Where freed time actually went

Before day 0, your aim named where freed time would go: the bottleneck, lower cost, better quality or growth. Ninety days later, people have saved time. This lesson checks whether that time went where you said it would, and how much of it you can actually count.

Saved isn't the same as converted

Almost every AI rollout can quote a number of hours saved. Far fewer can say what those hours turned into.

The hours are real. They just don't arrive anywhere the business can see. They go into more email, more drafts, more checking of AI output, and more work handed back. Without a record of where they went, nobody can tell the difference between time that changed something and time that quietly disappeared.

The conversion ledger

The answer is a conversion ledger. It's a simple table: one row for each chunk of freed time, filled in every two weeks. Each row says which team freed the time, how many hours, what changed to free them, where they went, and the evidence that they got there.

The last column does all the work. A row only counts when its evidence column holds something checkable. Hours saved with no evidence stay in the ledger as capacity: real time, recorded honestly, but not counted as value.

There's a second test, and it comes straight from the fourth principle: measure outcomes, never usage. Freed time only counts once it has gone where you promised. If your aim named the bottleneck and growth, hours that went to cutting costs are welcome, but they don't count until the aim is restated to include cost.

What counts as evidence

Checkable means someone outside the team could look it up. A count in the system of record, a contract that ended, an error rate, a signed order. A team that feels less stretched is real, and it matters, but it isn't evidence. Each destination has its own kind.

• The bottleneck: more output at the bottleneck, in the aim's unit. Quotes out up from 22 to 31 a week, from the CRM.
• Lower cost: a cost that stopped. Overtime that ended, a contractor not renewed, a licence cancelled, with the date it stopped.
• Better quality: fewer errors. Rework, corrections or complaints, measured the same way before and after.
• Growth: new work won. Bids submitted, customers served, orders signed.

Write the evidence the way you'd want to defend it to your finance lead: what changed, by how much, and where to check it. If you can't, the row is capacity for now. That's not a failure of the row. It's the ledger working.

Fill in your ledger

Start with what the teams around the bottleneck freed up since day 15. Add a row for each chunk of time, then be strict about the evidence column. The totals and the chart update as you go.

When most hours don't convert

If most freed hours never convert, that counts against the framework. The rule on where freed time goes didn't hold, and the readout says so in its first line.

It usually has an ordinary cause. Managers weren't told what the destination meant for their team. The time arrived somewhere the aim had nothing for it to do. Or nobody looked at the ledger until day 90. Logging it every two weeks, as the pilot kit does, means a row without evidence gets noticed in week 4, while there's still time to point those hours somewhere.

Whether a finance lead will trust a conversion ledger is one of the questions the framework's first pilots exist to answer. The ledger earns that trust the same way the prediction does: by counting less than it could, and by showing exactly where to check what it does count.