In almost every report two fundamentally different kinds of number sit side by side without being labelled. One is observed: a click happened, an order came in. The other is computed: 40 per cent of that order goes to the first touchpoint.
Both are useful. But they bear different weight. About a measured figure you can argue whether it is complete. About an attributed one you can argue whether the rule makes sense. That is a very different discussion.
The third layer that does not exist here
Many tools know a third kind: the modelled figure. Where consent is missing or a signal fails, the gap is filled by calculation. The result looks like a measurement but is a model on top of a model.
Metrunio does not have that layer, and this is not a technical limitation but a decision. Where a signal is missing, the gap stands. A conversion with no measured touchpoint is called “unattributed” and is not spread across channels on the grounds that it must have come from somewhere.
Alongside these, Metrunio shows figures that are neither measured nor attributed, and they carry their origin too. Reported is what an ad platform says about itself. Set is a value you fix yourself, such as what a sign-up is worth to you. Self-reported is what a customer says, for example in the post-purchase question. None of them fills a gap, and set and self-reported values never appear in revenue.
What that changes day to day
- The total of attributed conversions is at most as large as the measured one, usually smaller. That is correct and not an error.
- The “unattributed” share is a metric in its own right: it says how much of your measurement is missing.
- A figure you pass on needs the model stated with it. Without that it is not comparable, not even with itself from last month.
Key point
Filling a gap so the table looks complete improves the table and worsens the decision.