Analysis Across the Chain

A reconciling chain makes questions answerable that were not. Some of them are worth answering.

What becomes possible

Planned against worked against paid, per site and period, which is the core variance.

Whether overtime comes from absence, from understaffing or from rota design, which requires three modules and is the finding most operations want.

The real cost of absence, including the cover arranged and what it was paid.

Whether a low labour percentage reflects good scheduling or unfilled shifts, which are opposite situations that look identical in one number.

Seasonal patterns across years, which need consistent reference data more than they need sophisticated modelling.

Why site level and not person level

Each module's data carries its own error: schedule variance, clock accuracy, absence classification, rounding.

At site and period level those errors average.

At person level they multiply, and the result is a confident number about an individual that none of the sources supports.

Which is the note on requests to refuse, and it is worth restating here because a reconciling chain makes the wrong analysis technically easy for the first time.

What to build first

The variance report, because it is the one every department asks for in a different form.

With a stated period definition, since a payroll month and a scheduling week do not align and the comparison has to pick one.

And with the reconciliation status shown alongside, so a reader knows what weight to apply.

Data quality on the page

Which joins reconciled this period and which did not.

Known gaps.

The period definition used.

A report that states its limits is trusted; one that does not is tested once and then discounted entirely.

What not to build

A dashboard before the joins reconcile, which produces confident numbers from inconsistent data.

Individual comparisons.

Predictive models on top of data with an unreconciled identity mapping, which is building on sand with more arithmetic.

And anything only one department can reproduce, which becomes the subject of meetings rather than an input to them.

The measure of whether analysis works

When two departments cite different numbers, can the difference be explained in one conversation?

If yes, the chain reconciles and the analysis is real.

Show the reconciliation status

On the same page as any cross-module analysis.

Which joins reconciled this period and which did not.

Known gaps. The period definition used.

A report that states its limits is trusted; one that does not is tested once and discounted entirely.

And in this subject the limits change month to month, which makes a static caveat useless.

Follow the difficult record

Use employee activity monitoring to frame one representative case. The useful evidence is the record created when a value is corrected, approved and exported.

Independent reference

For a thematic point of reference, see the Office for National Statistics. Consult the source directly because technical, operational and legal details change.