What You Need to Know About Cohort Analysis and Predictive Marketing



Cohort Analysis chart


As marketing becomes more data-driven, businesses are increasingly demanding better insights about their customers. Top-line metrics such as revenue and total customers will always be important, but these numbers may not show recent successes, or more critically, can hide looming engagement problems that require urgent attention. Becoming more sophisticated in tracking customer behavior and trends is increasingly important in predicting where your business is headed.

One common method to observing changes in customer behavior is the cohort analysis. First used in the world of medicine and research, cohort analysis observes patterns between different groups of people over time. By monitoring multiple cohorts, researchers can observe the effects of different procedures and treatments across multiple groups, and identify common patterns.


Cohort Analysis

So what exactly defines a cohort?

At its most basic level, a cohort is a group of individuals who share a common characteristic within a defined time period. This “characteristic” is usually a similar action taken in a time period, such as “making a first purchase in January 2015”.

Cohort analysis is an important tool for understanding how customers behave over their lifetime with your business. It allows marketers and executives to better observe the impact of campaigns and projects, and to monitor the performance of their customers. In fact, many of the businesses we work with use cohort analysis to help project their future revenue and customer engagement.

Note that cohort analysis is different from demographic analysis. Demographic analysis is the practice of analyzing behavioral differences across age, income, gender, location, and more. Cohort analysis focuses on analyzing differences between groups when they were at the same stage in the customer lifecycle. Nonetheless, incorporating demographic analysis (such as location, income, age, and more) as an additional layer helps to make cohort analyses even more powerful.


Why This Matters

How does this play out in practice?

Say your marketing team launches a new 60-day welcome campaign in September, with a series of discounts and offers to drive customer growth. Backed by a strong customer acquisition campaign across display advertising and social media, your company is attracting hundreds of new customers every day. Five months on, your number of customers has gone up significantly, and management is happy with the results.

However, a closer look at each monthly cohort might reveal that the newer customers, brought in through the acquisition campaign, make fewer additional purchases after the initial two months. By contrast, customers who joined prior to the campaign, such as the August cohort, continue to show high purchase engagement through to Month 5.


If this company only looked at total revenue generated per month, they would have seen revenue growth simply from the influx of new customers making their first purchases. However, the behavior of newer cohorts indicates that revenue will fall once the acquisition campaign ends. This sharp drop-off in revenue after 60 days should concern any marketer hoping to grow their loyal and engaged base of customers.

As shown above, cohort analysis helps businesses to monitor real customer trends over time, and to avoid making false assumptions based on a simple look at overall numbers. By analyzing each cohort and monitoring behavioral differences, marketers can segment how they reach out to different groups, and make changes to ensure that they are serving each cohort differently. Clearly, in this case the marketing team will need a new strategy to keep newer cohorts engaged past Month 2.


Applications of Cohort Analysis

As seen above, cohort analysis helps to generate new business cases and customer insights beyond those shown in top-line metrics. With cohort analysis, businesses can drill down their analysis to monitor a single action or variable, across multiple sets of customers, to find patterns and pro-actively address problems. So how can you use this in your business?

To give just a few examples, cohort analysis can help you answer questions like:

  • Are our customer retention rates going up?
  • How valuable are the customers we acquired in the latest campaign?
  • Are our newest customers spending more in their first month than previous ones?

For instance, if you notice that your two most recent cohorts are averaging lower 90-day spend, this may be a warning sign that your outreach campaigns are becoming less effective over time. Alternatively, analyzing the behavior of cohorts from the previous two holiday seasons might help you to predict an influx of new customers in December, but to also know that 80% of them will drop off by January. The applications of cohort analysis are endless.

In a world where marketing practices and campaign effectiveness are constantly changing, keeping an eye on cohort analysis helps businesses to monitor their performance in real-time and project future revenue.




Learn more about Cohort Analysis

An Intro to Cohort Analysis is brought to you by Canopy Labs, a predictive analytics company. We work with businesses of all sizes to better understand – and sell to – their audiences. If you have questions about this guide or are interested in learning more about customer analytics, feel free to get in touch!



Click here to read more Canopy Labs resources