How do you know when to shut down a side project?

You know by writing it down before launch, while you're still hopeful: by this date, this number, or I close it. The version of you on day 60 can't be trusted with that decision, because by then you'll have every reason to say "nearly there". I kept one product alive for 36 months after I knew it was dead.

The thank you in the payment reference

One customer paid by bank transfer, on the first of every month, and put "thank you" in the payment reference. Every month, for 3 years. She was a driving instructor in her sixties. My product texted her pupils the evening before each lesson, and she told me it saved her about a lesson a week. It's the nicest thing anybody has said about anything I've built.

She's also the main reason I kept it running. (Kindness is a terrible reason to keep a product alive. It's also the commonest.)

By month 3 I knew

I'd found instructors through a public directory and written to about 40 of them. Within 2 months a handful were paying. That was the high point. New instructors stopped arriving, and the places where instructors talked were associations and private groups I had no way into.

The honest move was to give the handful 3 months' notice, refund anybody who wanted it, and stop. I started a list instead. By month 14 it had 40 items on it. Every Sunday I opened it, added 2 or 3 things and closed it again, while the product took an evening a month just to stand still. It lived 39 months in all.

The exception

Some flat products should stay. My test for one that runs itself: no unplanned work for 8 weeks, under 40 minutes a week, and its costs covered. That one can stay without me. The driving instructors' tool failed on the evenings, every month, for 3 years.

Do it today

Open the file for the thing you're building and write one sentence: by this date, this number, or I do this. Then tell somebody where the file is.

Nobody Bought It has the autopsy. Ship Alone has the kill rule in full.