Support is the thing I planned to deal with later. For one person it decides what the product can be.

It is not a cost line, and it is decidable before you build anything.

The arithmetic that convinced me. Fix the customer count at 100 and sweep the minutes each customer takes per month. 5 minutes is 1.9 hours a week. 30 minutes is 11.5 hours. 60 minutes is 23 hours.

Thirty minutes per customer per month is not a crisis. It is one real exchange. And at a customer count most solo products would call a good start, that is more than a day of every week gone before you build or sell anything.

Two distinctions decide which row you are on, and neither needs usage data you do not have.

Is the ticket deflectable — can docs, onboarding or the product itself answer it so you fix it once — or does it need you, every time, forever. That second kind does not get cheaper with scale, it gets bigger.

And customisation risk: everyone gets the same product, or some clients need a setting nobody else uses, or each client needs work only you can do. The third rung is the definition of an agency, and nobody arrives there on purpose. You arrive one accommodating yes at a time.

Some evidence rather than just a warning: across 706 candidates our own gate killed on 23 real runs, hands-on support load was named in 187 of them — 26 percent, second only to the missing documented problem. That describes our gate on our own accounts rather than a market, but it is the rank I did not expect.

One product note that follows directly: every feature that needs explaining is a recurring cost priced in your weeks. The cheapest support work is the kind you do once in the product instead of every time in an inbox.

One honesty note about our own version: the contract explicitly tells it not to be over-pessimistic, because a gate tuned to find support problems everywhere would find them everywhere.

The sweep: https://whittleos.com/guides/will-my-saas-be-support-heavy