What our paying customers have in common (and what it changed about our funnel)

I went through every paying customer we have and looked for what they share, instead of guessing. The patterns changed how we think about the funnel. Sharing in case it is useful.

What paying customers had in common, that free users who churned did not:

• Fast activation: Almost everyone who pays installed the snippet and saw their first number within the first session or two. The ones who took days to install rarely came back to pay.
• Connected revenue: They connected revenue, not just traffic. The people who pay are the ones who hooked up the stripe side and saw money attributed to a source. Once they saw "this channel made you X", they stuck.
• Specific questions: They had a specific question, not general curiosity. "Which campaign drove sales" converts. "Let me check out an analytics tool" mostly does not.
• First week return: They came back within the first week. First week return was the single clearest line between pays and churns.

What this changed:

• Segmented follow ups: We stopped treating all signups the same. A signup who activates fast and connects revenue gets a different follow up than one who poked around and left.
• Earlier revenue connection: Our onboarding now pushes toward the revenue connection sooner, because that is the moment people commit.
• Specific messaging: Our messaging leads with the specific question (where does your revenue come from) instead of the general category (analytics).

What did not predict paying, surprisingly:

Company size, traffic volume, and how many features they tried. Some of our happiest payers use a fraction of the product. Using more features did not mean paying.

The honest meta lesson: Your paying customers are a dataset you already have. I spent weeks guessing what would convert people when the answer was sitting in our own customer list.

Two co founders, bootstrapped, tool is Zenovay. Happy to share how we segment the follow ups.

For the folks here who sell self serve: What is the earliest signal in your data that predicts someone will pay?