Which process do you automate first? Five questions as a scorecard.
Every audit surfaces the same pattern: twenty ideas, no ranking. Most companies then pick the idea of the loudest voice or the most visible process. That's almost never the best first project. I use five questions, each scored one to five.
1, How often does it happen?
Twice a day is nothing; two hundred times a day is a business case. Automation has fixed build costs and almost no cost per run, so volume is the biggest factor. A process that occurs five times a week rarely earns back the build, however irritating it is.
2, How much variation is in it?
Does it run the same way every time, or does it depend on judgement and exceptions? Low variation is good news: the outcome is predictable and testing is simple. With high variation you don't automate the process but the groundwork for it: gathering, sorting and preparing data for someone who decides.
3, What happens when it goes wrong?
A misfiled email is annoying. A wrongly sent invoice is a problem. Wrong medical advice is unacceptable. The higher the risk, the more human checkpoints you must build in, and the less time you save. So start low-risk: your first project should build trust, not ask for it.
4, Is the data where you need it?
This is the quiet project killer. Is the information locked in a system you can't reach, in a spreadsheet on somebody's laptop, or only in the head of your longest-serving employee? Then your project is about unlocking data rather than about AI, and it takes three times as long. Pick something where the data is already in order.
5, Does the owner actually want this?
If the person doing this work feels they're automating themselves away, the project gets quietly resisted. Usually unconsciously: exceptions that “don't work”, feedback that never arrives. Talk it through beforehand and be concrete about what that person will do instead.
The arithmetic
Score each process one to five on volume, low variation, low risk, data availability and internal support. Take the highest total, not the nicest demo. Then add one figure: how many hours a week it costs now, and what that is per year in money. If the build price sits under the annual gain, you have your first project. One more rule: keep it small enough to be live within six weeks. A modest success returns more than a grand plan that stalls halfway.
The short version
Volume, low variation, low risk, data available, support present. Five scores, highest total wins. Live within six weeks, then the next one.
Shall we run the scorecard across your processes together?
Book your free intro call →