4 steps from data to success — the Mobile Analytics Value Chain
A beginner’s guide to mobile analytics
“We just successfully survived soft launch and now we want to grow with user acquisition. What shall we make from our data? How can we get started with analytics?” —we often hear things like this from clients at our data science agency. There is countless buzzwords around data analytics in mobile—but how can you use your data to make your app more successful? To help our clients answer this question we developed the Mobile Analytics Value Chain model. It summarizes the elements you have to look at to succeed: the right data infrastructure, the fitting analytics tools and models, to engage in the right marketing tactics to finally generate growth.
Your data comes from different sources and includes user data, campaign data and tracking data.
User data: Comes from your mobile CRM such as OneSignal, appboy, or Localytics etc.
Campaign data: Comes (in the worst case) from all the different ad networks or (in the best case) from an (integrated) attribution and provider for both installs and spend such as Tenjin, Appsflyer, Adjust, Kochava, Tune etc.
Tracking data: Comes from an analytics tool such as Mixpanel, Amplitude, etc.
(For more info on the the different tools have a look at the Mobile Marketing Tech Stack)
You can either store the data separated and pull and combine it on demand, or you can automate data integration and pre-processing in a data warehouse (which is recommendable once your number of sources reaches a certain number or if it is difficult to integrate the data coming from different sources manually for reports or analyses).
By “analytics”, we refer to both business intelligence and data science applications in mobile. Business intelligence provides nice charts, reports, and dashboards. It gives all employees access to the data and KPIs they need (“data democracy”). The appropriate BI tool can help you to save a lot of time for reporting processes by automating calculations and visualizations. There are various use cases for analytics in mobile — just to mention the most common ones:
LTV modeling: Customer lifetime value (LTV) shows the gross worth of a customer through his lifetime which means it is a function of revenues and retention. LTV can be modeled in different degrees of sophistication. Typically, you start by calculating LTV in a simple model from historical data, in a breakdown of different countries, campaigns, or segments. Then, you tune your simple model to a cohort view. In the next step, you can move to a more advanced version by predicting LTV from behavior and user features.
ROI estimation: Return on investment (ROI) is the proportion of profit to spend and reflects how much you get back from a Dollar spent on advertising. It can be simply calculated from historical data or forecasted using predictive methods.
A/B testing: A/B testing (or split testing) is a method that helps to evaluate which of two (or more) alternatives is more beneficial. There are various applications for A/B testing. The most popular ones in the mobile industry are comparing creatives for campaigns, UA features, and App Store appearance. You can also use A/B testing to find the optimal pricing once you start monetizing your app.
Behavior analytics: Analyzing user behavior means to describe and model the relations between of your product or campaigns and your business goals. This can be e.g. a model that predicts whether your users are likely to convert to paying customers, to churn or it can even be a dynamic pricing model for in-app purchases.
User segmentation: Data-driven approaches towards user segmentation cluster users into homogeneous groups based on behavior and (possibly) user attributes.
Campaign analytics: Your campaign performance can be broken down into different elements such as creatives, keywords or user IDs for Facebook Lookalikes or Google Customer Match Audiences.
What can UA and mobile marketing get from analytics? Some examples:
Target valuable users: Using LTV and ROI models, you can steer your campaigns effectively. With an LTV prediction model, you can predict whether the campaign is likely to be successful using the data collected through first days of your campaign. In a more simple setting, you can use ROI benchmarks per country or segment so see whether your campaign performance is in line with successful campaigns in the past.
Optimize campaigns: User segmentation and campaign analytics help you to optimize campaigns. Segments are groups of similar users which you can target with specific creatives and keywords. Behavior analytics help to get the most out of Facebook Lookalike campaigns and Google Customer Audience Match campaigns.
Increase virality: Knowing your most engaged users through behavior analysis, you can offer them incentives to recommend your app. With A/B testing, you can optimize your performance in the app store and attract more organic users.
Improve retention: Using behavior analytics, you can create customized user journeys through email or notification systems (e.g. through your mobile CRM or through Google Remarketing). You can offer discounts and retainers to customers who are about to churn or engage them with individualized push notifications.
How does that all make your company grow? More valuable users lead to higher revenues. Campaign optimization means that you either reduce spend and get the same amount of installs or you keep spend constant and get more installs. Higher virality leads to more installs. Higher retention leads to higher revenues. More installs, in the end, translate into higher revenues.
(To learn more about what we are doing at patya analytics, visit our website or drop me a line: firstname.lastname@example.org)