You might have missed this update from Google Analytics(GA). This happened when GA renamed ‘Unique Visitors’ to ‘Users’ last year. The terminology ‘Users’ has suddenly gained a lot of importance in the recent times. And the reason GA made it prominent is due to the fact that a lot of internet traffic is now being driven by apps.
The most valuable form of traffic comes from apps-
Imagine, browsing for a product on amazon. You can either do that using their app or use the site (mobile or the web) to get there. But when you visit through the app, amazon knows a lot more about you than visiting from the web. App traffic is different from web or mobile-web traffic in many ways. But the fundamental difference is- APPS NEED PERMANENT LOG-IN FROM USERS TO GET STARTED AND FUNCTION. A visit from the site does not guarantee that.
This is where differentiation between SESSIONS and USERS occurs. Traffic from apps has clear and distinguished USERS. Due to a clear and substantial identification of individual users, the dynamics of analytics and resultant marketing efforts have changed. Now, the entire focus is funneled into tracking individual app users over days and weeks.
Anonymous web users can only be identified with IP addresses that die within a short period of time. That is why they were referred to as ‘visits’ instead of ‘users’. It obviously makes sense that tracking users are more rewarding than visits. And this is where cohort analysis comes into play.
What is cohort analysis?
Many of you dealing with app analytics might already be aware of cohort analysis. A cohort is a group of people who share a common characteristic over a certain period of time. Cohort analysis is a study that focuses on the activities of a particular cohort.
I won’t discuss as to how a cohort chart is read but just for the sake of continuity, I will briskly explain and move on.
Here is a cohort chart snipped from Apsalar.com. This is how they typically look like. As you can see, the cohorts are segmented with respect to specific weeks. That means each horizontal line denotes a cohort. These are the number of users acquired each week. Over the period of several weeks, various matrices can be identified such as user retention, drop offs, engagement and so on. Hence, you can easily ascertain how long users from a particular cohort remain and use your app.
Google Analytics has recently added options for cohort analysis and Active Users. The purpose is clear- it wants to stay relevant for both app and site traffic.
The intertwined game of app retention and cohort analysis-
The only parameter that makes an App based business valuable is RETENTION. CEOs of app startups are always looking into analytics to monitor their MAUs and DAUs. User Drop offs are undesirable and worrisome. If an app is getting large numbers of sign ups and equal number drop off as well every week, it is not really a good sign either. An app with 10 million users with less retention can be less favorable than an app with 5m user base but high retention.
Generally in the beginning, app-based business startup sketch out a marketing plan to acquire as many users as possible. Over several weeks, the effectiveness of successive marketing activities can be correlated to user acquisition and app retention.And this is smoothly done using cohort analysis.
For example, in the above cohort from Appsalar (Initial Activity), out of all acquired users for the first 4 weeks, the 2nd week acquired the best lot of users. With 8.22% of users staying back till the 9th week, it is the best cohort available there as far as retention is concerned. I can look back to the marketing activities or an app update that made users stay back. On the other hand, I can also see a cohort that has the lowest retention and may conclude that my marketing efforts in that particular week got me more irrelevant users who left my app quite early.
You can now understand as to how a cohort chart can give you insights into retention and overall progress of an app-based business.
Cohorts do not intermix results-
Apps and app-driven businesses invite simultaneous changes and campaigns and this is where cohort analysis comes to save your day. There is no intermixing of results. Even if your marketing efforts are paying off in its 10th week after deployment, it will reflect back in some cohort. If you ran a successful marketing campaign that made you acquire a lot of users on any week, it won’t get reflected on users acquired before that.
For e-commerce sites or apps, one can clearly find out how long does it take for users to add products to cart and make a sale. If you are running a parallel google AdSense campaign, you may find out how it affects your overall sales. Does it speed up the process from cart to sale? If yes, then by what degree. Again, you can clearly distinguish users without intermixing.
Application of Cohort Analysis within the closed loop of Acquisition, Marketing, and Retention-
Every app based business model runs through this triad. You might know about the Indian e-commerce giant Flipkart. Like its main competitor in India, Amazon, Flipkart also employs both mobile as well as web channels to get users. However, the company pays more importance on mobile app usage and selling stuff through the app itself. Hence, there is obviously a bigger role of cohort analysis for Flipkart and similar app-based businesses.
Flipkart has made a lot of efforts to acquire users to sign up for its mobile apps. Once they had sizable chunk of acquired users, they continually rolled out a lot of attractive offers to them. They market them through notifications. Flipkart sells a lot of products from these offers time to time. This is how they retain their app users and create further marketing campaigns to keep the loop running. In fact Myntra, a leading fashion and lifestyle product site acquired by Flipkart, runs entirely on their mobile app now.
To drive the entire campaign for such business models, can you think of anything more effective than Cohort Analysis?
A similar logic applies to game applications as well. Game apps demonstrate the least retention amongst major app categories. They also run through a similar triad of acquisition, marketing, and retention. Game updates that go bad can also get reflected through cohort charts. Calculating MAU and DAU for games is crucial and Cohorts gives quick insights here.
When other matrices are combined-
There are a lot of tools in the market today to create cohort charts for your app(s) or site(s). However, my favorite of the lot is Mixpanel. Mixpanel integrates a lot of critical features to record user behaviors. These, when correlated with respective cohorts, gives substantially conclusive insights. Check out this link https://mixpanel.com/codeless-mobile-analytics/ to see what consolidated data along with cohort analysis might feel like.
App based startups are drawing huge amounts of money from venture capitalists. But at the same time, traffic patterns are different for apps. The motives are different and the way to analyze them is also contextually different. Apps have shifted the focus on individual users to create substantial and conclusive analysis. However, the only way to cash into the analytics part largely depends on your understanding of cohorts and its application.
Some other good content on Cohort Analysis-