The model we built made it possible to predict over 65% of unsubscribers based on change of customer behavior. Every time the specific client shows the behavioral pattern that is likely to lead to unsubscription, she was classified as potential unsubscriber and treated differently from the rest of the clients. Model implementation dropped email unsubscription rate by about 40%.
Details
The unified system collecting information from multiple sources was developed. The sources included prediction of site visitation (split by specific pages and page zones), evaluation of the site demographics (from 3 sources), estimation of unique visitors by sites, estimation of audience intersections (by sites and by media types), information about percentage of views of the video ads to the end, estimations of expected discrepancies of visitor counts depending on the counting system.
Details
Attracting new clients is costly, so retaining clients is absolutely critical for the business success.
The predictive model of churn rate was built. It made possible to predict over 70% of unsubscribers based on change of customer behavior. Project implementation dropped the unsubscription rate by about 50%.
Details
There is hardly any monitoring of online diapers sales in Russia. The key task of the client was to cover at least 60% of highly sparse small online diaper retailers and find data for own brands and competitors. The database with automatic reporting and notification system was created to collect information about availability and price for over 200 SKUs of diapers across 50 key online stores automatically.
Details
The smart algorithm allows calculation of the fair market price of the complex product using the Market Price Calculator (MPC) approach. The demonstration is devoted to the laptop market.
Details
Individual hotel ranking solution for each visitor. The most relevant options placed on top of the search output empowered by algorithmic combination of hotels’ features, visitor’s profile and search query characteristics.
Details
The benchmark research of Market Price Calculator — machine learning algorithm, which shows high statistical significance in explaining new car model price variation. Whenever an industrial product is introduced to the regional market, key competitors positioning and price optimization strategy are to be carefully calculated.
Details
Short introduction to optimization of online casino operations: revealing major tendencies and finding most significant groups of users.
Details
The analysis helps reveal the most significant groups of e-mail receivers in terms of their actions and preferences. Can be used as the first step toward creation of personalized emails.
Details