One report credits search with an enquiry. Another report includes the same enquiry after a social interaction. The customer mentions that a colleague recommended the business.
There does not have to be one false report in this story. Each view may be applying a different rule to the information it can observe. The difficulty begins when those views are treated as independent counts or complete explanations of why the person acted.
Write down what each view includes
Start with the outcome definition and the reporting period. Is the number a submitted form, an accepted enquiry or a later sales outcome? Which interactions can the report see, and how does it decide what receives credit?
Use the documentation and configuration relevant to the actual reporting setup. Do not assume that two columns share a definition because both are labelled conversions.
Keep observed outcomes distinct from allocated credit. A crediting rule can help compare activity within a report, but it is not the customer's full account of the decision.
Reconcile at the business outcome where possible
A central record of enquiries can establish the number the business actually received and how those enquiries progressed. Handle identifying information only where it is needed and authorised; a management summary generally needs grouped results.
Channel totals may still overlap or omit journeys. Explain that rather than adding them together and calling the sum unique customers.
The purpose of reconciliation is to establish a reliable shared outcome, not to force every report into a certainty it cannot provide.
Treat missing information as missing
An unobserved interaction is not proof that no interaction occurred. A direct visit can follow an earlier conversation, a remembered message or a saved link. Avoid filling those gaps with a confident narrative simply because the report needs a tidy explanation.
Customer feedback can add useful context, while carrying its own limitations. People may remember a broad impression rather than the sequence the analyst would like to reconstruct.
Let the decision determine the next investigation
If the team is deciding whether a message helps recognition, investigate that question. If it is reviewing enquiry quality, connect the relevant requests with their later outcomes. A broader channel allocation decision may need a carefully designed comparison and a clear statement of uncertainty.
Reading the gaps makes the reports more useful. It helps the team choose a question the evidence can answer, while keeping a partial view from becoming an invented complete journey.
A thought worth keeping?
