Why product-led growth stops working around $2M ARR

The Mistake

Most founders who hit a wall with PLG around $2M ARR assume it's an optimization problem. Better onboarding. Smarter nurture flows. More in-app prompts.

Karel Papik ran that playbook at Product Fruits. And they had every advantage going in. Product Fruits is a digital adoption platform. They were using their own product inside their own product. If anyone should have been able to optimize their way through, it was them.

It didn't work. Free trial conversion slid from 24-25% down to around 15%. Growth stalled between $1.7M and $2M ARR. And the fix had nothing to do with PLG at all.

The mistake is thinking of PLG as a product decision. At a certain point, it becomes a product complexity decision. And the two have opposite signals.

Why Founders Make It

There are three reasons PLG feels like the right answer when it isn't.

PLG is the thing that got you here. Product Fruits scaled from zero to $1.7M ARR on PLG plus PPC. When something works that well for that long, you trust it. You assume a slowdown means you need more of it, not less.

The complexity creeps in invisibly. Five years ago Product Fruits was simple. Product tours. Hints. Feedback widgets. Today it does AI-powered discovery calls, onboarding that adapts in real time, and churn surveys you build by prompting an AI agent. Every feature looked like an improvement in isolation. The cumulative effect broke self-serve.

Your best prospects stop knowing what they want. Karel describes sales calls where customers say "wow, what is even possible to do now?" If your best prospects can't describe what your product does, they can't figure it out alone in a free trial either. PLG assumes the product teaches itself. Complex products can't.

How Karel Lost Conversion

Here's how it actually played out.

At the start: 24-25% free trial conversion. 8-9 month payback. A lean PPC engine doing the work. Life was good.

Then the product kept getting deeper. Karel puts it directly: "Five years ago we had the product tours and hints feedback. Maybe now our AI is able to have discovery call with you and ask you what you are looking for."

Free trial conversion slowly dropped to around 15%. Still not terrible in isolation. But combined with the ceiling on PPC spend ($1M to $1.5M a year, capped by search demand), it meant growth just stalled out.

The real tell showed up on sales calls. Every time an AE got on the phone with a qualified prospect, the prospect said some version of "what is even possible to do now?" That's when Karel realized customers needed to be shown the product, not given access to it.

So they built a real sales team. 15 people including implementation engineers. Not CSMs bolted onto PLG. An actual sales function with AEs who sit with a customer and put together the use case together.

The Fix (If You're Making It Now)

1. Measure free trial conversion quarterly, not just once. A slow drop is the signal. Product Fruits' conversion didn't crash. It slid.
2. When conversion drops, don't A/B test the onboarding flow. Ask whether your best prospects can even articulate what your product does before the demo. If they can't, the problem is shape, not copy.
3. Hire implementation engineers alongside AEs, not after them. Complex products need people who can show, build, and hand off in the same conversation.
4. Expect CSM and support costs to balloon if you try to force more customers through PLG. Karel is direct about this trap: "you would need too much people in CSM and tech support."

The Signal to Watch

Listen for the phrase "what is even possible" on your sales calls. If qualified prospects keep saying some version of it, your product has outgrown self-serve. PLG won't save you. You need a sales team to show people what they can't find on their own.