Product-led growth: implementation patterns and common pitfalls
Product-led growth gets treated like a philosophy, a go-to-market motion, or a pricing strategy depending on who you ask. I've heard it described as "let the product sell itself" so many times that the phrase has become meaningless. So let me try to be more precise about what PLG actually demands from a product — and why most teams that attempt it fall short.
PLG is fundamentally about three things: fast time-to-value, inherent shareability, and a self-serve expansion path. Miss any one of them and you don't have a PLG product. You have a product with a free tier.
The clearest example I keep coming back to is Figma. Not because it's a convenient cliché, but because it's such a clean illustration of all three elements working together. You can open Figma, create something real, and share it with a collaborator in under ten minutes. The sharing mechanism is embedded in the product's core value proposition — you can't get the most out of Figma without sharing, so the product naturally expands. And the expansion path is obvious: free users hit limits, teams upgrade, organizations consolidate on enterprise plans. Each step happens because the product itself creates the pressure to grow.
Compare that to a product where value takes weeks to realize, where the core workflow is inherently solo, and where there's no obvious reason for an individual to pull in colleagues. That product might be excellent software. It's not a PLG product.
The flywheel and what drives it
The PLG flywheel is real, but it's also frequently misunderstood. The loop goes: users experience value → they invite others or share outputs → new users enter → they experience value → repeat. What powers the flywheel is the combination of value density and virality. You need both.
Value density is about how much value a user gets in a short period. If it takes ninety days of configuration before a user knows whether your product is useful to them, no amount of shareability will save you. The first session has to deliver something real. Slack is good at this — you're having actual conversations with colleagues within minutes of signing up. Notion is decent but has a higher floor of setup required. Enterprise tools that require IT provisioning, SSO configuration, and training before anyone can use them are structurally incompatible with PLG regardless of what their marketing says.
Virality is about whether the product creates natural sharing moments. This can be output sharing (sharing a Loom video, a Canva design, an Airtable base), collaborative workflows (Google Docs), or network effects (Calendly — when you share a scheduling link, you introduce someone to the product). The sharing needs to feel like a byproduct of normal product usage, not a marketing prompt. If you're relying on "refer a friend" campaigns to drive growth, you're not doing PLG. You're doing referral marketing.
The activation problem
Here's where most PLG attempts actually break down: activation. Getting someone to sign up is easy. Getting them to the moment where they genuinely understand the value of your product is genuinely hard.
Every PLG product has an activation threshold — the point at which a new user has done enough to actually get it. For Dropbox, it was putting a file in the folder and seeing it sync on another device. They called this the "aha moment" and every product decision in the early days was oriented around getting users to it faster. For Slack, it was roughly 2000 messages sent — apparently that's when teams start to realize they actually communicate better this way. The specific threshold varies, but the concept is universal.
The problem is that most teams don't know their activation threshold with any precision. They have a vague idea that users who complete onboarding are more likely to retain, but they haven't done the work of connecting specific early behaviors to long-term retention. I worked with a product team that spent months optimizing their onboarding flow without knowing what "activated" actually looked like. When we finally dug into the data, we found that users who connected their first integration within the first session retained at 3x the rate of those who didn't. The onboarding flow didn't emphasize integrations at all. We were polishing the wrong surface.
The activation problem is also asymmetric. It's much easier to lose a user during onboarding than to recover one who bounced. Users who hit friction in the first session rarely return. This means activation work has an enormous ROI — improving your activation rate from 30% to 40% has more business impact than most features teams spend quarters building.
What makes a good activation target? It should be: a specific action (not a page view), correlated with retention (validated with data, not intuition), achievable within the first session for motivated users, and reflective of the product's core value proposition. If your activation moment requires a feature that only power users ever touch, it's the wrong metric.
Freemium is a design problem
The free tier is where most PLG product decisions get made and where most strategic mistakes happen.
There are essentially three approaches to freemium, and each makes a different bet about what will drive conversion.
The first is the capability gate: free users get the core product but are blocked from advanced features. This works when the advanced features are genuinely compelling and the core is genuinely useful on its own. Notion operates roughly in this model — the free tier is functional for individuals, and teams need to upgrade for collaboration features. The risk here is designing a free tier so good that users never feel the pull to upgrade. I've seen enterprise sales teams complain that their biggest accounts were happily using the free tier indefinitely because it covered 80% of their needs. That's a pricing and capability gate design failure, not a PLG failure.
The second is the seat gate: free up to some number of users, then pay per seat. This is Slack's classic model. The bet is that as a team grows, the value of the product grows too, and expansion feels natural rather than punitive. The risk is that teams game the seat limits — keeping a small group on free, managing two-tier access internally. When Slack moved to limit free message history, they were trying to close exactly this gap. It worked, but not without pain.
The third is the usage gate: free up to some consumption threshold (API calls, storage, projects, reports). This is increasingly common in developer tools and data products. The bet is that usage and value are correlated — heavy users are getting more value and should pay more. The risk is that the usage limits feel arbitrary to users and create a jarring cliff rather than a gradual ramp.
The worst freemium implementations I've seen try to be all three simultaneously, applying capability gates, seat limits, and usage caps in a way that makes the pricing page look like a hostage negotiation. Users don't upgrade from confusion. They churn.
The organizational implications most companies miss
PLG isn't just a product strategy. It rewires how an organization has to operate, and that's where a lot of the difficulty lives.
The most significant shift is in how growth gets measured and who owns it. In a traditional sales-led organization, growth is owned by the sales team, measured in pipeline and bookings, and driven by outbound activity and relationship-building. In a PLG organization, growth is owned by the product, measured in activation and expansion, and driven by product decisions. This changes the power dynamics in ways that many organizations aren't prepared for.
I've seen PLG initiatives fail not because the product strategy was wrong but because the sales team continued to treat the free tier as lead generation for their pipeline rather than as the actual product. They'd engage with free users, try to accelerate them through a traditional sales process, and end up short-circuiting the PLG motion entirely. Free users who get a demo call and a contract negotiation don't go on to become the organic champions who pull in their colleagues. They become a closed deal that doesn't expand.
The other organizational shift is in how product teams are structured. PLG requires a growth function that sits at the intersection of product, data, and marketing — what some companies call a "growth PM" role. This person is responsible for the metrics that bridge acquisition and retention: activation rate, time to activation, expansion revenue per cohort, viral coefficient. In a PLG company, these metrics are as important as feature adoption metrics, and someone needs to own them with the same rigor.
Support also changes. PLG companies often see a barbell distribution in their user base: a large number of very light free users who need minimal support, and a smaller number of heavy power users who need sophisticated help. The middle — the moderate user who needs hand-holding through a sales process — is less common. This requires a support model that's primarily self-serve (documentation, community, in-app help) with a premium tier for enterprise accounts.
How AI is changing PLG
The dynamics I've described above are being reshaped by AI in ways I find genuinely interesting — not just incrementally better but structurally different.
The most important change is personalized onboarding at scale. Historically, onboarding was a one-size-fits-all flow with maybe a few branches based on role or use case. AI makes it possible to adapt the onboarding experience in real time based on what a user actually does, not what they claim they want in a survey. If a user spends three minutes on the templates page and then bounces to the integrations section, an AI-powered onboarding system can route them differently than a user who dives straight into a blank canvas. This kind of adaptive onboarding materially compresses time to activation — which is the single biggest lever in PLG.
The second change is AI-assisted activation. Some products are now using AI agents to actively help users get to their aha moment. Rather than relying on users to discover the product's value themselves, the AI can surface the relevant workflow, generate a starting template, or complete the first task on the user's behalf to show what's possible. This is genuinely new. Previous onboarding relied on checklists, tooltips, and empty state screens. AI onboarding can generate a personalized first artifact from the user's own data — their real use case, not a generic demo. When a project management tool can look at your existing workflow and generate a first board that's actually relevant to your team, the activation moment is dramatically closer to signup.
The third shift is in viral mechanics. AI-generated outputs — documents, designs, analyses, code — are inherently shareable because they're interesting. When Gamma (the AI presentation tool) generates a polished deck from a user's bullet points, the user is immediately inclined to share it because it's genuinely impressive. The sharing feels like showing off something you created, not spreading a marketing message. This creates a more natural viral loop than traditional PLG virality, which often feels more manufactured.
What this means for PLG product strategy is that the activation timeline is compressing. If your time-to-value was 30 minutes before AI, it might be 5 minutes now. If it was a week of configuration, it might be a day. This shifts competitive advantage from whoever has the best self-serve setup flow to whoever can deliver the most impressive first-session output. The product that makes a new user feel effective in the first five minutes will win, and AI is what makes five-minute competency possible for complex tools.
There's also a new risk worth naming: AI-assisted activation that feels manipulative. If the AI generates something so polished and impressive in the first session that it doesn't reflect what the product will actually do for the user in normal usage, you've created a bait-and-switch. The activation metric improves; the 30-day retention suffers. The goal is still to get users to genuinely understand the product's value, not to wow them with AI magic that disappears after they convert to paid.
What separates PLG companies from companies with a free tier
The difference usually comes down to whether the product team genuinely understands why any individual user would, unprompted, choose to pay more or expand their team's usage. If the answer requires a sales argument — "you'll get better support," "you'll unlock more collaboration" — that's not PLG. That's a sales pitch attached to a free trial.
In a real PLG product, the expansion feels inevitable rather than persuaded. Users hit natural limits imposed by growth (more collaborators, more data, more projects), and the path forward is obvious. The upgrade isn't a decision to buy software. It's a decision to continue doing what's already working.
Getting there requires the activation work I described above, combined with honest pricing design and organizational changes that most companies underestimate. It also requires patience. PLG growth curves are often slow at first and exponential later, and that shape makes them unpopular in organizations where quarterly results matter. Many companies abandon PLG before the flywheel starts spinning because they can't sustain the patience required.
The ones that don't abandon it are the ones that generate compounding growth without proportional growth in sales headcount. That's a structural advantage that's hard to replicate. When it works, it really works.
This article is part of a series on product management in an AI-transformed landscape.