Choosing your first automation

Your first automation has to prove the idea works, not just remove your least favourite task. Four tests for picking the process most likely to succeed.

Once a business decides to try automation, the next question is where to start, and the instinct is usually wrong. People reach for the process that annoys them most, or the one a senior person wants gone. Neither is a good way to pick a first project.

The first automation has a job beyond saving time. It has to prove to your team that this works. So you want the one most likely to succeed cleanly, not the one that would feel best to be rid of.

Look for high volume before anything else

Automation pays back on repetition. A task done fifty times a day is a far better candidate than one done twice a week, even if the twice-a-week task is more painful each time. Count how often each candidate process runs across a real month, and let the volume do the ranking.

High volume also means you find the problems fast. If something is off, you will see it on day one instead of discovering it three weeks later when the process next comes around.

Favour rules over judgement

The best first automation follows clear rules that you could write down and hand to a new starter. If this, then that. Data entry, sorting, matching, routing, sending a standard document when a standard trigger happens. These are the tasks where a machine shines and rarely surprises you.

Save the judgement-heavy work for later, once your team trusts the approach. A first project that depends on nuance is a first project likely to disappoint.

Pick something with a clean edge

The ideal starter process has a clear beginning and a clear end, and does not tangle into six other systems. A task that starts when an email arrives and ends when a record is updated is contained. A task that touches your accounting, your CRM and a supplier portal in one breath is not a first project, it is a programme of work.

Contained processes are cheaper to build, easier to test, and far easier to point to afterwards and say, that clearly works.

Make the win easy to see

Choose something where success is obvious to the people around it. Hours given back, a backlog that clears, a report that now arrives on its own. A visible early win buys the goodwill and the budget for the harder projects that follow.

High volume, rules based, contained, and visibly useful. A process that ticks those four boxes is almost always the right place to start, whatever your gut says about the one that annoys you most.