You've embraced AI. Now where's the money?

In my last post I said that handing out Copilot licences and hoping for efficiency wasn't going to work. Since then I've had a lot of conversations with business leaders who are feeling somewhat underwhelmed with deployments so far.

So why exactly is AI failing to live up to the hype?

The problem isn't the technology

Most businesses have now spent something on AI: licences, tools, a pilot or two, maybe some training. But if you ask the CFO what that spend has delivered in the accounts you usually get a pause........

For the record this isn't just my experience. McKinsey's 2025 State of AI survey found that over 80% of organisations report no material enterprise-wide EBIT impact from generative AI. Gartner expects over 40% of agentic AI projects to be cancelled by the end of 2027 because of cost, unclear value or weak risk controls. Closer to home, research by Klarus this summer found that 73% of UK and Irish mid-market businesses have deployed AI, but many projects are stuck at pilot stage.

So what is going wrong? Currently there appear to be three key reasons;

1. No one owns the value. AI usually arrives through the Tech function or managed service provider. They can switch it on, but they can't make the finance team close the month faster or the customer service team handle more calls. Nobody is accountable for what the spend should deliver.

2. Hours saved are not cash saved. If Copilot saves each person an hour a week, that is only worth something if the hour is redeployed, a vacancy goes unfilled or a cost goes away. Otherwise, as I have said in previous posts, you are paying people to do less while still paying the licence bill.

3. The wrong people own the tools. As agents and automations creep into processes, there is a tendency to assume the Tech team should own them. But if an agent gives a customer the wrong answer or posts a wrong figure in a spreadsheet, that's an operational problem. The function head needs to own the outcome, with the Tech team owning the platform underneath it.

The businesses getting real returns are doing something different. McKinsey's research shows that redesigning workflows is one of the clearest drivers of AI-related EBIT gains, and that senior ownership of AI goes hand in hand with financial impact. It's the same answer as every other transformation: clear ownership, redesigned processes and benefits that someone tracks.

Put simply this is a change programme and it needs to be treated as such. Someone needs to define the problem, identify the benefits, create a plan for change and then drive through the new ways of working. This doesn't just 'happen'. Resources with the right skills need to be deployed.

Introducing the AI Value Diagnostic

At Imago we've applied our Diagnose:Define:Deliver methodology specifically to this tackle the problem.

• Diagnose (2–3 weeks). We baseline AI spend and usage, including the tools people are using that you may not know about. We map who owns what today, identify where time and cost could come out of two or three functions, and check whether your data and controls are ready.

• Define (2-4 weeks). We give you a simple operating model: business heads own outcomes, the Tech team own the platform, and risk owns the guardrails. We set out three to five priority process changes, each with a business case in cash terms. And we help you decide which licences are earning their keep.

• Deliver. If you want support to make it happen, we run 90-day value sprints, embed ownership in the functions and track benefits all the way to the P&L.

As an accountant, I'm only interested in benefits that show up in the numbers. That's what makes our approach different to training or technology-led AI programmes. And if the Diagnose stage shows there's no money to be saved, we'll tell you.

What next?

If you're an FD, CFO or MD who has spent money on AI and can't yet see the return, I'd like to talk. A 45-minute conversation is usually enough to tell whether a diagnostic would be worthwhile.

Email info@imago-consulting.co.uk to book a time and we can establish exactly where your 'AI efficiency' could be hiding.