Market Intelligence is an essential tool for a successful ASO strategy. But still reliable data on app markets for trends, competitors and usage statistics are extremely scarce. Mobile Action’s market intelligence tool has helped thousands of users to understand what is happening in the mobile ecosystem. Seeing how beneficial it was for our users, upon popular request we made a new year resolution to increase the coverage and accuracy of our Market Intelligence estimations to provide our users with an even better experience.
The New Updated Models
Download and Revenue Estimations
Thousands of apps are connected to Mobile Action via integrated App Store or developer accounts. We use this data to fuel our Market Intelligence data that are run by powerful machine learning algorithms. With the increase in our connected apps and widening of our geographic coverage, we began having a better idea about how markets in the mobile world change. Because of this we decided to embed this idea in to our estimation models and started over.
One problem we faced while trying to increase our coverage was lack of sufficient data points for some of the countries and categories. We were able to solve this issue when we noticed that the problem was actually a sudoku puzzle. Because, even if the data points were not enough for a specific country or a category, we could simply use our global knowledge about the category and our knowledge about other categories in the chosen country. Using this method, we were able to generate robust estimates. For instance, even if we do not have enough data about Japanese news apps, we are still able to provide our users with reliable data. As we can use our knowledge on news apps and Japanese app market to provide our users with the insight they need.
We have also discovered a correlation among paid rankings and free rankings as well. Making it possible for our models to cover apps accurately in the paid charts.
Another aspect we have improved upon while updating our models was the incorporation of daily user behavior changes, of users throughout the week. Even though our former model had also taken this into consideration, now with more data, we are able to have a deeper level of granularity with regards to these changes.
Usage Estimations (DAU and MAU)
For usage estimations we have built an entirely new model. Our former model was based on retention data and we were not entirely satisfied with it, since:
- Retention statistics are highly variable even for the apps in the same category
- Retention primarily indicates installs and not usage
To overcome these we have redefined retention and let our model understand:
- User ratings and reviews
- Install to activation patterns for different apps.
The Outcome
Because of the new models, there has been a drastic increase in our estimation coverage for apps in both Apple App Store and Google Play Store. For Apple Store, the new model has doubled our coverage for Downloads and nearly tripled it for Revenue estimation, a whopping 99.8% and 168.9% respectively.
Similarly, our coverage for Download estimation in the Google Play Store has increased by 12.9% while the increase in Revenue estimation is a colossal 256.3%.
What does this update mean for the user and what to expect in the future?
At Mobile Action, we are set on providing the best experience for our users. With our updated estimation models for Market Intelligence, the users will be able to have a much wider coverage and will have access to data that is increasingly accurate and reliable. With our constantly growing user base from around the world, we are also excited to announce that we are currently in the works of introducing new countries to our Market Intelligence tool. We are committed to constantly improve our tools so that our users always have all the data to stay on top of the competition.
