Solving complex business issues
with Data Science and Big Data

When we can help
Data
You have data for analysis and models applicable to business issues
No team
You understand that your current processes are suboptimal and you struggle to improve them with internal resources
Need impact
The potential financial gains from improvement is significant, so engaging an external team looks plausible
Our solution
We specialize in customized solutions that take into account the specifics of your business and provide it with optimal results
Contact us
If you have questions or suggestions,
feel free to contact us through the feedback form

Our experience
Virtual client database
The client has several business lines, each with its own database, but they have no ability to build cross-business analytics.
At the same time the client wants to keep the client base structure unchanged.

  • We built the probability graph of customer profiles' matching based o  ID data and action history.
  • We conducted comparison and clustering of possibly duplicated profiles.
  • Offline and online data were combined in near real-time stream processing mode.
  • Increased model accuracy and overall analysis quality with enhanced user profiles.
  • We created a tool for making different representations of the connection graph for separate business task: advertising, marketing etc.
  • New data warehouse increased analytics processing speed, both due to the logical model adaptation and due to high-performing technical solution.
Marketing communications solutions
Inefficient email and sms marketing (low response rate).

  • We built a model, which allows to define the client response probability.
  • As a result conversion rate was significantly increased (email and sms). We utilised an approach which allows to estimate the immediate effect of new communication types.
Anonymous visitor identification
Majority of website visitors are anonymous, therefore the capability to estimate the effect of marketing activities, personalise offers etc is low.

  • Client websites were enhanced with webtracker code, which tracks user's behaviour. Connection graphs of online and offline user profiles were enriched with collected user tracking logs.
  • Returning anonymous customers identification.
  • Significantly improved model for channel effectiveness estimation.
Sales force effectiveness analysis
Client needed to optimise sales force motivation system through insights from data analysis.

  • We made a sales force effectiveness model based on their characteristics, behaviour dynamics, training etc...
  • The results allowed to identify several problems and adjust recruitment, training and motivation policies.
Scoring of goods redemption
In the delivery service in the web store, there was a large percentage of non- redemption goods in the "order online with POS pickup".

  • We have built a scoring model of redemption probability estimation and optimised processes of communication with the risk group.
Demand forecasting
On top of demand forecasting model, the client required to take into account goods amortization and supply chain processes. Key features: estimation of safety stock according to economic factors and the effective distribution of goods in the deficit conditions.

  • The implementation of the system increased cost savings by 25% from EBITDA.
Airport imitation model
An airport required a simulation model that would allow it to determine the effects of changes to the configuration of ground transportation and airport network so it would not depend solely on the expert opinion.

  • We built a high performance simulation system.
  • Implemented visual "What-If analysis" tool.
Forecasting ATM cash demands
Major ATM network requested to improve an existing expert model for replenishing of cash levels at ATMs.

  • We built a model for optimal cash inventory levels estimation and replenishment procedures forecasting.
Educational content recommendations
A company has lots of training courses, which are advertised to existing users in a non-effective manner, resulting in low conversion rates.

  • We developed personal course recommendation algorithms that took into account user profile data, history of their behavior, and the content of the courses.
  • Learning conversion rate has been significantly improved.
Optimization of targeted banner advertising algorithms
Company (top5 in RuNet) implemented its own targeted banner advertising system, which had low margins from advertising campaigns.

  • We conducted a research of advertising banner traffic system features.
  • We built mathematical optimization models of ad impressions and pricing campaigns and bids for RTB.
  • We implemented new high performance algorithms, which increased the company's margins.
Optimization of product recommendations for remarketing
The company had a B2B-service for remarketing of products with sufficiently low proposal conversion (close to breakeven limit).

  • We developed machine learning models for product classification according to their compatibility etc., based on product data, consolidated from various sources.
  • We developed models and algorithms to personalize product offers based on user behavior history available on the company's customers and its partner sites. We included solvency, LTV and other integrated customer data.
  • Proposals conversion was increased by 34%.
Predicting server failures
The company has a large pool of data centers with server hardware (hundreds of units). It was crucial to provide an optimal SLA, which would take into account failures, aging, and seasonal characteristics of load balancing.

  • We developed an adaptive model of server hardware failures based on trend analysis of its functional characteristics. After the model implementation, server failures decreased by order of magnitude.
  • Data Center Service Management received a service that optimized its work and reduced the peak load on the staff due to the massive failures etc., maintaining the awareness level of deviations within 3σ.
  • Company management received reporting system for failures and maintenance of server hardware, depending on many factors, including design features, components layout and temporary factors, which allowed to make better decisions about purchase and design of equipment in the long term.

Contact us
If you have questions or suggestions,
feel free to contact us through the feedback form
Made on
Tilda