To begin creating a geo-testing experiment:
Navigate to the Experiments page from the Nav bar and click on the “Create Experiment” button.
On the Set-up page, select “Geo Testing” under Experiment Type.
Select the date range for the experiment, i.e., the date range from which you want to start the experiment and when it should end. It should be a minimum of 15 days.
Upload the historical data in the form of a CSV file.
Map the schema by associating your input data with Lifesight's data types and input number of Geo's to experiment.
Name the experiment and click on ‘Next’ to proceed.
On the campaigns, define your Test Cell. Select the platform and type of experiment (hold-out or scale-up). Input ‘Expected ROAS’ from the campaign.
Click on ‘Add Cell’
Click on 'Find 'Markets' to proceed
Once the experiment is created, the status of the experiment will change to ‘Finding Markets’. The status will change to'Markets Ready’ once the test geos are identified. It usually takes 10–15 minutes for lifesight to identify the test markets for the experiments.
On the Markets page, select the test markets of each cell and click on Next.
Review the details of the experiment you want to start on the review page and click on ‘Save’ to begin.
The user then activates experiments (campaigns) for each cell on target platforms, with selected markets as holdouts or scale-ups for the suggested duration.
Data Table: Create a table for each data set, including at least three key variables: time, location, and the measured Key Performance Indicator (KPI). Enhance accuracy by adding covariates.
Pre-Campaign Data: Use historical data that is four to five times longer than the test duration. Minimum requirement: data from 25 pre-treatment periods across at least 20 geographical units.
Granularity:
Time: Daily granularity.
Geography: Use the most detailed geographical unit available (city, state, zip) for precise insights.