Cohort analysis is technically any analysis performed in regards to a certain group. One very useful way to use cohort analysis is to look at two measurements of time.
To perform cohort analysis, it is necessary to first have two date fields in the dataset (for example Customer Created Date and Charge Created Date). Simply drag on those two fields as well the field you're interested in measuring (for our example, we will use Amount).
In your chart, the x-axis will show dates as new customers are created while the vertical access (in a stacked column chart) shows you the cohorts as they progress over time.
In the chart below, you'll notice that the x-axis is broken down by month because in this example, customers are billed monthly. This metric should be aligned with the timing of your data to make the most sense.
Below is a chart of Stripe payments broken down by when the charge was created, when the customer was created, and the amount. This shows you how valuable your customers are by when they signed up for your service.
For more information on how to perform cohort analysis, check out this blog post.