A funnel rarely fails all at once. It leaks. A few percent lost at signup, a few more at activation, another slice at the paywall: none of them alarming on their own, all of them compounding into a number that never shows up on a single dashboard. By the time revenue growth stalls, the leaks have been running for months.
The hard part isn’t fixing a leak. It’s finding it. Aggregate conversion rates hide the exact step where users give up, and revenue reports tell you what happened without telling you where. Below are the five places we see product-led funnels leak most often (roughly in the order a user encounters them) and the signal that tells you it’s happening.
1. The signup step that asks for too much
The first leak is the one everyone underestimates, because it happens before anyone is really a user. Every additional field, every “verify your email before continuing,” every forced workspace-naming step is a place where intent quietly evaporates. The people who leave here were interested enough to click, and you lost them to friction, not to a bad product.
What makes this leak invisible is that it looks like traffic that “wasn’t serious.” It usually isn’t. It’s serious traffic meeting a form that asks for a company size before it has shown a single ounce of value.
The signal: a gap between signup started and signup completed. If more than one in five people who begin signup never finish, the form is the problem, not the audience.
2. The activation gap between “signed up” and “got value”
Activation is the single most important (and most neglected) step in a product-led funnel. It’s the moment a new user does the one thing that makes your product click: sends the first message, connects the first data source, ships the first project. Users who cross that line stay. Users who don’t, churn quietly and are almost never won back.
The leak here is brutal precisely because it’s silent. These users signed up. They count as “conversions” in your acquisition report. They inflate your MAU for a week. And then they’re gone, and no report flags them because, technically, they converted.
Acquisition tells you someone walked in the door. Activation tells you whether they found the light switch. Only one of them predicts revenue.
The signal: the time-to-value curve. Plot how long it takes new users to reach the activation event, cohort by cohort. If most of your paying customers activate within their first session but the majority of signups take days, or never get there, you’ve found where the funnel is bleeding out.
3. The habit that never forms
Between the first “aha” and the decision to pay, there’s a stretch most funnels don’t measure at all: the habit loop. A user who activates once but never comes back has technically been activated and will still churn. Value experienced once is a demo. Value experienced repeatedly is a reason to pay.
This is the murkiest leak because it lives in the space between activation and the paywall, where most analytics setups have no defined event to track. Teams instrument the signup and the checkout and leave the middle (the part that actually determines whether checkout ever happens) as a black box.
The signal: second-week return rate for activated users. If people love it on day one and vanish by day nine, your problem isn’t the paywall you haven’t reached yet. It’s the habit that never took hold.
4. The paywall that surprises people
By the time a user hits your paywall, they should already know what they’re paying for. The leak happens when the paywall is a surprise: a wall that appears mid-task, gating a capability the user didn’t realize was premium, at a price no earlier moment prepared them for.
There are really two failure modes here, and they need opposite fixes. Some paywalls are hit too early, before the user has felt enough value to justify the cost. Those need to move later or offer a taste. Others are hit too late, after users have quietly extracted everything they needed from the free tier. Those are leaving money on the table. The revenue number alone can’t tell these two apart. The journey can.
The signal: what users did in the sessions before they hit the paywall, and whether the ones who bounce off it had experienced the feature they’re being asked to pay for. If they’re bouncing on a feature they’ve never used, the paywall is arriving before the value.
5. The expansion revenue you never ask for
The last leak isn’t a drop-off. It’s an absence. A customer who pays and stays but never expands is a leak measured against the revenue they would have generated if anyone had noticed they were ready to grow. The account that hit its seat limit three times last month is telling you something. The team that’s using a feature far beyond its plan’s intent is telling you something. Most funnels aren’t listening.
This one hides better than all the others, because the customer is happy and the MRR is stable. Nothing looks broken. But “stable” for an account showing every signal of readiness to upgrade is its own kind of leak.
The signal: usage that has outgrown the plan. Track accounts pressing against their limits or leaning hard on their most valuable features. Those are expansion conversations you’re currently leaving to chance.
A five-minute diagnostic
You don’t need a new tool to start. You need to be honest about which of these five you actually measure. Walk the list:
- Signup: Do you know your started-vs-completed rate, or only how many finished?
- Activation: Have you defined the single event that means “this user got it”? Can you see how long it takes?
- Habit: Is there any event between activation and the paywall that you track on purpose?
- Paywall: Do you know what users did before they bounced off pricing?
- Expansion: Can you name the accounts that are ready to upgrade right now?
Every “no” on that list is a leak running with the lights off. The funnels that compound aren’t the ones with the flashiest top-of-funnel. They’re the ones where every one of those five questions has an answer, and someone is watching the number move.
That’s the whole idea behind how we built FullVision: not another dashboard of aggregate rates, but the full journey, so the leak has nowhere to hide.