Cohort Analysis has become one of the most favorite analytical tools with entrepreneurs lately. One can find quite a few entrepreneurs swearing by insights gathered from cohort analysis. Before we dive into how we are using it at our start-up, ThanksForTheHelp, let’s first take a look at what it means and what could be some of the use cases.
What is cohort analysis?
By definition, a cohort is a group of people that share some commonality. Put differently, a cohort can be defined as a subset of users from the entire database with a shared characteristic. Therefore, cohort analysis is the behavioral analysis of this subset of users, instead of analyzing the entire user database.
Now let’s take a look at how cohort analysis can be useful.
Cohort analysis utility & use cases
Not every user behaves in the same fashion, therefore, to group them in one unit and then to do behavioral analysis over that one group often leads to erroneous conclusions and insights. And that’s where cohort analysis comes into the picture.
The users’ database is categorized into different segments (or cohorts) and then the analysis is done on each of those cohorts to gain more realistic and actionable insights.
It is one of the favorite business analytical tools for entrepreneurs and especially digital entrepreneurs. Most frequent use cases are –
Do customers acquired from different marketing channels behave the same way?
This is perhaps the most frequent use case of cohort analysis. Users acquired through different marketing channel behave differently, for example, someone who is acquired through PPC (via AdWords on Google) would behave differently from someone who is acquired through Social Media (say FB). To optimize the marketing spend therefore it becomes crucial to analyze these segment of users differently.
How do customers acquired along the sales funnel behave?
- Purchase Decision is a long process which is often broken down in the following steps –
- Information Seeking
- Mind Firming
- Actual Purchase
- Needless to say, that users acquired at different steps would behave differently. This analysis helps a company to gauge things like how many interactions are required to convert a user acquired at say information seeking phase to a paid
Is there any variation in users’ behavior depending on the month they were acquired?
This is particularly true for those businesses that are cyclical like ours. Users acquired in lean business months would naturally lead to lower LTV than users acquired just at the beginning of the business cycle. Such analysis helps companies plan their user acquisition and LTV numbers better.
Now let’s take a look at how we are using cohort analysis at our business.
How are we using cohort analysis?
We at ThanksForTheHelp, regularly employ cohort analysis to gauge LTV of a customer, retention metrics and growth. We use it to test the quality of our services and see if we are doing a good job of making our customers happy.
Here is a snapshot of the cohort we follow –
(* Figures are not real, yet representative of the business)
Before we take a deep dive into the analysis part, let’s see how we create this cohort.
So, this represents a group of customers acquired month wise and how they give us business over a period. The first row of every column represents business generated by customers in the first month of acquisition. The second row represents total business from those customers over 2 months and so on…
Therefore, it would be read like –
“Customers acquired in month M1 gave $110 in the first month (i.e. M1 month). Also, those same set of customers resulted in the business of $135.6 over M1 and M2. Eventually, they resulted in the business of $342 over 13 months of stay with us”
In case you are still wondering how to make a cohort, here are some useful links.
Now let’s look at how we use this to gauge the quality of our services, retention and business planning.
LTV of customer
The last row of every column represents total business given by customers that were acquired in a particular month. This is nothing but LTV of a customer. Though we could find LTV by simply pulling out the data from DB but seeing it like this makes it more actionable.
It helps us know how much customers would pay each month to reach their LTV value. And this progress helps us plan our revenues betters. As they say
“The means are as important as the end”.
Similarly, just knowing the LTV wouldn’t make that much sense as knowing how that LTV is reached.
So, it is visible that our LTV of a customer is tentatively around $350. But we are reaching to that number faster as is visible when we move from left to right in the cohort. What does this tell us about the business?
This tells us that not only we are retaining the customer but also getting a bigger piece of the pie – an individual’s entire business. Since we are in study help industry, this means that a student is coming to us more frequently that he used to do in the past. And as an entrepreneur, there is no better news than this.
We see that customers acquired in different months behave differently, and it is expected (after all that’s what cohort analysis is for). So, when we use this data to plan for future, we know how much business in a month would come from existing customers and how much would need to be brought from new customers in order to achieve our targets.
This sort of business planning gives immense confidence to an entrepreneur as it is based on real data (your own set of customers) and the deviation wouldn’t be much thereby lending much more credibility to the plan you make.
There are ample more use cases for cohort analysis and its use is limited to one’s imagination. If you use it some other way, do mention it in the comments. Probably I will also implement it in my business.