Insights · The business case
Hours saved is a weak business case.
Hours saved is the most common AI business case and the weakest. It is easy to model, hard to find afterwards, and quietly threatening to the very people whose adoption decides whether the project lives. Customer outcomes make a stronger case, and they usually contain the efficiency anyway.
Why does everyone reach for hours saved?
Because it fits in a spreadsheet. Multiply a task time by a headcount by a percentage and the model produces a satisfying number. Finance can audit the arithmetic, the slide looks rigorous, and the case sails through the first meeting.
Then it meets reality. The saved hours are spread thinly across hundreds of people, in fragments too small to redeploy. Nobody can point to where they went. A year later the same case is back, promising the same hours, and everyone in the room quietly knows it.
What is wrong with it beyond the arithmetic?
It frames the technology as a way to do the same work with fewer people. That is a message every member of staff can decode, and it is why hours-saved projects so often meet polite, immovable resistance. You are asking people to adopt the thing that measures its success by their absence.
And it is fragile in the budget round, because a cost saving that never materialised is the easiest line to cut. Cases built only on efficiency die in their second year.
What does a stronger case look like?
It starts with what the customer feels. A decision that took weeks now takes a day. A question answered at first contact instead of the fourth. A communication that is relevant instead of generic. These outcomes are visible, measurable, and connected to revenue and reputation rather than only to cost.
They also carry the staff with them, because serving people better is a purpose employees will adopt. In two decades of building personalisation systems for household-name brands, the programmes that endured were the ones framed around the customer. When customers saw what was genuinely most relevant to them, they responded, and the efficiency came anyway, as a by-product nobody had to fear.
How do you rebuild a weak case?
Take the hours-saved line and ask what those hours were failing to do for a customer. Slow answers? Repeated questions? Generic service? Name that, measure that, lead with that. Keep the efficiency in the case, but as supporting evidence rather than the headline. Then give the case one sentence a sponsor can repeat to a sceptical colleague, because in the end that sentence is what gets funded.
Common questions
Is efficiency a bad reason to adopt AI?
No, an incomplete one. Put the customer outcome first and let the efficiency ride inside it.
How do you measure customer outcomes?
Pick the delay, error or friction the customer actually feels and measure that. These are harder numbers than modelled hours, not softer ones. The general form of this is that the value is downstream: measure at the hand-off, where the outcome was already valued.
What makes a case survive a budget round?
An outcome a senior person can repeat in one sentence, and at least one number finance did not take on faith.