The question that ends the call
A founder told me he was doing three to four discovery calls a week. People were booking through his website. Some were using the free product. A few were even copy-pasting outputs into their actual workflow during the call. He was excited. He wanted to talk about which channel to scale next.
I asked him one question: of the people who'd done discovery calls with him in the last two months, how many had come back to the product without him prodding them?
Silence.
He didn't know. He hadn't tracked it. When we walked through it together, the answer was maybe two out of thirty.
This is the post version of a conversation I have every other week. Founders bring me "we're getting interest" and want help scaling it. The harder question - whether the interest is real or whether it's noise that resembles real - usually doesn't get asked, because the answer is unpleasant. This is one of the more painful versions of what I called the calls-not-converting bottleneck: the calls aren't bad, they're just not what they look like.
Real interest and noise look identical from the outside for the first six to twelve months of an early-stage SaaS company. The metrics overlap. People show up to calls. Some of them say nice things. A few of them might even pay you. If you don't run the cheap experiment that tells you which is which, you'll spend the next year scaling the wrong thing, and find out only after you've spent the money and burned the time.
What real early traction looks like
First, the version that is real. This post isn't about ignoring early signal. Some signal is real.
Real early traction has a specific texture. People come back to the product without you reminding them. They use it on days you didn't email them. They tell other people unprompted. They ask you to charge them before you've asked them to pay. They get frustrated when something breaks, not because the experience is bad, but because they were relying on it.
The most reliable tell: someone uses your product, and then a week later uses it again, and you did nothing in between to make that happen.
That's signal. That's the thing that compounds. If you have that, with any consistency, you don't need this post. You need to figure out how to find more people like them.
This post is for the much more common case: the founder who has discovery calls, engagement, even occasional paid conversion, and isn't sure whether any of it means anything yet.
What noise looks like when it mimics traction
The noise version looks like this.
People book discovery calls because they're curious. They've seen something on LinkedIn, or a friend mentioned you, or they came across your site and thought "huh, what does this do." They take a 30-minute call. They're polite. They say things like "this is interesting." They might even agree to try the free version. Maybe they log in once. Maybe they don't.
Free users use the product because it's free. The cost of trying is zero, so the bar for "engagement" is basically nothing. They open the dashboard, run a few queries, look at the output, maybe copy it into a deck or a Slack message. Then they close the tab.
Some users come back. Not because the product is essential to their work, but because they remembered it existed and had a thing to do. They run another query. They close the tab again. This isn't a habit. It's a tool they reach for occasionally, like a free online image compressor.
A small number of these people convert to paid, especially if you ask them directly. They say yes because they like you, because the price is low, because they figure they'll get value out of it eventually. They use it for a month. They churn. You never quite figure out why.
This is what noise looks like in B2B SaaS. It's not bad people. It's not even a bad product. It's just engagement that doesn't compound into anything that grows on its own.
From the inside, on a Tuesday afternoon when you're checking your dashboard, this looks indistinguishable from real traction. The numbers move. People do things. Some of them pay you. You feel like you're making progress.
You might not be.
Why founders fall for this every time
Three reasons.
Humans are nice in meetings. A founder takes a 30-minute call with someone. The someone doesn't want to make it awkward. They say "this is really useful" and "I'm definitely going to try it" and maybe even "I could see this fitting into our workflow." None of those statements predict whether the person will be using the product six weeks from now. But they feel like predictions, because in any other context that's how people talk when they mean it.
Engagement metrics look identical whether it's curiosity or commitment. A user who runs five queries because they're seriously evaluating the product looks the same in your dashboard as a user who runs five queries because they were procrastinating from something else. You can't tell them apart from the data alone. You can only tell them apart by asking.
Founders are pattern-matching for survival. They want this to be real, because the alternative is uncomfortable. So they see the version that's real. They notice the user who came back twice and they don't notice the twenty-eight users who never opened the product after the first session. This isn't dishonesty. It's the same bias that makes you read your own writing as clearer than it actually is.
The result of those three things: founders run for months on signal that isn't there, because the noise looks like signal and the experiments that would distinguish them haven't been run.
The cheapest experiment in B2B SaaS
There is one experiment that converts the noise into signal in about three hours of work. I give it to almost every founder I diagnose who has free users and is trying to figure out whether to charge.
Pick 30 of your most active free users. Not people who signed up and disappeared. People who've actually used the product more than once in the last month.
Write them a short personal email. Two or three sentences. Tell them you're considering pricing the product at €25 a month starting in two weeks, and you want to know if they'd keep using it at that price. Ask them to reply with "mostly yes" or "mostly no." Anything else is fine, but those two words are the ones that matter.
Send it on a Tuesday morning. Wait three days.
What you get back is the signal you've been trying to extract for months.
If 18 out of 30 say "mostly yes": you have something. The product is solving a real problem for a real customer. You don't need to scale calls. You don't need to add channels. You need to start charging.
If 6 out of 30 say "mostly yes": you have a more complicated situation. The product is useful to some people, but the people who think it's useful aren't a coherent enough segment to build a business on yet. You need to figure out what's different about the six, then go find more like them.
If 1 out of 30 says "mostly yes": you've learned the thing the discovery calls weren't telling you. The engagement was noise. It wasn't a lie - those people were really using the product. They just weren't using it because it solved a problem worth €25 a month. The strategy needs a rethink, and you've found that out in three days instead of six months.
Most founders avoid this experiment for one reason: they're afraid of the "mostly no" answer. The "mostly no" answer is exactly what they need most. It's the only thing that tells them where the real work is.
What happens when you scale on noise
The founders who don't run this experiment end up doing the same thing every time. They look at the engagement, they decide it's traction, and they go to scale it.
They add channels. They run paid ads. They hire someone for outbound. They invest in SEO. They push product updates that should, in theory, make engagement deeper. All of these things are reasonable when you have signal. They are catastrophic when you have noise.
What happens is the noise scales with them. Now they're getting twelve discovery calls a week instead of four. Their free user base is growing 30% month over month. By every dashboard metric, the business is working. And the paid conversion rate is unchanged.
Eventually they realize the people they did convert to paid (the early ones, the ones who said yes because they liked the founder) are the wrong customers. They're using the product in ways the founder didn't intend. They're asking for features the founder doesn't want to build. Rolling them off the platform is harder than not having converted them in the first place, because now they have expectations and contracts and slack threads.
A founder told me, on a call, that some of his earliest paying customers had turned out to be the worst thing that happened to his company. Not because they were bad people. Because converting them had locked him into a version of the product he didn't want to build, and getting out of that was going to take him a year.
The hardest part: by the time you find this out, you've lost six to twelve months you could have spent figuring out who the right customers actually were. You can't get those back.
The diagnostic question to ask yourself today
If you have any free users, even a handful, you can run the willingness-to-pay test this afternoon. The email takes ten minutes to write. The replies take three days to come in. The decision the answer leads to is bigger than any growth experiment you've been planning.
The question to ask yourself first is the one I asked the founder I opened this post with: of the people who've engaged with my product in the last sixty days, how many have come back without me prodding them?
If you don't know, find out. If the answer is most of them, you have signal. If the answer is fewer than you hoped, you have a free, fast experiment available that will tell you whether to scale, whether to pivot, or whether to charge.
What you don't want to do is the thing most founders default to, which is to assume the engagement is real, scale the channels that produced it, and discover six months later that the only thing you scaled was the noise.
Fix the funnel before you scale the volume. Even a 1-in-100 conversion rate doesn't get better when you turn it into 1-in-1000. It just gets bigger.
