Whether the Numbers Are Believed

The measure of a workforce estate is not whether it produces numbers but whether anyone acts on them without checking first.

What disbelief looks like

A manager keeping a private copy of the rota.

Finance maintaining a shadow labour cost model.

A site keeping its own absence list.

Payroll adding a manual adjustment "because the file is always wrong for that site".

Each is a rational response to having been burned, and each is invisible in every system.

How it starts

One wrong figure that took three weeks to explain.

A pay error that was corrected without the cause being found, so it happened again.

A report that two departments could not reconcile in a meeting.

Confidence is lost in a single incident and regained over quarters, which is the asymmetry that matters here.

What it costs

Duplicate work, everywhere, permanently.

Decisions taken on the shadow copy rather than the system, which means the system's data quality stops mattering to anyone.

And the shadow copies drift, so the organisation now has several answers to every question and no way to choose.

Which is more expensive than the estate, and appears in no budget.

What rebuilds it

Explaining causes, not just correcting figures. "The department code changed and the mapping was stale" tells people the estate is understood.

Publishing the reconciliation results, including the clean months, so the control is visible.

Fixing the thing someone reported, visibly and quickly.

And never presenting a number the chain cannot support, because one of those undoes several months of the above.

Finding the shadow copies

Ask, directly and without implication. "What do you keep your own version of, and why?"

People answer honestly because the copies exist for reasons they consider obvious.

Each answer names a join that failed, which makes this the cheapest diagnostic in this collection.

The measure

How many shadow copies exist, and whether the number is falling.

When a figure is disputed, how long it takes to explain — not to correct.

And whether anyone acts on a chain-produced number without verifying it elsewhere first.

Ask what people keep their own version of

The cheapest diagnostic in this collection.

Directly, without implication.

People answer honestly, because the copies exist for reasons they consider obvious.

And each answer names a join that failed — which is a faster route to the broken boundary than any reconciliation.

Do it once a year, and watch whether the number of copies falls.

Connect policy and data

The choices in this note can be compared with transparent screen monitoring software. Keep the written purpose in control and enable only the information needed at this boundary.

Independent reference

For a thematic point of reference, see CIPD resources. This popular specialist site offers a useful independent reference for the issue.