Cohort analysis

Cohort analysis is a method of marketing research of user behavior that helps to evaluate the effectiveness of a business. The bottom line is that you need to analyze the behavior of groups of people who have performed the same action in a certain period of time. Such groups are called cohorts.

What is cohort analysis for?

Cohort analysis allows you to investigate changes in the behavior of user groups over time and find patterns. The data obtained helps to improve the marketing strategy, refine the customer’s path and the sales funnel. With the help of cohort analysis, you can:

• analyze the impact of marketing decisions on retention, ROI, LTV, CAC and other metrics;
• find the most effective channels to attract buyers;
• evaluate the effectiveness of advertising campaigns;
• find ways to increase user engagement with low activity and minimize customer churn;
• see the need to connect a new promotion channel;
• increase the conversion rate;
• to assess the impact of seasonality on the level of sales;
• get to know your target audience better;
• it is more efficient to distribute the marketing budget.

Thanks to cohort analysis, companies can analyze the behavior of new users, evaluate customer loyalty, track an increase in repeat sales among new customers, or, conversely, detect a decrease in the activity of long-standing ones. This research method is used by online stores, online schools, educational platforms, SaaS services and other companies.

In the next section, you will learn about the key metrics that are necessary for cohort analysis.

Key indicators of cohort analysis

Cohort analysis uses many different indicators. There is no single list of metrics, because their choice depends on the scope of the company, the specifics of the work and goals. To conduct a cohort analysis, it is important to choose those indicators that will help to investigate the behavior of new and old customers, as well as collect information to solve current problems. Below we have collected only some metrics that can be tracked:

• ARPU of each cohort member;
• statistics of site visits;
• site page views;
• channels for attracting users;
• repeat purchases;
• ROI;
• date of purchase;
• customer outflow;
• the number of attracted buyers for a specific period;
• LTV;
• CAC;
• retention rate and so on.

For a qualitative cohort analysis, it is necessary to correctly define goals and key indicators, otherwise the collected data will not allow you to see the full picture and improve your marketing strategy. Next, you will learn more about how this research is conducted.

What is needed for cohort analysis?

Since each business is different, first of all you need to set the goal of cohort analysis, choose a tool for creating a report, form cohorts, set their size and the time of the study. It can be a day, a week, or a month.

The choice of tools for cohort analysis depends on the scope of the business and its capabilities. Today there are enough analytics services on the market, among them Google Analytics, Adjust, AppsFlyer, OWOX BI Smart Data, Kissmetrics. If it is not possible to use special tools, then you can create reports in the form of Excel tables or Google Sheets.

In the next section, you will learn more about the cohort analysis process and see what it is with examples.

Stages of cohort analysis

The process of preparing a cohort analysis consists of four main stages. To prepare for the study, follow the step-by-step guide below.

• Determine the signs of cohort formation.?At this stage, select the action based on which users will be grouped into one group. As a sign for the formation of a cohort, you can use a site visit, purchase of goods, registration, and so on.
• Set the cohort sizes. Select the time interval during which users will fall into the same group. You can set the day, week or month.
• Determine the period for generating the report.?Choose the time during which you want to conduct research and collect the necessary information.
• Set the key metrics.?Based on the goals, select the desired marketing indicator. You can analyze LTV, ROI, CAC, retention rate, time spent on the site, number of transactions, average profit, and so on.

To analyze the behavior of your audience and identify certain patterns and differences, compare metrics in different cohorts.

How to apply cohort analysis and which tools to choose

There are different complexity tools for cohort analysis. These can be the simplest reports in the form of Excel tables or Google Sheets, as well as analytics services, such as Google Analytics, AppsFlyer, Adapt, AppMetrica.

How to conduct cohort analysis:

• Highlight the signs by which cohorts are formed. There may be several of them: for example, the installation of the application, the first visit, the first purchase, and so on. It all depends on which indicator you want to improve.
• Set the cohort size — the time interval during which users will fall into the same group. It can be anything from a week to a month, depending on the business and the task being solved.
• Determine the reporting period — the time during which you will conduct research.
• Identify the key metrics for which the analysis will be carried out. This can be, for example, the time spent on the site, the number of transactions, etc.
• Then compare the metrics in different cohorts — so you can identify patterns and differences in the behavior of your audience.

Let’s look at an example of a report in Google Analytics. It analyzes the behavior of new visitors to the site from March 5 to March 11, 2020. Let’s assume that the metric we are evaluating is the percentage of those who bought a certain product on a stock. On March 5, 3,356 users visited the site, of which 6.41% bought the product on the same day. The next day, only 1.85% of these visitors made a purchase, and on the third day — even less: 1.37%. However, then there is a slight increase, and on the sixth day the indicator is 2.09%. The report allows you to set a longer period of time, so you can find out exactly at what point users from this cohort will stop making purchases.

Conclusion

Cohort analysis is a tool that requires preparation: long data collection, understanding which metric needs to be investigated now in order to improve business performance in the future. But the costs are worth the result — a deep and detailed understanding of the company’s marketing, proper budget allocation and effective data-based strategies.