The cheapest process step is the one you delete

There is an order of operations for making work better, and almost everybody starts at the wrong end of it.

The order is: delete, simplify, automate. In that order, because each step is cheaper, faster, and less risky than the one after it, and because automating something you should have deleted is the most expensive way to keep it forever.

Deletion is the only free win

An automated step still costs something. It has to be built, it has to be maintained, it breaks when the system upstream changes, and somebody has to understand it when it does.

A deleted step costs nothing. It cannot break. It cannot drift. It does not need a runbook. It is the only change that makes the system genuinely simpler rather than merely faster.

How to find the deletable steps

Ask, of every step: who reads this, and what do they do differently because of it?

Not "who receives it", who reads it. And not "is it useful", what decision changes. A weekly report that nobody has ever acted on is not a low-value step. It is a zero-value step with a maintenance cost, and it has been quietly charging you rent for years.

You will find three kinds of thing:

• Steps for a decision nobody makes any more. The situation they were built for ended; the step did not.
• Steps that exist to check another step. Often the cheaper fix is to make the first step trustworthy.
• Steps that exist because someone once asked. They asked in 2021. They have left.

Every process accumulates steps the way a house accumulates cables behind the TV. Nobody added a useless one on purpose. Nobody has ever pulled one out either.

Then, and only then, automate

What survives that cull is real work, and real work is worth automating properly. But you will be automating a smaller, clearer process, which means less to build, less to maintain, and less to go wrong.

That is the whole argument. It is not glamorous, and it is not what an agency is supposed to say when it would rather sell you a build.