Required Data Points
The experiment requires only three essential data points:
Date
Total output variable (e.g., Total Revenue, Total Orders)
Geo-Granularity
<aside> 💡
Important: This should be total figures, NOT channel-attributed numbers
</aside>
Example: Use total daily revenue for each geography, not “revenue attributed to Meta ads”
Historical Data Requirements
The data should cover atleast one year, and should be as up to date as possible when designing the experiment. Avoid designing experiments today that you plan on running in 2 months time.
Geographic Granularity
Best practices:
Keep total locations under 240 (DMA granularity or less).
Avoid zip code level granularity in large markets since revenue and conversion volumes can be very low at this granularity.