Automation makes things faster. It also makes mistakes faster, makes bad processes more efficient, and makes organizational dysfunction harder to see because the dysfunction is now running at machine speed. Before you automate, you need to know what you're optimizing for. Most companies skip this step.
The Automation Trap
The pattern is predictable: a manual process exists that is slow and expensive. Someone identifies that this process could be automated. The automation project is approved. The automation is built. The process is now fast and cheap. Leadership declares success. Three years later, the automated process is running at scale and nobody remembers why it works the way it does, why certain decisions are made the way they are, or what the original business logic was designed to achieve.
You have automated institutional knowledge out of your organization. The process runs efficiently. It may no longer be doing the right thing.
What Strategy-First Automation Looks Like
Before automating any process, a strategy-first approach asks: what is this process trying to achieve? Is it achieving that goal today? Is that the right goal, given where the business is heading? If the answers are clear, then automation accelerates good work. If the answers are unclear, automation should wait until they are clear — because automating an unclear or broken process at scale creates a problem that is much harder to fix than the original manual one.
“Don't automate a process you don't understand. Don't automate a process you're not sure you should keep.”
The Right Sequence
- Clarify: document the process as it actually works, not as it was designed. Identify what decisions are being made and why.
- Simplify: before automating, remove steps that don't need to exist. Automation of unnecessary steps is expensive overhead.
- Standardize: document the rules and logic. Automation requires explicit rules — the act of documenting them often reveals inconsistencies that should be resolved before they are baked into code.
- Automate: only now, with a clear, simplified, documented process, does automation deliver consistent value.
The Measurement Imperative
Every automation should have a measurement layer that tracks whether it is achieving its intended outcome, not just whether it is running. An automated onboarding process should be measured against customer activation rate, not just completion rate. An automated approval workflow should be measured against decision quality, not just decision speed. If you cannot measure the outcome, you cannot know whether the automation is working or just running.