When I talk to college students about what they’re studying, I frequently give them this piece of advice: pay attention in statistics class. I loved my liberal arts education. It prepared me for a lifetime of learning on the job. I can learn quickly and adapt to new situations and that has served me well throughout my career. But if there is one class from my undergraduate days that I wish I had really really dug into, it was my statistics class.
Once I left the world of low volume enterprise software and entered the modern world of transactional SaaS software, I suddenly found myself in a world where customer data was far more plentiful and we could compare different groups of leads, opportunities and customers to study their behaviors. And this is when my statistics class regret really started to rear its head.
One thing I quickly realized is that I had been ignoring a key concept for doing reliable, trustworthy analysis. And that is the concept of cohorts. If there was one reasonably complex aspect of doing analytics that I wish every person understood and internalized, it is cohorts.
A cohort, when doing analysis, is a group of similarly defined subjects. Most commonly, when doing analysis of sales results, you might compare opportunities that were created in a common time frame. In this case, your cohort is defined by the creation date of the opportunity.
Tip 1: Each type of analysis you execute might require defining your cohorts differently.
Though opportunity creation date might seem like the obvious choice when analyzing opportunities, In some cases, you might choose to cohort sales data based on when the opportunities have closed. One reason for choosing cohorts based on closed date is to ensure that you are analyzing opportunities that have progressed through the entirety of their sales cycle.
In other scenarios, you might also choose to cohort based on close date because you’ll want to assess behaviors after the opportunities have become customers, for customer satisfaction analysis or analysis of product usage.
It is important to be mindful of your cohort definition to ensure it properly fits the question you are trying to answer.
Tip 2: Cohorts become powerful once you use them to compare different time-based cohorts, e.g. the outcomes of opportunities created in January vs. those in February vs. those in March.
You are changing your business, working to improve it. And independent of your actions, the environment within which you are running your business is constantly changing. For example, you might change your lead generation tactics from month-to-month. Comparing the conversion rates of different cohorts based on their creation date will best allow you to assess the impact of that change because the opportunities you are grouping together are most likely to have experienced similar lead generation treatments.
So if you want to improve the manner in which you do your analysis, be mindful of your cohort definition.
Don’t want to worry about defining cohorts and the inherent complexity? Did you fall asleep in statistics class? Let us do all the hard work for you.
Image courtesy of flickr user sampsyo