Your First 90 Days of Automation: A Roadmap That Survives Reality
Most automation programmes stall in month two, when the easy win is done and nobody agreed what comes next. Here is the sequence that does not.
The first automation is usually a success - it was chosen because it was obvious. The second is where programmes stall: the obvious win is spent, the remaining candidates all look plausible, and without a way to choose, the effort dissolves back into daily work.
Days 1-30: measure, do not build
- List every recurring task that takes more than fifteen minutes and happens more than weekly. Ask the people doing them, not the managers.
- For each, record frequency, duration, who does it, and what breaks when it is skipped.
- Identify the three with the highest annual hours. It will not be the three you expected, and that surprise is the point of measuring.
- Check data quality for each. Half of automation candidates are blocked by data problems that must be fixed first.
Days 31-60: build one thing, completely
Pick a single process - ideally high frequency, low complexity, one owner - and automate it end to end, including the exception path and the handover to a human. A finished automation that handles reality builds confidence; three half-finished ones build scepticism that costs you the rest of the programme.
Days 61-90: prove it, then decide
- Measure the same numbers as month one for that process. Actual saving, not estimated.
- Ask the people affected what got worse, explicitly. Something always does, and hearing it early is what keeps them engaged.
- Document what you learned about your own data and processes - that knowledge is a real asset for everything that follows.
- Choose the next two from the measured list, not from whatever became irritating this week.
What derails it
Starting with the most complex process because it is the most painful. Automating around an organisational problem instead of fixing it. And building without an owner: every automation needs a named person who notices when it breaks, because it will, quietly, on a Tuesday. This is the shape most of our process automation engagements take.
- 30 days measuring before building
- 1 process automated completely
- 1 named owner per automation
Frequently asked questions
What if we cannot spare 30 days for measurement?
Then measure for two weeks - but do not skip it. Companies that start building immediately routinely automate a process that was about to change, or one whose real cost was a tenth of what they assumed.
Should we hire someone or use an agency?
For the first three automations, external help is usually faster and teaches your team what the work involves. Hire internally once you have a continuous pipeline of work rather than a project - that is the point where the economics flip.
More on this topic: Growth & Strategy.
Keep reading
Want this built for your business? See what we do.