Requirements for Testing
Before launching a test, it’s important to ensure you have the right data and media setup in place. In this guide, we walk through the data and operational requirements for both GeoTests and Difference-in-Differences tests within Rockerbox.
🌍 Requirements for geo-tests
To run a geographic incrementality test, you’ll need:
- KPI Data with Geographic Identifiers
- One year of data for your primary KPI, by date and location
- Acceptable geographic levels: country, state, media market, zip code
- Additonal detail for all KPI types being analyzed in the test, if needed
- Data Availability in Rockerbox
- If using the Rockerbox Data Foundation, this may already exist in your conversion tables
- If not, you may need to share store-level or retail partner data (e.g., Amazon or in-store purchases), including store name, address, and date
- Media Buying Capabilities by Geography
- You must be able to target media at the same geo level as your test (e.g. State or metro)
- Digital media is typically easier to geo-target; affiliate, print, and influencer may pose challenges
🔁 Alternative: Time Based or Fully Bespoke Tests
If you can’t control media geographically, Rockerbox can run a time-based test using conversion data and univariate modeling, or use attribution or other data to create counterfactuals for channel performance.
Examples include:
- Univariate tests for media shut off
- Brand vs. non-brand search
- Coupon strategy in one product category vs. another
In these cases, Rockerbox uses:
- Attributed conversions from your MTA tables
- Product-level purchase data from your Conversion tables