Which AI ideas should you start with? How we prioritise

When we meet a new client, there is almost always a long list of ideas for AI and automation. Some come from management, some from employees and some from a supplier or a conference. The problem is rarely a lack of ideas. It is choosing the right ones.

That is why the report after a Shadow Walk includes a prioritised list of your opportunities for AI and automation. Here is how we make it.

1. How big is the impact?

We start by looking at what a solution will actually change. How many hours does it save per week? How many errors do you avoid? Can you answer customers faster or take on more orders? We start from what we saw during the Shadow Walk and from your own numbers, so the impact is not just a guess.

2. How big is the effort?

Next, we look at what it takes. Not just building the solution, but also getting the data in place, connecting it to your systems and teaching employees to use it. A solution that needs five systems to talk to each other is a very different task from one that works in a single system.

3. How realistic is it to put into use?

The third question is often forgotten: Will people use it? A solution can have a big impact on paper, but if it does not fit into the working day, it will not be used. That is why we look at who will use the solution, where they sit and what they need. It is one of the reasons we follow employees in their work instead of only talking to management.

From list to plan

Once each idea has been assessed, we place them in a simple model with four boxes: high or low impact and high or low effort. It gives an overview everyone can understand:

• High impact, low effort: Start here.
• High impact, high effort: Plan them, but not as the first thing.
• Low impact, low effort: Good to include when there is room.
• Low impact, high effort: Leave them.

Based on the list, we recommend what you should build first, with a clear rationale, a defined scope and concrete success criteria. That way everyone knows what it takes for the first project to be a success.

It is not about finding the smartest AI. It is about finding the place where it makes the biggest difference for you.

If you want your own prioritised list, a Shadow Walk is a good place to start. It takes ½–2 days at your site, and about a week later you have the report.