How to choose which business process to automate first

When we sit down with a business owner, the first thing they usually mention is the task that winds them up. Often it is something they do themselves, at the end of a long day, and it sticks in the memory for that reason.

Meanwhile the real drain is often a quiet job someone in the team does every morning without complaining. Nobody raises it because it has always been done that way. Choosing by frustration means you can spend money automating something that saves an hour a month while the big one carries on. A repeatable method takes the guesswork out.

Start by writing down every task the team does in a normal week. Not just the painful ones. Ask each person to keep a rough list for five working days: what they did, roughly how long it took, and whether it happens again next week. You will be surprised how much of a week is the same work repeated.

Then score each task out of five on three things. Keep it rough. You are ranking, not auditing.

Across different industries, the same kinds of work keep rising to the top of the list. They have one thing in common: they follow rules. The same inputs go in, the same steps are followed, and the output looks much the same each time. Software is very good at that kind of work and does not get bored or make typing errors at four in the afternoon.

Before you build anything, look at where the inputs for your top task actually live. If the details for one letter come from three spreadsheets, an inbox and somebody's memory, automation has nothing reliable to work from.

Some tasks need judgement on every single case. Deciding whether to take on a client, approving an unusual deal or handling a complaint all rely on experience. These are poor first candidates for full automation, and pretending otherwise causes problems.

That does not mean leaving them alone. Keep a person making the decision and automate the preparation around them: gathering the information, drafting the paperwork and sending the outcome once it is approved. The decision stays human, and the hour of digging before it disappears.

PreAction, a B2B debt mediation business we work with, is a good illustration. Their documentation scored high on all three counts. It happened constantly, it took a long time, and every document followed the same rules using details already held on the job. Before automation, it took two days for every hundred records, and added up to two to three months of working time each year.

We first moved their records out of a limited old system into one database, so every document had a single, reliable source. Then we automated the documentation end to end. It now runs from a single button.

Write down how long the task takes before you change anything. Time it over a normal week. After the automation goes live, measure it again the same way. Without a before figure you are relying on a feeling, and feelings are easy to argue with. Once the numbers show it worked, go back to your scored list and pick the next one.

This is close to how we work with clients. It starts with a free initial consultation. If there is a fit, we run a paid discovery of one to five days, where we map the processes, score them and check where the data lives. That produces a full software development document before any build begins, so you know exactly what is being made and why it comes first.