devtodev has implemented an LTV prediction tool based on machine learning. Its accuracy is up to 90%

The devtodev analytical platform has introduced a calculation model that, according to its developers, allows predicting the revenue of mobile applications with an accuracy of 85-90%. The tool is based on machine learning.

To make a forecast of the model, three days’ worth of data about the application is enough. However , the most accurate forecast is obtained under the following conditions:

• availability of data for 90 days;
• MAU of the project from 30 thousand;
• the presence of integrated events that inform about making payments and launching events.

To predict LTV, the devtodev platform uses information about purchases, as well as about the actions, activity and properties of users.

“The innovation will allow devtodev customers to choose only those traffic sources that can pay off, as well as compare LTV indicators before and after making changes to the product,” said Alexander Mezin, devtodev Business Development Director.

The model is already available for use by all users of the platform.