Skip to content

Process

When not to automate

The fastest way to earn trust is to talk someone out of work they do not need. Including the automation that will never pay itself back.

AP

Alex Pavlov

May 4, 2026 · 5 min read

I have talked more clients out of automation than into it, and it is the best marketing I have ever done by accident. The fastest way to earn trust is to tell someone not to spend money, especially on the shiny thing they came in asking for.

Automation has a payback period, the same as a delivery van or an espresso machine for the office. If the thing you want to automate happens twice a month and takes ten minutes, the spreadsheet has already won. Automating it would be a hobby, and I have enough of those.

The three questions before we quote

  1. How often does this actually happen? Real number, not the dramatic one.
  2. What does it cost when a human does it, including the cost of doing it wrong?
  3. What will it cost to run and maintain the automation after the invoice is paid?

If the maintenance cost is bigger than the manual cost, we have just designed a more expensive way to do the same task, with the bonus of a pager that goes off at 2am. That is not a win. That is a subscription to a problem.

What is worth automating

High frequency, real stakes, and a stable process. The thing your team does fifty times a day, where a mistake is expensive and the steps do not change every week. That is where software earns its keep. The smoke tests, the reconciliation, the intake that gets duplicated across five systems by hand. Automate the boring high-volume work, leave the rare judgment calls to people.

The honest sign-off

I have automated other people's expense reports for a living and still not automated my own. The cobbler's kids, and so on. If you have tried everything and the manual process is still eating your week, call us. If it is eating ten minutes a fortnight, keep the spreadsheet and buy yourself a coffee with the difference.

Frequently asked

How do I decide whether a task is worth automating?

Compare three numbers: how often the task happens, what it costs when a person does it including errors, and what the automation costs to build and maintain. If maintenance outweighs the manual cost, the automation is not worth it.

What kinds of work are best to automate?

High-frequency, high-stakes work with a stable process: things done many times a day where mistakes are expensive and the steps rarely change. Rare judgment calls are usually better left to people.

Have a workflow that needs this?

Tell us the shape of the problem. Scoped estimate, usually within 3 to 5 business days.

Estimate project