How to pick your first automation project without wasting a quarter

The most common mistake with a first automation project is picking something too ambitious: a full customer journey overhaul, a company wide AI rollout, a system that touches every department at once. These projects are exciting to talk about and slow to deliver, and a slow first project tends to be the last one a business attempts.

A better approach is to pick something small, specific and already visibly annoying. Ask around: what task does someone dread doing every week? What report takes longer to build than it should? Where does the same piece of information get typed into more than one system? These are usually small in scope and high in daily friction, which makes them ideal first projects: fast to build, easy to measure, and immediately noticeable to the person who used to do it manually.

A good first project also has a clear before and after. Before: someone spends forty minutes every Monday building a report by hand. After: the report generates itself, and that forty minutes goes somewhere else. That comparison is what builds internal confidence in automation as a way of working, which matters more than the project itself, because it's what makes the second and third projects easier to greenlight.

Once the first project is live, the natural next step is usually obvious. The work of finding it disappears, because the team has already started noticing where else the same kind of friction shows up.