Welcome, we are glad to see you here!

Here is some information on what we do, you can expect the same methods in a pilot project on a small scale.

We are data analysts with business, economics and statistics background.
Our specialty is in applying advanced methodologies to business problems.
Our goal is above all to find answers to your questions and find those new questions that need to be answered for businesses to succeed and compete better.

In order to better imagine what you can expect, we will talk about our methods, how we do projects and bring you methodology examples.


  • Our standard methods include:
    • Linear regression methods
    • Decision trees
    • Random forests
    • Randomized experiments
    • Propensity score matching
    • Difference-in-differences regression
  • Some of our more advanced methods
    • Bayesian statistics and reinforcement learning
    • Cox proportional hazard model and its extensions
    • Buy-til-you-die model and the RFMC approach
    • Discrete choice models and association rule learning

How we do projects:

  • Business and data diagnostics phase
    • We always start by establishing a comprehensive understanding of your business, organization, and challenges, so we can work more like business consultants instead of just data analysts
    • In the data diagnostics phase, we scan your data treasure and find the ways to make your data work to serve your business need, identifying any bottlenecks and data problems.
  • Closely working together with your team
    • We rely on interviews and workshops where ideas and results can be shared with those who need to apply them
    • We work in sprints and update you in regular (bi-weekly) meetings to validate our findings and let them work for you instantly
  • We pay attention to pass knowledge over to you efficiently
    • We understand that the most advanced machine learning algorithms are hard to interpret
    • We love data analysis because it brings clarity rather than confusion – we would like to keep it that way
    • We coach your staff in the project as much as the implementation of analytics results requires it
    • We build simple models that everybody can use on their computers and test the basic features of even very sophisticated models
    • We bring you simulations to help understand complex problems and prepare for the challenges
    • We discuss models with your team and help to bring ideas to the surface
  • Back-testing phase at the end
    • Before finalizing analytics recommendations and implementing them we do thorough quality assurance at the end of each project so your business is secure

Methodology examples:

Methodology and support for A/B testing (split testing)

  • Problem
    • You got ideas how to improve your business’s bottom line – but does it really test well in real life?
    • Be it product, design, pricing, or content – fast prototyping and testing is key to improvements and value creation
    • This service is particularly important for companies with strong online presence, e-commerce
  • Solution
    • Support with standard and Bayesian statistical methodology and related reinforcement learning algorithms, full evaluation if needed
    • Implementing a clear and intuitive methodology that everybody can understand
    • Bayesian statistical methodology with the help of machine learning lets you run multiple A/B tests at the same time without sacrificing performance
    • With adjustments for your unique situations, we work with you to find the right metrics to measure success and find best alternatives
    • We consult on test design, monitoring and interpretation of the results
    • With a dependable solution you will be able to focus on being better and more proactive

Bayesian evaluation chart of the A/B tests performance

Survival analysis and LTV calculation for SAAS companies

  • Problem
    • Do you know how much your visitor or customer worth? – Customer and Visitor Lifetime Value should be the basis of business decisions
    • Yet crunching data to get to reliable LTV estimates and to identify the main drivers behind differences is as much a science as an art
    • As marketing and operations is both involved at the same time, besides data analytics a deep business understanding is also needed to tackle the problems
  • Solution
    • We employ survival models like the Cox proportional hazard and the Buy-til-you-die algorithms for estimation and prediction
    • Our analysts calculate your segmented LTV values and our consultants advise you on the business aspects as well
    • Bringing you methodologies that forecast trends earlier so you can react faster to changing environment and customer tastes
    • Valuation will be put on objective grounds where debate drives understanding not hunches

Visualize customers’ churn and survival by customer categories

Choice models for product conversion

  • Problem
    • Do you know how and why your customers choose between your products? – Business responses needs to be based on deep understanding of demand
    • However, multiple relevant factors are at play and influence customer demand and so simple methodologies can lose their credibility and predictive power
    • Optimization of prices and targeting can only be based on models that capture the real drivers of the choice
    • Abstract statistical models cannot be interpreted widely in your organization and will not be acted upon
  • Solution
    • With the combination of A/B testing and discrete choice models, we can optimize your pricing for maximum customer portfolio LTV
    • If you are working with affiliates with this tool you can make thoughtful and solid decisions on their commission structure
    • We employ discrete choice models like multinomial logit regression and association rule learning algorithms
    • We also build you models and dashboard that your people can use to visualize and understand the results of the analytics

Explore the demand model by setting example parameters to explanatory variables

Potential: lung cancer detection based on CT images (example)

  • Problem
    • Time-consuming and data intensive tasks like preliminary diagnosis can be partially automated and specialists supported
  • Solution
    • CT images can be transformed into data and analyzed with standard methods
    • Machine learning algorithms can be trained based on private and public samples
    • Implemented algorithms provide decision support for specialists and can improve their understanding as more data is collected (reinforcement learning)