Time Saved Is Not a Result

PwC recently published a piece on the uncomfortable truths of AI adoption. I agree with most of it. One line I want to push back on: "If it reduces a five-hour task to one hour, that creates value."

Not necessarily. It frees up four hours of capacity. Whether that capacity becomes value depends on a decision most organizations never make.

Some time ago, a CHRO invited me to a session with the HR leadership team to talk through AI opportunities. Most ideas were about employee efficiency: less manual administration, quicker routine work. All sensible, easy to pilot and easy to count.

I put a rough example on the table to make one point: where AI saves time and where it creates the most value are often not the same place. That matters because every organization has limits. Money, people and attention spent in one place cannot be spent in another. If every team uses AI to save a little time, everyone takes a small step forward, and the organization misses the leap it could have made.

Take an HR department of 40 people. Half their week is administration, and AI cuts that work by a quarter. That frees up the equivalent of five people: perhaps €400,000 a year, assuming €80,000 in annual employment costs per FTE. If you can actually take the cost out.

Now consider the core business of grocery retail, where I spent most of my career. Take half a percent off €1 billion of purchasing spend. Per product, that is a fraction of a cent. In total, it is €5 million. Year after year. And that is a deliberately conservative example. The largest savings I have seen came from exactly this kind of work: not from faster office work, but from cross-brand data work that made joint sourcing possible. That was before AI even entered the picture.

On a dashboard, though, the first reads 25% and the second 0.5%. Those percentages have different denominators, but they can steer attention all the same. And in this case the larger percentage may be the smaller result.

Time saved measures output. The business measures outcomes. The sourcing case is harder and slower to build and deploy, I know. But when money, people and attention are limited, that is exactly where they should go.

The HR leadership team saw the point. Still, every leader would prefer an improvement in their own scope over a bigger gain elsewhere. That is not a criticism. It is how mandates work: even with its CHRO in the room, that HR team could not decide to invest in the sourcing case. That question had to be answered above them, by the people who own the P&L.

Within its own scope, that team did something many leadership teams skip. They had the conversation together, instead of each team chasing its own gains, and they asked where time saved would matter most. What they did not settle was where that time would go. Like most organizations, large and small, they assumed hours saved would automatically turn into impact.

Sometimes that happens. But that is an assumption, and assumptions don't deliver results. In 2025 the UK government published its findings from a Copilot experiment: participants reported saving 26 minutes a day, but it was not possible to identify how that freed-up time was spent. A separate evaluation in one department found no robust evidence that time savings had improved productivity. Some people used the time for training or a lunchtime walk. If that was the intent, good. If it just happened, it is not value the organization can point to.

Elsewhere, one person's saved time becomes another person's work. Stanford and BetterUp researchers found that AI-generated "workslop", polished-looking work without substance, costs the person receiving it almost two hours each time.

If you want to turn freed capacity into measurable business value, it needs an intentional destination. There are three main routes, each with its own measure.

Out as cost: fewer contractors, less overtime, roles you don't refill. Measure the spend that went down.

Into throughput: more of the same work, or the same work faster. Measure the backlog or the cycle time at the same cost.

Into different work: name that work before you start, track that the hours actually go there, and measure what it delivers.

If nobody chooses, the time gets absorbed anyway. Saved hours don’t sit in a bank account, waiting to be spent.

PwC is right that demanding perfect ROI up front kills good projects. And not every gain has to be financial: time that makes work better is valuable in its own right. But if you claim value, you should be able to show it, not assume it. That takes one question, asked before the pilot and again whenever leadership is shown an AI dashboard or showcase: where do the hours go, and on which line of the P&L can we see them? Ask it consistently and the portfolio reorders itself. Efficiency cases with no destination drop down the list. The unglamorous improvements, harder to build but clearly visible in a budget or the P&L, move up.

Most AI business cases promise hours saved or use cases live. Nothing wrong with that, but it is output. Most AI business cases promise hours saved or use cases live. Nothing wrong with that, but it is output. The good ones point to where the value will land in the P&L, and when. Only those can carry a strategy.

Recognize this? I help leadership teams turn data and AI investments into results the P&L can show. Happy to think it through with you in a 30-minute call. No pitch, no deck. Book a call →