Fixie earns less than any client project we take on. It teaches us more than most of them.
Fixie is our own goal-tracking app. It has 400 and some users, a free trial, and an annual subscription. By agency math it is a bad deal: the same hours billed to a client would earn a multiple of what Fixie brings in. We keep running it anyway, and we think every studio should run at least one product it cannot bill anyone for.
The structural problem with agency work is that it ends. You design, you build, you hand over, and the results of your decisions land on someone else’s desk. If the onboarding flow leaks users, the client feels it. If the pricing is wrong, the client feels it. The agency has moved on to the next project, and its beliefs about what works never get tested against a churn graph.
Fixie is where our beliefs get tested. We picked the pricing. We wrote the onboarding. We decided which features to build and which to skip. When a decision is wrong, nobody softens it for us: the trial-to-paid conversion just sits there, being what it is. That feedback loop is uncomfortable in a way client work almost never is, and the discomfort is the point.
The clearest lesson is what retention really hangs on. Fixie’s monthly look-back, a Spotify-Wrapped-style review of your past month, is the reason people come back. We did not know that when we built it. We suspected it, we shipped it, and the usage told us. Meanwhile features that felt obviously necessary, more settings, more customization, more goal types, moved nothing. We would have defended some of those features in a client meeting. Running the app ourselves is what proved us wrong.
The second lesson is that a subscription is a promise, not a transaction. Someone who pays for a year expects the app to be alive: updated, responsive, slightly better each month. The one-time-purchase mindset that agency projects quietly train into you does not survive contact with a renewal date.
The third lesson is about support. When a product has 400 users, every complaint is from a person you could name. That scale is a gift. Patterns are visible by hand, and a single conversation can change the roadmap. We now tell clients with early products to stop hiding behind dashboards and read the individual messages, because we learned to do that ourselves.
Our own product loses the scheduling fight against paying clients. Every time. A client deadline is a date with a person attached; Fixie’s roadmap is a promise we made to ourselves, and promises to yourself are the easiest ones to break. There have been stretches where Fixie sat untouched for weeks because invoiced work came first. The fix was not discipline, it was structure: Fixie gets planned like a client, with scoped work in the calendar, or it does not happen at all.
We should also be honest about what Fixie is not. It is not a growth story we can put in a pitch. It grows slowly, because we spend nearly nothing on marketing and our distribution skills lag our product skills. That gap is itself a lesson we pass on: founders underestimate distribution because builders always do, and we are builders.
When we tell a founder that their settings screen can wait, or that their pricing needs a decision rather than a survey, or that retention lives in one feature and not ten, that advice comes from operating a product, not from reading about one. Clients can tell the difference. Fixie’s subscription revenue is small. What it buys is the right to say “we run one of these ourselves”, and in our line of work that sentence is worth more than the invoice we could have sent instead.
We build websites, software and companies. Tell us which one you need.