A time study finishes and the numbers look solid. Someone has stood on the floor, watched the work and measured it. Then the data reaches the forecasting system, and the structure it arrives in does not match the structure the system expects.

This is a gap I look for between a good study and a working forecast, because it can be easy to miss. Both sides are describing the same store. Neither is wrong.

Two true descriptions of one store

The study is organised the way the site is managed: departments X, Y and Z. That is how the rosters read, how the managers think and how the observer was shown around.

The forecasting model groups work by what drives it: service, replenishment and online fulfilment. In this fictional example, service includes transactions, assistance requests and returns; replenishment follows cartons; online fulfilment follows collection orders. Each activity keeps its own demand unit and standard.

Department Z is not a function. Replenishment is not a department. The names do not line up, and nothing in either system will tell you that.

A matching total is the weakest possible check

Here is the part worth slowing down for. The study measures fifteen hours. The system receives fifteen hours. The reconciliation passes.

Those fifteen hours can still be sitting in the wrong places.

Assign a department’s total to whichever function seems closest and the store total stays honest while the distribution quietly drifts. The forecast then sends labour to the right building at the wrong hour, or asks replenishment to absorb work that customers actually generate. The total is a necessary check, but it cannot validate the allocation by itself.

A correct total proves the arithmetic. It says nothing about whether the work is in the right place.

Map the activity, not the department

The fix is to drop one level down. Instead of assigning a department total, take each measured activity and answer three questions:

  1. What is the activity? Not the department it sat in. The work itself.
  2. What drives it? Transactions, delivered units, hours open, something else.
  3. Which function should carry it? The group whose demand signal actually moves that work.

This takes longer than assigning totals, and it is the step that tends to get compressed when a study is already late. That compression is usually what gets paid for later, either as a forecast nobody trusts or as a bill to rebuild the model.

See it happen

Make the study fit the system lets you assign department totals first, then switch to mapping activities, and compare the two. Watch the store total stay the same while the hours move underneath it. The experiment also estimates the rework if the mismatch is found after go-live rather than before.

A question for your next study: when your measurement finishes, who checks that the work landed in the right place — and what do they check besides the total?