Relic Solution: Cohorts AKA Faceting by Time

Timeseries charts are a great way to visualize trends over time, but sometimes you want to see the buckets themselves in order to see what activity looked like for a given time range. Cohort analysis allows you to group query results based on different time-based criteria. These queries rely on FACETing - or ‘grouping’ results by something like the month accounts were created in, the date of last account access, or any other time-based criteria. Here’s a video that shows basics.

Maybe you want to know what the throughput was for your application, but want to know it hour-by-hour to see if there are particularly busy hours of the day. Do this by getting today’s data and FACET with the keyword hourOf

You’ll get something that looks like this:

Similarly, you can do this for day-over-day analysis or week-over-week analysis. For week-over-week analysis, change the amount of time in the query (SINCE last month) and then FACETing with the keyword weekOf

You’ll get something that looks like this:

Another thing you can use cohorts for is to exclude particular windows from your query. For example, maybe you want to exclude a window from 6am-7am when you are doing site maintenance:

You’ll get something that looks like this:

Or perhaps you just want the total number for the day with your maintenance window excluded:

You’ll get something that looks like this (the total number with the transactions from that hour excluded):

With cohort analysis, you can hop in the Tardis and have fun playing with time.

A tip of the Nerd-Relic cap to Henry Shapiro, Insights Product Manager, for sharing this tip.