dedicated engineer, with a designer and user research possibly shared. She prefers promoting an internal candidate, even one from outside product, and otherwise matching an external hire to your starting focus area.

Qu's most-recommended books are The Almanack of Naval, from which she learned about leverage through writing, code, capital and team, and How Women Rise, along with her own Chinese-language book. She enjoyed the sci-fi film The Wandering Earth 2 by Three-Body author Cixin Liu, and her favorite recent products are ChatGPT and Lululemon yoga pants.

Her favorite interview question for growth PMs and analysts asks about an experiment with a very unexpected result and what they did next. She wants candidates who run many experiments, can explain why the result was unexpected, which reveals the depth of their customer understanding, and can describe how they learned from it. A small change she made to her process was adding to tickets the success metric and which growth lever, such as acquisition or activation, the work supports, which sometimes led teams to drop projects.

Her favorite growth concept is the north star metric, which she applies to career decisions and parenting as a prompt to think long term and about what is valuable beyond salary. Lenny relates it to the question "what are you optimizing for?". Qu can be found on LinkedIn, and she is happy to talk with founders looking for a growth advisor.