Running a successful business is associated means dealing with the need to analyze large amounts of data (big data). No matter what type of business you own, you have to be prepared to explore and explain the behavior of your customers. However, knowing truly useful patterns in data collection requires additional analytical techniques, one of which is cohort analysis.
A comparison of several cohorts will allow you to discover trends and take appropriate action.
Let’s start by explaining the basic concepts like what a cohort is. Generally speaking, it is a collection of individuals who are experiencing the same phenomenon in the same period of time. In the case of business solutions usually we have people that have established an account or made a first purchase at a particular time, for example, on the first day of the month or the first day of a week. Thus are so-called static fields formed – unchanging group of customers who represent certain behaviors at the same periods. Importantly, each person can belong to only one cohort, therefore, they can not migrate to other groups. It should also be remembered that the basis of the analysis here is the time interval and not, for example, a location.
Comparing different cohorts
An analysis of cohorts can be applied to many types of data: sales volume, number of page visits, subscriptions to a specific services or conversion rates. The purpose of the analysis calss for the appropriate measures. For example, if you want to discover the real value of customer life (CLV), you can rely only on the general and average indicators. However, comparing levels of sales in different cohorts – for example, clients who make their first purchase in the next five months – can help uncover deeper mechanisms. An example could be the phenomenon, consisting of the fact that previously acquired clients are increasingly less likely to make purchases, and sales growth is heavily dependent on the time of year.
A comparison of several cohorts will allow you to discover trends and to estimate the revenue in the coming months, and thus take appropriate action. As the table presents, customers who made their first purchase during the Christmas and New Year’s holidays (December – January), spent the most money. However, in subsequent months, revenues begin to decline. The graph shows that new customers spend less and less, and in March sales collapse. This situation can show the high seasonality of a business.
With this knowledge you can take preventive action, for example by designing automated and personalized promotional campaigns, which start at the beginning of March. Additionally, thanks to the advanced marketing automation platforms, for example Synerise, you can vary the content of messages and promotions depending on the client’s date of registration. As a result, you will have a chance to a reach current clients and a new ones who spent relatively less during the period of March-April, than clients who made their first purchase in December or January.
As already mentioned, the analysis of cohorts can be used for different purposes. Let us consider another. Let’s say you want to make an estimation of the average value of the shopping cart in the individual sales channels. You run a clothing store, and since last year you have been offering a range of products in online channels, such as the official e-shop, mobile application and a popular auction site:
The graph shows the average value of the shopping cart in different sales channels divided into the four quarters of a year. It shows the transformation of client behavior to three cohorts of clients (i.e. Mobile application clients and buyers in an e-shop and auction sites) in the following periods of the year. While the behavior of online buyers is quite predictable, we can see a major transformation among mobile application users and buyers on the auction site.
The value of the shopping cart among the latter is steadily falling, while clients using applications buy more. How can you explain this correlation? The increase in the value of the shopping cart in the mobile application may result from getting used to clients with this channel, which can further be enhanced by the growing popularity of mobile solutions for the e-commerce market. On the other hand, decreasing the average value of the shopping cart buyers on auction sites needs a response. If you were in that situation, you would have to decide whether to maintain this sales channel as profitable, and if so, the next step would be to take action aimed to improve the situation in this distribution channel.
What are the effects of cohort analysis?
The cohort analysis grows into an important type of marketing tool, indicating the quality and effectiveness of chosen business activities, as well as giving the ability to predict future events. It helps to avoid rash and superficial decisions, which often happen when we only rely on general indicators.
If you are considering the profitability of your project, you want to determine the consequences of a given decision and uncover the factors that influence the behavior of your clients you should use the described way of analysis. As a result, you will become more aware of the actual processes and mechanisms. This is the first step towards success in business.
Reaching the described method, you should know that it won’t be useful in a time of dynamic changes in the market or your company, so to get the most reliable results you should base them on data from at least several months. Think about regular updates of your base, meanwhile search for the differentiation of downloaded data.
In the next text we will show how you can use the analysis of cohorts to count one of the most important indicators of marketing, which is the indicator of the value of customer life – CLV.