To understand how to work with cohorts, you need to understand how it is organized. First, a cohort consists of objects that are collected by some attribute, usually they are clients or potential clients who came to the product during the same period of time. 

Then we observe the change of properties of these objects over time. In this case, we have the opportunity to look at the average value of the property (metrics) for the entire time of the object in the cohort, and in a particular period of time, in addition, we can see what value had a property at a given point in time.

All of this allows us to compare cohorts, but it is important to remember that cohorts can only be compared over the same period of time, so relative periods are used to represent cohorts. 

If I need to understand what metric values the business as a whole has at a given point in time, then I use absolute cohort period representation, which allows me to summarize metric values in cohorts for each period separately.