Or how to start analyzing a mobile application.
Mobile app analytics should be your best friend if you are planning or already developing an app for iOS or Android. It will help you understand what needs to be optimized and in which direction to move to achieve the project's goals.
Without analytics, you can try new features, play with optimization, or conduct experiments for a long time without understanding how all this affects the key metrics of a mobile application.
After reading this article, you will get a general idea of what mobile application analytics is, where to start and where to move, and what you should definitely not do.
Mobile vs Web
If you have ever used Yandex.Metrica, then a further analogy will help you better understand why mobile analytics needs to devote a lot of time and attention and why you cannot do just setting the usual “counter” for websites.
Installing the analytics system into the application
It all starts with installing the analytics system into the application.
Even a banal installation of analytical systems code into an application is a rather laborious process that will require the involvement of developers. There are many pitfalls here. And if you consider that any change also requires re-moderation of applications in Google Play and the App Store, then the process is not only laborious, but also time-consuming.
Complexity of analytical services
Service interfaces for collecting and analyzing data are often quite complex. You won't be able to figure them out in one day.
And if Yandex.Metrica for websites is a mass product with an intuitive interface, then all analytics services for Mobile are primarily focused on specialists and require many days of studying the documentation.
Functionality of mobile applications
Most of the sites are of the same type: landing page, corporate site, online store, etc. The approaches to the analysis are also template, and therefore you can simply install a counter on the site, set goals in a couple of clicks and start receiving data for analysis.
It's not like that with apps. Each mobile application is specific and has its own set of functional elements. A large zoo of technological stack options, specific functionality, various tasks ... All this does not allow unifying analytics systems. That is why each application is a new project for data analysis.
The analysis requires a set of services
Today, no service can close all data analysis tasks in a mobile application - applications are too complex, and the analysis tasks are specific.
Analysis requires a combination of several services, databases, integrations, etc. This complex of services and applications must be carefully designed, then properly implemented and maintained.
How to build an analytics system in an application
Analytics of a mobile application cannot be done overnight. It will not work just to "fasten" an analogue of Yandex.Metrica. By inviting a specialist a week before the planned release, you may be very surprised to receive a technical specification for developers for a couple of weeks.
Let's take a look at what needs to be done to make everything work as it should. But since this is the “right thing” in each case individually, I propose to consider 3 main options for the development of analytics in your mobile application.
Note that within the framework of this article, I plan to outline only the essence of each of the options, implementation details are topics for separate articles.
Basic analytics
At the initial stage, you can get by with installing one of the available analytics systems into the application and marking up the events that you plan to track. This is a good option if you are just launching an application or have a limited development budget.
There are several systems for tracking data in mobile applications on the market. They are free, shareware, and paid.
The most popular mobile analytics systems in the Russian-speaking segment:
- Yandex AppMetrica (free)
- Google Firebase (shareware)
- Amplitude (free up to 10 million events per month)
- AppsFlyer (Paid, from $ 500 per month)
What needs to be done?
- Decide on a data tracking system.
- Prepare a technical assignment for installing an analytical system SDK for developers.
- Prepare an event map for markup in the application.
- Implement analytics into the application.
- Test data collection.
What are the costs?
The cost consists of the cost of the data tracking service (if you choose a paid service), the cost of the programmers' work on the implementation of the analytics system and the analyst's services, who will perform points 2 and 3.
In the most economical option, you can try to do without an analyst. Then the implementation of the system will cost within 10-15 hours of the developer's work and your time to prepare all the necessary technical specifications.
What tasks will help to solve?
Tracking user actions in the application and sources of installs at the initial stage will allow you to understand such basic things as:
- Traffic sources (which of them are effective and what is the conversion for them)
- User activity . Information about DAU, MAU, Retention and other metrics based on user actions in the application.
- Profitability . If your app offers in-app purchases, it might be possible to rate Revenue, ARPU, ARPPU, etc.
- Audience and behavior . What users come to your application and what are their patterns of interaction with the product.
A properly selected and configured analytics system will allow you to close up to 80% of analytical tasks that you may have in the first months and even years of the application's operation.
Advanced analytics (more data)
If your application has been functioning for some time and is generally developing successfully, it makes sense to think about the development of the analytics system. Further improvements will require additional resources, but with the right approach, the money spent will more than pay off due to the insights obtained from analytics.
If you already have Basic Analytics implemented, you can start adding new data to your system or enriching existing ones. This can be data about your users from your own database, expenses from ad offices, data from external systems, etc.
How can you strengthen your analytics system?
- Set up a unified analytical data warehouse (DWH). A database that will collect data on user actions from various sources.
- ( , , .) DWH.
- .
?
- .
- . , .
- , , .
What tasks will help to solve?
By connecting new data sources, you will be able to combine information about user behavior with their client profile from the application, build end-to-end analytics on users, understand exactly which users from which channels you are attracting and how much it costs you.
If we evaluate the ratio of the Basic and Extended options according to the Pareto law, then the Basic option is the very 80% that can give the main result. But when you have a stable product that makes money, an additional 20% growth from analytics can dramatically improve the performance of your application.
Endless prospects
This option should be considered if you have a successful project and a sufficient user base. At this stage, you go beyond simple data analysis to find insights and move on to using the data in the product itself.
With the help of the accumulated data, you can start building predictive models, recommendation systems, that is, use the already accumulated data in order to predict the behavior and increase the value of attracted users.
Development in this direction goes beyond product analytics and flows smoothly into the field of Data Science.
data usage
In all options, we only consider approaches to data collection. But the data is collected in order to use it for the benefit of the project.
The most common option is data visualization using BI systems . Pivot tables, graphs and charts are what are most commonly used in companies to make business decisions. This may seem like a trivial task, because we all know how to build graphs in Excel, but the task is not so easy if you study it in more detail. Therefore, the project involvement of a specialist in BI-systems for primary development will avoid many mistakes.
However, analytics is not limited to visualizations, graphs and data dumps in the form of pivot tables. If you collect enough data about your users, you can design effective recommendation systems for users, which will increase the average check, increase user retention, etc.
In addition to using the data in the moment, if you have enough data, you can think about building models. that will be able to predict certain events in the future with a certain degree of probability - this is predictive analytics.
And that's just a small part of how data can help you grow your business.
Abstract about the article
If you are short on time, here is a short summari:
- , ? , — , .
- , , . , — .
- — , , .
- — . .