Simulated worked examples
See how measurement
can change the story.
Follow the tempting conclusion, find the weak comparison, and work through a better test.
These are learning scenarios, not accounts of real campaigns. Each uses simulated data to make a measurement problem reproducible. The assumptions and limits are explained in the example.
Did search advertising create new demand?
Explore how existing purchase intent changes the apparent value of a paid search campaign.
Work through the example 02 / AttributionWould those shoppers have returned anyway?
Separate the effect of retargeting from the behavior of people already close to buying.
Work through the example 03 / ExperimentsDid the campaign lift sales—or did the season?
Check how market selection and pre-period demand change the interpretation of a geo test.
Work through the example 04 / SurveysDid opinions change, or did the sample?
Follow how differences in survey recruitment can look like brand persuasion.
Work through the example 05 / Audience selectionDoes claiming an offer mean it caused a purchase?
Inspect why offer claimers and non-claimers may differ before a promotion starts.
Work through the example 06 / Audience selectionWere the people targeted already likely to spend?
Examine how transaction-history targeting can overstate a promotion's causal impact.
Work through the example 07 / Audience selectionWas it the reward, or existing brand affinity?
Work through the difference between selected participants and a credible comparison.
Work through the example 08 / Outcome qualityDo more form fills mean more new business?
Trace the distance between early lead volume and the outcome the campaign was meant to change.
Work through the example