Before You Start Your AI Project, Measure What Matters

One of the first questions I ask organisations before they begin an AI project has nothing to do with AI.

It’s much simpler than that.

“What are you measuring today?”

The question often catches people off guard.

We’ve already started talking about AI strategy, customer experience and automation, yet I’m asking about numbers that already exist inside the business.

There’s a good reason for that.

Because if you don’t know where you’re starting from, how will you know whether AI actually made a difference?

You Can’t Improve What You Don’t Measure

Imagine implementing an AI customer service assistant.

Six months later, someone asks:

“Has it been successful?”

How would you answer?

Is success measured by the number of conversations?

Response time?

Customer satisfaction?

Sales?

Employee productivity?

Without a baseline, the discussion quickly becomes subjective.

People start saying things like:

“It feels faster.”

“Customers seem happier.”

“The team thinks it’s helping.”

Those are useful observations.

But they’re not evidence.

The Numbers Tell the Story

One of our clients took a different approach.

Before implementing AI, they measured how many customer enquiries they were handling every month.

The number was around 6,000 enquiries.

After implementing the solution and continuing to refine it, that number grew significantly.

Today, they’re handling approximately 30,000 enquiries every month.

That’s a fivefold increase.

The numbers didn’t just demonstrate that the AI was working.

They did something even more important.

They gave the organisation confidence.

Confidence that the investment was worthwhile.

Confidence that customers were benefiting.

Confidence that the business had the capability to take on even bigger initiatives.

That confidence matters.

Because success creates momentum.

Success Builds Belief

Interestingly, the most valuable outcome wasn’t simply handling more enquiries.

It was what happened afterwards.

That client is now working on their fourth AI project.

Not because AI is fashionable.

Not because they want to experiment.

Because every successful implementation has given them the confidence to tackle the next challenge.

That’s something I see repeatedly.

Organisations rarely transform because of one large AI project.

They transform through a series of successful projects that build trust over time.

Each success makes the next decision easier.

AI Is a Journey, Not a Single Project

Many businesses approach AI as though it will be one large transformation.

In reality, the organisations I’ve seen achieve the greatest results tend to take a different approach.

They start with a problem that matters.

They define what success looks like.

They measure it.

They implement.

They learn.

Then they repeat the process.

Each project teaches the organisation something.

Each project strengthens internal capability.

Each project builds confidence for the next.

Over time, AI stops feeling like an experiment.

It simply becomes part of how the organisation improves.

Measure Before You Build

One of the biggest mistakes organisations can make is waiting until after an AI project goes live to think about success.

By then, it’s too late.

Success should be defined before the first prompt is written, before the first workflow is automated and before the first chatbot goes live.

Ask yourself:

• How long does this process take today?
• How many enquiries are we currently handling?
• How much time do employees spend on repetitive work?
• How quickly do we respond to customers?
• What does customer satisfaction look like today?

Those numbers become your starting point.

Without them, it’s difficult to know whether you’ve improved or simply changed the way work gets done.

Final Thoughts

I’ve found that successful AI projects don’t begin with technology.

They begin with clarity.

Clarity about the problem you’re trying to solve.

Clarity about what success looks like.

And clarity about how you’ll measure it.

Because AI shouldn’t be judged by how impressive the technology is.

It should be judged by whether it makes the business measurably better.

So before you begin your next AI project, ask one simple question:

What are we measuring today?

Measure first.

Then improve.

Because the organisations that succeed with AI aren’t just the ones that build the technology.

They’re the ones that can clearly demonstrate the value it creates.