In conversations about AI with business leaders, I keep hearing different versions of the same questions. Where do we start, how much should we invest, and how will we know whether it's working? Some organisations have already spent a considerable amount and are still trying to get people to use what they've bought. Others are being asked to move faster when trading conditions are difficult, budgets are tight and their teams have more than enough to deal with already.
Even where there is strong support for AI at board level, not everyone agrees on what success looks like. Some directors want a clear financial return, while others focus on customers, risk, the organisation's long-term position, or what the changes will mean for employees. These aren't competing concerns that can be brushed aside. They all factor into whether an investment is worthwhile.
This is why I find some of the usual measures of AI progress frustrating. We hear how many licences have been bought, how many people have been trained and how much time AI might be saving. Those things tell us something about activity, but they don't tell us whether the business is actually better off, or whether we've spotted an opportunity to do something entirely different. I think we need to be clear about what we're trying to achieve before deciding how to measure it.
Start with what you want to achieve
When I talk about AI business value, I mean what actually changes for the organisation as a result of the investment. It might be lower costs, better customer outcomes, stronger growth or less risk, but it could just as easily be a product or service that wasn't possible before. Sometimes the starting point is a problem we understand well; sometimes it's an opportunity we haven't worked out how to pursue yet. Both are valid, and I don't think we should assume every investment needs a problem to solve.
There's a practical reason we tend to start with problems, though. Most leaders are flat out running the business, and it's easier to get attention and budget for a cost, a backlog or a risk that everyone already recognises than to find space to imagine something quite different. I wouldn't want that to limit the ambition. The more useful question for a board is what are we trying to change or make possible, where could AI help, and what evidence would tell us it was worthwhile? If the investment serves several purposes, it's still worth agreeing on the main one so that you know what success means.
1. Reducing costs
If the business is investing to protect margin or reduce expenditure, the financial measure needs to be a cost that actually changes. A reduction in supplier spending, contractor costs, or overtime may be visible in the accounts. Avoiding future cost increases can also be valuable, but it should be described differently from a saving made today.
One of the biggest traps is taking a calculation of hours saved and treating it as money saved. A team might complete its work more quickly while the salary bill stays exactly the same. That could be an excellent result if the time is used to improve service or take on more demand, but it isn't a cash saving. It is also important to include the cost of getting AI working properly, from integration and training through to ongoing support, checking and correcting its output.
2. Growing without increasing costs at the same rate
For a growing business, the question may be whether it can support more customers or more transactions without adding people and infrastructure at the same rate. That is a different reason to invest, and it calls for different measures. You would want to look at the volume of business being handled, the resources required to deliver it and whether service levels and quality are being maintained.
This is where the discussion about capacity becomes important. In one client conversation, I talked about the choice leaders have when AI frees up time. They can deliberately use that capacity for customers, sales or improving the business, or they can leave things largely as they are and watch the time get absorbed into other activities. The technology doesn't make that decision for them, and the return depends on what they choose to do next.
3. Improving quality and managing risk
Some investments have the strongest case not as cost reduction. A business may want fewer mistakes, faster customer responses, more reliable decisions, or better control of a process that has major risk. Those improvements can be commercially important, even when they are harder to express as an immediate saving.
Boards are also having to weigh new risks against the benefits. Privacy, security, accuracy, regulatory obligations and trust don't disappear because a technology delivers productivity gains. When some board members focus on commercial results, and others worry about employees or customers, both sets of questions deserve a proper answer. It makes sense to measure errors, complaints, service outcomes and control failures, while also checking whether the AI itself has introduced problems that were not there before.
4. Changing how the business operates
Some organisations have a much bigger opportunity than helping people complete existing activities faster. Processes may have been designed around old systems, departmental boundaries and the need to pass information from one person to another. AI may create the opportunity to rethink the whole process, including parts that could be removed rather than simply accelerated.
A customer doesn't particularly care that one team has saved fifteen minutes if the request still takes a week to resolve. What matters is whether the experience becomes quicker, simpler or more reliable, and whether it costs the business less to deliver. We have written before about how AI can shift bottlenecks rather than remove them. That is why I would measure the end-to-end outcome, rather than adding up isolated improvements and assuming they amount to transformation.
5. Creating new products, services or revenue
I wish more of the conversation started here, because some of the opportunities have little to do with making existing work faster. AI may make a personalised service affordable, enable a business to reach customers it couldn't previously serve or make an entirely new product possible. That's a very different proposition from reducing a cost, and it needs a different measure: whether the organisation is creating something customers value and, where relevant, will pay for.
You might start by looking at customer take-up, repeat use and feedback, but at some point those measures need to connect to revenue, retention or margin. You also have to consider the full cost of delivering the new offer and whether it brings in new business or shifts customers away from an existing product. You can build something people find impressive without creating a commercially attractive business around it.
6. Remaining competitive
Some of the hardest board decisions are about what could happen if the organisation does nothing. A competitor might find a much cheaper way to serve the same market, customer expectations might change, or a business model that has worked for years may become less attractive. There may be very good reasons to invest before the financial return is obvious, but that doesn't mean every strategic concern should automatically become an AI programme.
I would want the assumptions to be clear. What exactly do we think is changing, what are we seeing from customers and competitors, and what will we need to do differently? Then you can set sensible points to review the evidence and decide whether to increase investment, change direction or stop. There may never be a precise financial figure for the market share you managed not to lose, but boards can still make a considered decision about the risk of standing still.
How should boards measure AI business value?
To measure AI business value, a board needs to agree on the outcome it expects, understand how the business performs today, include the full cost of change and review what actually improves. In the conversations I'm having, leaders also want to know where the opportunities are, which ones matter and whether the organisation has the capacity to act on them. They should be able to say an investment isn't working, rather than feeling they have to defend it because it was once labelled a strategic priority.
A useful report should make it clear what the organisation is trying to achieve, whether that's solving a problem or pursuing an opportunity. It should show where the business is starting from, what has been invested, what's changed so far, the full costs and the outcomes still expected. Someone also needs to be responsible for turning the opportunity into a result. If the business is claiming a saving, where has expenditure changed? If it is claiming growth, where is the additional output or revenue? If the claim is about lower risk or a stronger competitive position, what evidence supports that judgement and what remains uncertain?
I would also keep potential separate from results. An assessment might show that AI could reduce the time required for a group of activities, but that is only the beginning of the business case. A further step lies between showing that something can be done differently and proving that the business has benefited. This is also where attribution matters: sales may have grown for reasons unrelated to AI, and the same released capacity cannot be counted once as a saving and again as additional revenue.
None of this requires pretending that every benefit can be reduced to an exact dollar figure. It requires being clear about what is known, what is estimated and what decisions are still needed. That gives boards a much better basis for allocating capital than a long list of pilots and optimistic productivity calculations.
A practical place to start
We built GoFIGR to make that first assessment much easier than people might expect. There's no need to commission a time-and-motion study or ask hundreds of employees to spend weeks documenting what they do. We start with information organisations already have, such as role descriptions and other existing documentation, and use it to assess where AI could change tasks, skills and roles. Most clients have their initial results back the day they upload their data. That gives leaders something concrete to work with quickly, including where costs, capacity and quality might change, and which capabilities they'll need to develop.
If you're deciding where AI is worth investing in your organisation, you don't have to start with industry averages or another theoretical business case. Book a 30-minute executive briefing and we'll analyse a role you choose, show you what the data reveals and discuss which opportunities might be worth exploring before you commit to a wider assessment.





