How Cohort Analysis can help you build a more sustainable business

You built a website to do a lot of the heavy lifting for your business or brand—to promote, spark interest and ultimately generate sales. Perhaps you also run promotions to boost your numbers and nurture long-term relationships. But how hard are your website and promotional efforts working for you? How can you know which parts of your site or promotional efforts are performing best against your business goals and which are under-delivering along the same lines?

The answer: Cohort Analysis.

Simply put, Cohort Analysis is an analytical technique used by digital marketers that observes and analyzes how the behavior of a group of people with common characteristics changes over time. I’ll explain.

In Cohort Analysis, groups of people—usually users/customers of a website or digital platform— are called “cohorts.” Marketers create cohorts by grouping people by a shared common characteristic, such as the month they first purchased, the source channel of their acquisition, or their demographic profile. Then, they track the cohort’s collective behavior to draw marketing insights they can use to inform improvements to their marketing efforts (e.g., website, promotion, etc.). 

For instance, say you’re running ads to acquire new customers on a given digital channel, and those ads are performing well and driving new customers to your website. You can confidently say that the channel appears to be a good source of new customers. Next, you want to know whether it is also a good source of loyal customers; whether customers acquired through the channel are sticking around and becoming loyal customers. 

This is where you turn to Cohort Analysis. First, you would group all customers acquired from that particular channel into one cohort, and then you would track that cohort over time against metrics that measure loyalty, like average lifetime value (LTV) or repeat purchases. For example, if the cohort’s LTV increased over time, that would signal that those customers are becoming more loyal as time passes. For a paid subscription business, an increase in LTV is an important metric that reflects business health by showing that paid subscribers are sticking around longer and churning less. Another example is the average number of repeat purchases for the cohort. If the average increases over time, it means first-time customers are returning to buy again, which is essential for any e-commerce business. 

Of course, these examples are oversimplified to illustrate a concept. In the real world, you’d probably group customers by more than just one common characteristic. For instance, creating a cohort that combines all customers who signed up who meet the the following criteria: 

  • (1) Was acquired from a given marketing channel (as used in the previous example)
  • (2) Signed up at a given time of year
  • (3) Signed up when a given promotion was running on that channel

The latter two commonalities could impact the type of person attracted to signing up, and the type of person could very likely dictate their level of loyalty (as customers to a given product or service). Clearly, narrowing your cohort to those that match all three characteristics would give you a more in-depth picture from which to draw conclusions. 

Short-term growth metrics vs. long-term engagement metrics

The Cohort Analysis technique is an excellent way to get a clearer picture of the audience attributes, benefits and conditions that help drive your different loyalty marketing goals; metrics like average lifetime value, average rate of retention (the inverse being average rate of churn), average rate of engagement, average revenue per customer, average profit per customer, and so on. 

Say, for example, you added a new feature to your website because you believed it would prompt more trial users to become paid subscribers. Now, you want to confirm that hunch. Cohort Analytics can help. Or maybe, you want to know whether a particular promotion attracts loyal customers. For instance, maybe you’ve been running a one month promotional discount rate on your streaming service to motivate people to try it. The promotion appears to be working well; it’s prompting a lot of new users to sign up. This might seem like a grand success. Yet, when you look a little further using Cohort Analysis, you realize a lot of those same new users are canceling their subscriptions when the discount rate expires at the end of the first month. What initially looked like a great way to acquire new customers doesn’t look so promising when you look at retention. You realize the promotion predominantly attracts price-sensitive new customers who generally don’t stick around once they have to pay full price. The only way you might’ve been able to see this is through the employ of Cohort Analysis. Cohort Analysis works by separating your short-term growth metrics, in this case, the number of new users acquired during the promotional period, from your long-term engagement metrics, in this case, the average number of months those new users you acquired during the promotional period actually stuck around (also called “Average Customer Lifespan”) .

Lumped together short-term growth can mask long-term engagement problems. In other words, when short-term growth metrics and long-term engagement metrics are not pulled apart, any lack of activity by older users can be easily overlooked when new user growth looks robust. In effect, the new user growth conceals a lack of loyalty in longer term users.

More on cohorts

Now, let’s take a deeper look at cohorts. As mentioned, a cohort is a group of users who share a common characteristic and there are several types. Which type you use for your analysis depends on what you’re trying to measure. Some examples of cohort types include: 

  • “Time-based”: Time-based cohorts are customers who signed up for a product or service during a particular time frame. Here’s how this would translate to actionable data. For example, let’s imagine that 90% of customers who signed up with your company in the first quarter were still with the company in the fourth quarter, but only 10% who signed up in the second quarter were still around in the fourth. That tells you that Q2 customers weren’t satisfied. Why? What changed? By zeroing in on these numbers, you can reassess your 2nd quarter promotions or offers and tweak them to be as effective as your 1st quarter.
  • “Segment-based”: Segment-based cohorts include customers who purchased a specific product or service in the past. You can segment these customers by type of product or service. If you find that customers who bought a particular product/service churned faster than those who purchased another product/service, it’s a red flag that changes are needed.
  • “Size-based”: Size-based cohorts refer to the size of organizational customers who purchase a company’s products or services. That is, small, mid-size, enterprise-level, or startup businesses. You can determine the channel leading to the most significant purchases by delving into size.

Defining the question

To get the answers you need, you must ask the right questions. Using Cohort Analysis to measure retention aims to determine how many customers are loyal to your brand and will provide repeat business. So paying attention to metrics related to customer retention will reveal telling trends and patterns. Retention metrics to watch might include:

Churn Rate 

  • Average number of first-time customers that never come back (as measured after a given time lapse)? Is that number increasing or decreasing?
  • Average number of repeat customers who stopped coming back (as measured after a given time lapse)? Is that number increasing or decreasing?

Repeat Purchases 

  • Average number of purchases per customer? Is that number increasing or decreasing?

Frequency

  • How often repeat customers return to your website? Is that number increasing or decreasing?

Customer lifetime value (LTV)

  • Average total revenue value that a customer represents to your business?