We're living in this unfortunate split where tools like Google Analytics tell you where people came from, and Stripe tells you who paid, but there's no simple way to connect the two.
As a solo founder building SaaS products, I kept asking myself:
“Which traffic sources actually drive revenue?"
“Which traffic sources bring in the highest-value customers?”
UTMs and custom attribution params get lost after signup. Stripe doesn’t track attribution. And wiring together GA, BigQuery, and Stripe is way too much for a small team.
So I built Boone — a lightweight attribution tool that fills the gap:
- One-line JS snippet captures UTM/referrer data
- Stripe Connect pulls in revenue events
- Boone links the two via customer email
- You see exactly which sources drive paying customers
And here's the fun part: once the data's in, you can actually chat with it. Boone includes an AI business coach (powered by GPT-4) that answers questions like:
- “Which channels are bringing in the highest LTV?”
- “Did Twitter traffic drop off this week?”
- “Where is most of my churn coming from?”
So now instead of staring at dashboards, you just ask your questions and get real, instant insights.
What was harder than expected:
1. Making the AI chat actually useful: The first version just repeated the stats: “$500 from Google, $200 from Twitter.” I had to iterate a lot on prompt engineering to get it to give real advice like “Double down on Google - LTV is 3x higher.” It now feels more like a growth-minded cofounder than a data parrot.
2. Resisting the urge to overbuild: There were so many tempting features: predictive modeling, graphs, anomaly detection, more insights. But I had to keep asking: “Does this help validate the core problem?” Saying no to scope creep over and over again was difficult.
3. Making the data actually connect end-to-end: Capturing traffic data → linking it to a signup → tying that to Stripe revenue seems simple. But real-world edge cases (like multiple signups, missing attribution, or webhook failures) ate up a surprising amount of time. The AI stuff gets all the attention, but the plumbing took just as much work.
4. Solving the cold start problem: Boone’s value lives in attribution — but when someone first signs up, we have none. Waiting for post-snippet traffic wasn’t acceptable, so we focused on delivering instant value from Stripe data alone. That meant surfacing insights like which products had the highest LTV or churn, and letting users chat with that data out of the box. It gave new users a reason to engage immediately, even before attribution kicked in.
Happy to answer any technical questions or go deeper on how it works.
- Lock
PaulHoule•14h ago
I think you don't want to advertise that it is an AI-related product. Right now HN feels really flooded by AI products which usually aren't that good, you might get a less positive response than you would otherwise by including that keyword.
Lrodd•13h ago
Thanks for the feedback. Leaving that aside, do you feel the overall value prop is useful?
PaulHoule•12h ago
Yeah, for someone who is using Stripe for e-commerce it looks like a good idea. That's not me.
Lrodd•14h ago
As a solo founder building SaaS products, I kept asking myself:
“Which traffic sources actually drive revenue?" “Which traffic sources bring in the highest-value customers?”
UTMs and custom attribution params get lost after signup. Stripe doesn’t track attribution. And wiring together GA, BigQuery, and Stripe is way too much for a small team.
So I built Boone — a lightweight attribution tool that fills the gap:
- One-line JS snippet captures UTM/referrer data - Stripe Connect pulls in revenue events - Boone links the two via customer email - You see exactly which sources drive paying customers
And here's the fun part: once the data's in, you can actually chat with it. Boone includes an AI business coach (powered by GPT-4) that answers questions like:
- “Which channels are bringing in the highest LTV?” - “Did Twitter traffic drop off this week?” - “Where is most of my churn coming from?”
So now instead of staring at dashboards, you just ask your questions and get real, instant insights.
What was harder than expected:
1. Making the AI chat actually useful: The first version just repeated the stats: “$500 from Google, $200 from Twitter.” I had to iterate a lot on prompt engineering to get it to give real advice like “Double down on Google - LTV is 3x higher.” It now feels more like a growth-minded cofounder than a data parrot.
2. Resisting the urge to overbuild: There were so many tempting features: predictive modeling, graphs, anomaly detection, more insights. But I had to keep asking: “Does this help validate the core problem?” Saying no to scope creep over and over again was difficult.
3. Making the data actually connect end-to-end: Capturing traffic data → linking it to a signup → tying that to Stripe revenue seems simple. But real-world edge cases (like multiple signups, missing attribution, or webhook failures) ate up a surprising amount of time. The AI stuff gets all the attention, but the plumbing took just as much work.
4. Solving the cold start problem: Boone’s value lives in attribution — but when someone first signs up, we have none. Waiting for post-snippet traffic wasn’t acceptable, so we focused on delivering instant value from Stripe data alone. That meant surfacing insights like which products had the highest LTV or churn, and letting users chat with that data out of the box. It gave new users a reason to engage immediately, even before attribution kicked in.
Boone’s live. I’m looking for early users to try it out and start tracking revenue across Twitter, LinkedIn, paid ads, or wherever you’re sending traffic. https://www.getboone.com/?utm_source=hackernews&utm_medium=s...
Happy to answer any technical questions or go deeper on how it works.
- Lock
PaulHoule•14h ago
Lrodd•13h ago
PaulHoule•12h ago