ORIGINAL TECHNICAL PRACTICE
Diagnose an aggregate metric change
Overall conversion fell, but conversion improved in every country. How can both statements be true?
Written for independent practice. The model answer presents one approach; clarify constraints and explain alternatives.
A REASONED APPROACH
Model answer.
The composition of traffic may have shifted toward countries with lower baseline conversion. An aggregate rate combines within-group rates with group weights, so a change in the weights can outweigh improvements inside each group.
Work through it
- Write conversion as total conversions divided by total eligible visits, and calculate country-level numerators, denominators and traffic shares for both periods.
- Compare the observed aggregate with a standardized calculation using fixed country weights. Separate within-country changes from mix effects.
- Check that eligibility, attribution windows and tracking definitions did not change. Present both the aggregate business result and the segmented explanation.
Tradeoffs to discuss
- Standardization supports fair comparisons but uses a chosen reference mix. The observed aggregate remains relevant when actual traffic composition is part of the business outcome.
Common pitfalls
- A simple average of country conversion rates generally differs from the overall rate. Segmentation alone does not rule out tracking changes or smaller confounding effects.