The Automation Business Case Starts With the Work, Not the Technology

Most leadership teams I know are being asked the same two questions: what is our AI and automation strategy, and what will it actually save?

The second question is where many business cases come unstuck, because they are built left to right. They start with a technology – a platform, a licence, a vendor demo – and work forward to a justification. The savings are optimistic, the board is disappointed, and scepticism about the whole agenda grows.

Last week Peter Brown, PwC’s Global Workforce Leader, shared a stage at UNLEASH Paris with Lloyds Banking Group and described their approach to redesigning work at scale: “starting with the customer and working backwards, ‘right to left’. Not putting AI onto today’s processes, but rethinking the work, working out the right role for people and technology and redesigning around it.”

It’s the best description I’ve heard of how an automation business case should be built too.

Building it right to left

1. Start with the outcome. What should be better – for the customer, the employee or the business? Faster, more accurate, easier, cheaper to run? If you can’t say, no amount of technology will tell you.

2. Understand the work that delivers it today. Volumes, time, hand-offs, and the exceptions people quietly handle. I’ve argued before that we may need to bring back the stopwatch; this is where it earns its keep.

3. Find the addressable workload. Which activities are rules-based, high-volume and stable enough to automate – and which depend on judgement, exceptions and relationships? Be honest: the addressable share is usually smaller than the demo suggests.

4. Decide where the released capacity goes, and who owns it. Hours saved don’t release themselves. A saving only becomes real when a named role or person in the operating line has signed up to it – whether that time is removed, reinvested or redeployed. (I’ll come back to this soon.)

5. Only then choose the technology – and price all of it. Licences, implementation, integration, running costs, and the people needed to support it. AI isn’t free, as I explored in The 30% Rehire Problem.

Read left to right, a business case asks: what can this technology save? Read right to left, it asks: what needs to be better, and what is the least we need to change to get there? Sometimes the honest answer is a simpler process rather than a new platform.

In the programmes I’ve run or been part of, we’ve carried out capacity analysis – something I’ve written about before. It’s challenging work. Leaders have to buy into it, or it simply doesn’t happen, because it takes time, energy and explanation. At first they’re usually very interested: they engage with the analysis and the ‘what if’ scenarios. But for leaders who don’t yet see the long-term change it supports, the interest then deflates like a party balloon, with little chance of re-inflating it. The analysis stays useful as an indicator for a while, but after the next organisational change it quickly loses its validity – and the effort, energy and value go with it.

Is that because those leaders aren’t commercial, are avoiding the inevitable, or are simply too busy? Whichever it is, the chance to identify the capacity that could be released is lost – and so is the explanation to staff. And as we know, nature abhors a vacuum.

And there’s a hidden cost. When the analysis isn’t used, it usually has to be done again – and the second time is harder. People remember giving their time and honest answers to an exercise that went nowhere, or they suspect it was really about headcount all along. The input is more guarded, and the results are less reliable.

The Budget on 28 October made the same kind of case at national scale. The Chancellor announced a £50 million fund for AI transformation across the civil service, saying it would “support saving a further £600m from back office costs”. I hope that case is being read right to left. The £50 million is the easy number to find. The £600 million only becomes real when someone has worked out what the back office actually does – and a named person has signed up to what happens to that work.

The opportunity from automation and AI is real. But it is realised by redesigning the work around the technology – never by the technology alone.

So before the next automation business case reaches your board: can you read it right to left?

Related insights

Maybe We Need to Bring Back the Stopwatch

The 30% Rehire Problem

Are We Finally Getting Better at HR Technology ROI?

Source for the quote: Peter Brown MBE, LinkedIn post, 23 October 2026

Source for the Budget figures: Civil Service World, 28 October 2026