I’m building the referral tool I always wanted, in public, starting now

I didn’t either. For years I ran marketing where the happiest customers were the quietest. They’d finish a project thrilled, tell me how good it was, and then nothing. No referral. Not because they wouldn’t. Because I never asked, and when I did I asked weeks late, by which point the warmth had cooled and the ask felt like begging.

That’s the whole problem in one line. Referrals don’t fail because people won’t refer. They fail because nobody asks at the moment of peak goodwill, and asking is friction.

So I’m building the thing that fixes it. In public. From zero. I’m not an engineer. That’s the point.

What I’m building

A tool that watches for the signal that a customer is happy right now, then fires the referral ask automatically, in that window, with copy that doesn’t make anyone cringe.

For the first build I’m doing it on Shopify, because the happiness signal is clean there. Delivery confirmed, no support ticket raised, a repeat order, a five-star review. Those are receipts that a buyer is in the good mood. The tool catches one of those events and sends a templated, on-brand referral request while the goodwill is hot.

No new behaviour from the merchant. No remembering to ask. The unscalable human act, the thing a great salesperson does on instinct, run by a system.

Why I’m the one building it

I’m not going to pretend I can out-engineer anyone. I can’t write the backend a real developer would. What I can do is see how the pieces should fit, then build it anyway with AI doing the parts I can’t.

That’s the show. A non-technical operator shipping real software, showing you every move, including the ones that break.

I’ll give you the trigger logic, the event I listen for and why. I’ll give you the referral copy that converts and the versions that flopped. I’ll show you the Shopify billing setup, the bits that fought me, the cost of each AI call. When something doesn’t work I’ll show you that too, because the teardown is worth more than the win.

This week

By Wednesday I want one thing working. The app installed on a test store, listening for one signal, firing one referral email. Not polished. Alive.

Everything after that is in the open. You’ll watch me go from one firing trigger to something a real merchant pays for, or watch me find out it doesn’t work. Either way you see the whole thing.

Founding members

I’m opening a founding tier today. $150 for the year.

Here’s what it is, straight. Right now it buys you this build, live, with all the artefacts as I ship them. The code, the trigger logic, the copy that works, the billing teardown, the numbers. It also locks you into every build after this one at the founding rate while the public price climbs.

Here’s what it isn’t. It isn’t a finished product. If you want a polished tool to install today, this isn’t that, and I’d rather you waited. This is for the operators who want to watch the machine get built and take the parts for their own work along the way.

Twelve months from now the founding members will have the full archive of every build, every system, every teardown. That’s what $150 buys. Not issue two. The trajectory.

[Become a founding member]

Next

Wednesday I’ll show you the first firing trigger. The signal I chose, why I chose it, and the one I rejected that I think most people would have picked by mistake.

That rejected signal is where most referral tools get it wrong. More then.