Your First Hundred Customers Should Be Manual

Onboarding fifty customers personally teaches you which steps confuse people, which questions recur, which stages actually matter and which can be skipped. Automating first bakes in your assumptions about all of that, and assumptions at this stage are usually wrong in ways that are expensive to unwind.

Automate the process you have observed, not the one you imagined.

Track where the manual hours go. The first thing to automate is whatever consumes the most time and is most repetitive, which is frequently not what the team expected. Building in that order means every hour of engineering removes an hour of known, measured cost.

A spreadsheet, a shared inbox and a diligent person handle a surprising volume — often well past a hundred customers. The pressure to automate usually comes from a feeling of inelegance rather than from an operational limit that has actually been reached.

Handling customers directly puts founders in contact with objections, confusions and reasons for churn in a way no dashboard reproduces. Automating early removes that channel exactly when the information is most valuable and least available elsewhere.

The signal is not that the process feels amateurish; it is that it is failing — things being missed, delays growing, a person becoming a bottleneck. Wait for that signal and you will build the right thing, because by then you will know precisely what breaks.

Even at scale, keep a route for the cases automation cannot handle. Products that automate completely tend to fail their most complicated and often most valuable customers, who then leave without a clear explanation of why.