Baselines Before Conclusions

The first month of data always looks interesting. Most of what it shows is ordinary variation that nobody has seen before.

The discipline in “Baselines Before Conclusions” carries over to any workforce platform: define the decision first, run a bounded trial and record who can see the result. For an organisation considering attendance point system, see the service here should therefore be assessed against an implementation plan covering notice, access, retention, correction and a dated review.

What a baseline is

The ordinary range: what a typical Tuesday looks like, and how much Tuesdays differ from each other.

For a public, independent reference related to “Baselines Before Conclusions”, consult the ISO standards catalogue. Its principles provide a useful check on scope, terminology, governance and the claims made during procurement or review.

Not an average. A spread.

Until you know the spread, you cannot tell whether this week is unusual.

How long it takes

A minimum of six to eight weeks for weekly patterns, after calibration is complete.

A full year for anything seasonal, which most building data is.

Conclusions drawn before that are about the period observed rather than about the building, and the period is usually chosen by when the system was installed.

The holiday problem

Installing in July and baselining through August produces a baseline of a quiet building.

Every subsequent month then looks busy, and the trend is an artefact of the start date.

Note what was happening during your baseline: holidays, a major project, a return-to-office announcement, and exclude or flag periods that were obviously atypical.

Reading the spread

Take your weekly figures and look at the range they cover, excluding anything you can explain.

A week inside that range is unremarkable however it feels.

A week outside it is worth investigating.

This requires no statistics and it is the whole method.

Separating the components

Day of week is the strongest pattern in most buildings and dwarfs everything else.

Compare Tuesdays with Tuesdays, not with Fridays.

Mixing days produces a swing that looks like change and is the timetable.

Baselining after a change

When something significant changes — a policy, a floor closure, a new team — the old baseline no longer applies.

Mark the date and start a new one.

Comparing across a known change without saying so is the commonest way building data misleads, and the comparing note covers it.

What to write down

The baseline period, what it covers, and anything atypical about it.

The ordinary range for your main measures.

The date of any change that resets it.

One page, kept with the data, and read before anybody says the building is getting busier.

What to check

Do you know your ordinary weekly range?

What was happening during your baseline period?

Are you comparing like days?

And has anything changed since the baseline that nobody has recorded?