Strategy and governance
Before you automate: deciding what to fix first
A sequencing method for choosing your first automation
Context
Most businesses arrive with a list of automation ideas, not a single problem. Everything feels urgent, so the loudest idea usually gets built first.
Choosing badly is expensive in a way that rarely shows on an invoice: a first project that stumbles drains the team's appetite for the second one.
Sequencing is therefore a strategy decision, not a technical one — and it deserves the same rigour as the build itself.
What changed with AI systems
AI has widened what can be automated, which makes the queue longer, not shorter. More candidates means more ways to start in the wrong place.
Modern tooling makes a first version deceptively fast to ship. The constraint has moved from building to choosing.
The businesses getting real returns treat the first project as proof of a working method, not as the biggest possible swing.
How to approach this in your organisation
- 01Map the workflows candidates actually touch: who does the step today, how often, and what happens when it goes wrong.
- 02Weigh frequency against failure cost. High-frequency, low-drama work is the natural first target.
- 03Pick a first project the team will feel weekly — visible relief builds the mandate for harder work.
- 04Defer workflows dominated by edge cases and judgement calls until the operating rhythm exists.
- 05Review after the first month of live running and let the results reorder the remaining list.
Key metrics
- Hours returned to the team each week by the first automation.
- Exceptions raised versus items handled without intervention.
- Hand-off count: how many times work still changes hands in the automated flow.
- Team sentiment about the system after a month of real use.
Risks to consider
- Automating a process that was broken to begin with — the automation inherits the flaws faster.
- Starting with the flashiest workflow instead of the most frequent one.
- Designing around the people who run the process today instead of with them.
- Skipping the rollback plan because the pilot went smoothly.