Finding the AI opportunities worth chasing
Most businesses spend their AI budget where it is most visible, not where it pays. MIT found the biggest returns hiding in the back office. Here is how to find the opportunities actually worth chasing.
When a business goes hunting for AI opportunities, it tends to chase the exciting ones: the customer-facing, the visible, the projects you can show off in a board meeting. That instinct is an expensive one.
In its 2025 study, MIT found that more than half of corporate generative-AI budgets went into sales and marketing tools, while the biggest and most dependable returns were turning up somewhere far less glamorous, the back office (MIT, The GenAI Divide: State of AI in Business 2025). The repetitive operational work, the process costs, the outside agency spend that quietly adds up. The money was chasing visibility, not value.
The loud opportunities and the quiet ones
The loud opportunities are the ones everyone reaches for first, because they are easy to imagine and satisfying to announce. The quiet ones are the unglamorous, repetitive tasks buried inside your operations, the work nobody enjoys and everybody does.
Picture the jobs that eat hours without anyone questioning them: drafting the first version of routine documents, summarising long reports nobody has time to read, pulling the same numbers together every month, answering the same supplier question for the hundredth time. Individually, none of it is dramatic. Together, it is often where the largest and safest savings in the whole business are sitting.
This is not a rule that the back office always wins. It is a warning that the obvious answer is often the wrong one, and that the opportunities worth chasing are usually the ones you have stopped noticing, because they have always just been part of the job.
How to find the ones worth chasing
A real opportunity earns its place on two axes: how much value it creates, and how feasible it actually is. For each candidate, ask:
- Does it move a real number? Time, cost, revenue or quality, in a way you could put in front of a board.
- Is the data there? A brilliant use case on top of poor data goes nowhere (why your data strategy makes or breaks AI).
- Can a human still own the outcome? The best early wins keep a person firmly in charge.
- Will people actually use it? A theoretical saving that disrupts how people work rarely survives contact with a Monday.
Rank what is left without sentiment. A handful of unglamorous, high-value, feasible opportunities beats a flashy one that scores well on excitement and nothing else.
Start where it's dull and profitable
When we run an opportunity assessment with a business, this is the whole exercise: separating where AI genuinely moves your numbers from where it merely looks impressive, and ranking accordingly. We would rather find you three unglamorous wins that pay for the entire programme than one show-stopping pilot that returns nothing but a good story.
If you want a sense of where your own best opportunities sit, the free AI Maturity Assessment is a quick read on where you stand and what to do first.
Want to know where your team actually stands?