Cohort analysis: 3 cases
Cohort analysis is a business performance analysis method. The bottom line is to analyze the behavior of groups of people united by any sign in time.
Evaluation of the product is not on the final metric, and for each individual cohort of this metric. A cohort is a group of people who did the same thing in a certain period of time.
Users are divided into cohorts, for example, at the time of the first visit to the site / registration / installation of the application. In the future, the analysis of the actions of the user is carried out within each cohort.
')
3 case studies on the use of cohort analysis were prepared using the t.onthe.io service.
The result of email mailing on site X - the conversion of the sent letter to referrals was 12%. Users who registered 3 weeks ago (yellow graph) follow the links in the letter 2 times more often than users who registered 2 months ago (green graph).
Based on the data obtained, we can conclude that when planning such a newsletter, you need to focus on newer users. Since those who have registered before - either are more loyal to the product (they make up the core), or have switched from the letter by chance.
Company X has launched an ad campaign in Adwords. If we evaluate its effectiveness only on the user's profitability on the day of the acquisition, the results will not be indicative.
Users on the first day of life are most active and bring 30% of all profits per day. The next day they bring 10% of profit, the next - another 10%. Thus, the effect of advertising transitions accumulates, and money continues to flow from users attracted some time ago during the entire period of their use of the product.
The general graph of mailing letter conversion shows a stable number of transitions with little fluctuation. If you analyze the individual cohorts, you can see the subsidence on the graph of users of the second week. This was not visible on the general schedule, because on the same day an advertising campaign was launched from Case 2, which increased the number of new users with a higher conversion of letters received on the day of registration.
It is important to note that the efficiency of new users has not changed, but their share in the total mass has grown. As a result, subsidence in the distribution metrics disappeared by the marketing effect.
Number of transitions:
Clicks in percentage:
This analysis allows you to quickly look for sources of drawdowns, distinguish the effect on key metrics of changes in marketing or a product, and quickly correct the situation.
Evaluation of the product is not on the final metric, and for each individual cohort of this metric. A cohort is a group of people who did the same thing in a certain period of time.
Users are divided into cohorts, for example, at the time of the first visit to the site / registration / installation of the application. In the future, the analysis of the actions of the user is carried out within each cohort.
')
3 case studies on the use of cohort analysis were prepared using the t.onthe.io service.
Case 1: Mailing List
The result of email mailing on site X - the conversion of the sent letter to referrals was 12%. Users who registered 3 weeks ago (yellow graph) follow the links in the letter 2 times more often than users who registered 2 months ago (green graph).
Based on the data obtained, we can conclude that when planning such a newsletter, you need to focus on newer users. Since those who have registered before - either are more loyal to the product (they make up the core), or have switched from the letter by chance.
Case 2: advertising banner
Company X has launched an ad campaign in Adwords. If we evaluate its effectiveness only on the user's profitability on the day of the acquisition, the results will not be indicative.
Users on the first day of life are most active and bring 30% of all profits per day. The next day they bring 10% of profit, the next - another 10%. Thus, the effect of advertising transitions accumulates, and money continues to flow from users attracted some time ago during the entire period of their use of the product.
Case 3: trends inside the metric
The general graph of mailing letter conversion shows a stable number of transitions with little fluctuation. If you analyze the individual cohorts, you can see the subsidence on the graph of users of the second week. This was not visible on the general schedule, because on the same day an advertising campaign was launched from Case 2, which increased the number of new users with a higher conversion of letters received on the day of registration.
It is important to note that the efficiency of new users has not changed, but their share in the total mass has grown. As a result, subsidence in the distribution metrics disappeared by the marketing effect.
Number of transitions:
Clicks in percentage:
This analysis allows you to quickly look for sources of drawdowns, distinguish the effect on key metrics of changes in marketing or a product, and quickly correct the situation.
Abstract
- Cohort analysis is a relatively new method of effective analysis, more here .
- The most popular factor of division into cohorts is the first visit to the site / registration / installation of the application.
- Cohorts allow you to analyze trends within a metric and distinguish product metrics from project growth metrics.
Source: https://habr.com/ru/post/262025/
All Articles