Support is the specification
Running Wodloop means reading the support inbox. A box owner writes at 6 a.m. because a member could not book; the fix is a line of code and a reply, but the lesson is larger: every feature we shipped that nobody wrote in about was a feature nobody used. Since then the first question on a studio project is not 'what should it do' but 'what will they write in about'. The answer decides the first release.
It also killed a habit: the settings page. Products we run have few settings because every one of them is a support ticket waiting to happen. Products we build for others now start with none, and gain one each time a real client asks twice.
The second product is the cheap one
Wodlane cost a fraction of Wodloop because the world was known: the athletes, the coaches, the rhythm of a weekend, the payment habits of a box. m2 cost a fraction of Agora for the same reason: the public buyer was known, the questionnaire was known, the instance-per-client architecture was already paid for. The expensive part of a product is learning the trade, and it only has to be paid once per trade.
That is why the Lab's studio work now leans toward trades we already carry, public operators, training, sport, and why a first project in a new trade is quoted as the learning it is.
What we no longer build
Invoicing, payroll, e-mail sending, PDF rendering, authentication from scratch: we rent them, and we have stopped apologising for it. Four products taught us where the margin is, the two or three workflows that are the trade, and where it is not. A studio project that asks for the commodity parts custom gets a polite question about why, and usually a smaller quote.
The corollary is what we do build every time: the data model, in the client's own words, and the back office the operations team will live in. Those are never commodities, and they are where the four products earned their keep.