Most of the expensive mistakes I have seen in software were not coding mistakes. They were decisions that nobody could explain clearly before the work started.

So I use a simple test before any significant change, whether it is a new service, a schema change or a rewrite: can I explain what we are doing, and why, in two sentences?

If I can, we build. If I can't, the problem isn't understood yet, and writing code will only hide that for a few weeks.

Why two sentences

The first sentence says what changes. The second says why it is worth the cost. That is all a reviewer, a product owner or a new engineer really needs to start.

When the explanation needs a third sentence that starts with "but", or "except when", that is usually where the real design work is hiding.

What the test catches

  • Solutions looking for a problem. If the "why" sentence is vague ("it will be cleaner"), the change can usually wait.
  • Two changes pretending to be one. If the "what" sentence has an "and" in it, split the work.
  • Missing owners. If I can't say who notices when it breaks, it isn't ready for production.

Where the thinking happens

Very little of this happens at a keyboard. It happens in the ten quiet minutes before a meeting, on a walk, or sitting with a notebook. Teams sometimes treat that time as idle. I think it is the cheapest engineering we ever do.

Try it this week

Pick the next change on your board and write the two sentences at the top of the ticket. If you can't, talk it through with someone before anyone opens an editor. You will throw away less code.

I'm Pratyush Shivam, CTO at Konfidio in Berlin, where we build TrustedTrucks. More about my work at pratyushshivam.com.