What is a cohort analysis anyways and why should I care about it?
Cohort analysis is a method to see how different groups of customers in your business interact with your product over their life-cycle. This can help a SaaS company to take action for catering to the different groups of customers they may have. For a SaaS company, there are many ways to make cohorts of your customers:
- Acquisition date
- Acquisition channel
- If you have tiered product offerings, which plan they started with
- Their industry
- Use case for your program
A familiar example
One cohort analysis you might have seen before is the one you can find on your Google Analytics account. These cohorts are based on when the users first visited our website. Different rows that we see on the left are the different cohorts, seperated by when they first made their visit. Then, starting with 100%, we see week by week the percent of those from that cohort that visited our page again over a 12 week period. Each cohort analysis will start with 100% in the first column. Then, each next column will show out of the people who first visited our website between those dates, the percent of them that have visited again.
Cohort Analysis for SaaS:
This type of analysis can give us a great deal of information about our customer groups. For a SaaS business there are many different metrics you can run a cohort analysis on that will give actionable insights. Here are couple scenarios which you can answer by running a cohort analysis.
What does our churn rate look like across different cohorts?
Instead of having a single churn rate you track, you can track churn rate based on account size. This could mean creating cohorts of your customers based on how many users they have. Or, you could create cohorts based on how much revenue they generate. Since most SaaS companies operate on a monthly subscription basis, instead of having weeks, you would have months as the columns. You could run your churn analysis on revenue instead of number of customers. This would show you whether you had any revenue expansion, whether your churn rate is different across these cohorts, and when customers start churning rather than a single churn rate. As much as revenue and churn is a sales metric, seeing churn rate variance across accounts with different user sizes would have implications for both product, customer success and marketing teams.
As much as revenue and churn is a sales metric, seeing churn rate variance across accounts with different user sizes would have implications for both product, customer success and marketing teams.
Are accounts with different sizes getting different amounts of value from using our product?
Perhaps if churn rate is significantly higher in the first three months, and then it steadies out it could be a question of whether the customers were a bad fit in the first place or they left because they couldn’t get the value they were hoping for. If the customers were a bad fit, it should raise a red flag since in most cases customer acquisition costs are very high. A good lead qualification process is absolutely necessary to make sure no extra resources are going to be spent on customers who are not going to be adding value throughout their lifetime. Finding out why a customer who wasn’t a bad fit was targeted, qualified and sold to is a very important question. Customers that quickly churn will give bad reviews and hurt your business in many ways. A more common reason could be that as your users who are asked to start using your platform, something they are not familiar with and may show resistance at first, have not been properly onboarded. Perhaps they need more guidance after implementation and need biweekly checkups. Making sure that new customers quickly start getting value out of using your product is the best way of reducing churn which will usually be higher in the first 6 months.
Do the customers we acquire from different channels behave the same?
Let’s assume your marketing and sales teams have both inbound and outbound campaigns running concurrently. You might have search engine campaigns, online blog content, listings on different industry sites, paid content, and outbound email campaigns all going on at the same time. Even if you were tracking different acquisition costs for each channel by attributing each lead to it’s source campaign, you can learn a lot more by looking at how long customers from each source stay, how big their accounts are, what their lifetime value is, and a lot more by doing a cohort analysis. Suppose your outbound team only targets high value accounts and the search engine campaigns bring companies of different sizes. You know how your customers get value out of using your product and that can be quantified. Perhaps, number of logins, number of documents created, time spent using your app, any one of these could be a good indicator of a user getting value out of your product.
Comparing how customers from different acquisition channels use your product differently could give you access to a market you didn’t even know existed. Sometimes your product will have a fit with a group of users that you perhaps hadn’t intended. These insights could greatly change your marketing efforts and positioning. To sum it up, a cohort analysis has three components:
- The cohorts, or the rows. This is the group you define. It can be anything you want to group by. (industry, acquisition date, the plan they started with)
- Date range you want to run your analysis on.
- The metric you want to track over this time. Again, this could be anything, (page visits, user count, revenue, number of times they did something on your app) Part 2 will be about collecting and shaping your data for a cohort analysis.