In this lesson I’d like to point out another nuance when calculating unit economics metrics per cohort (campaign, channel etc). We already saw that the wrong type of join could remove some cohorts from the result completely (if campaign has no signups yet it’ll be filter out with
This is why we always should use
LEFT JOIN in such calculations. To ensure that our report contains data on all campaigns we’re running, even those with 0 signups or customers. Let’s consider the query for calculating CAC per campaign:
WITH spends AS (...