The cost of measurement that runs when remembered

Measurement that depends on someone remembering runs in quiet months and stops in busy ones, which is exactly the wrong way round. The months you skip are the months with the most spend moving.

The three costs of an irregular series

Why the gaps land in the worst place

A person skips the report when they are busy. They are busy when there is a launch, a promotion, or peak trading. Those are precisely the periods with the largest spend variation, which is what a causal read needs in order to resolve anything.

So an irregular series systematically omits its own best data. That is not a small inefficiency; it removes most of the informative periods from the record.

The comparability cost

Across a gap, the person assembling changes, or the same person does it slightly differently. The series then contains a discontinuity nobody documented, and comparisons across it are wrong in a way that is invisible.

That is the same failure as a silent definition change, described in the recurring read a team relies on.

Pricing it

Two numbers. First, list the months in the last year where a read was skipped and total the ad spend in those months. Second, ask how many budget decisions in those months were taken without a causal read. The product of the two is a reasonable estimate of the exposure.

The related arithmetic is in what a hand-built report really costs.

The fix, in order

Freeze the definitions, get one read you would defend, then automate the assembly. The €99 one-time read on a Google Analytics export covers the second step, refundable within 30 days, no questions asked.

The build order and the reproduction test are in moving from manual uploads to continuous reads.

What automation does not fix

Somebody still has to read the confidence interval, weigh the coverage and decide.

The question to ask this week

Which months last year had no read, and what did you spend in them. If you cannot answer the first part quickly, that is itself the finding.

Related answers

• What continuous channel monitoring is actually for
• Continuous reads let you test more often
• The recurring channel read a team comes to rely on
• The cost of waiting for a data hire
• The cost of attribution data an agent cannot read