Brand management BigData SaaS


Complex analytical model


Knowing your brand and having a hang of how to promote, develop and embrace new technologies always was the key to success. So, a lot of companies strive to make data science assist them for increasing brand value and getting better knowledge of the customer.  One of the FMCG leaders, Large tobacco company, approached us with a  need to expand their brand-line alongside with increasing the number of distribution outlets. The Initial and foremost goal was specified as prediction of customer response to various marketing collaterals and design elements, i.e. different advertisement types, external branding elements such as outlets, big boards, and even colour combinations used in visual materials. Large tobacco company possessed a significant amount of historical data comprised of sales data and description of branding action of a particular moment. The final model was supposed to be complex and intended to detect any response within the predicted deviation of customer’s behaviour or purchasing activity which might have been caused by promotional and/or advertising activities.


Having considered the Client’s current and future needs, the key points of the solution were defined as:

  • fine-tuning of current Client ETL solution;
  • multicomponent predictive model containing several sub-models;
  • price elasticity and sensitivity model;
  • embedded Bi-solution showing customer’s response to any applied activities in both short-term and long-term time frames.

In order to implement outlined solution Unicsoft involved SME (Subject Matter Expert) to help building more precise model by identifying all dependencies and nuances in the Client’s business domain. Alongside with, a team comprised of Data Scientists and BI experts was set. Considering constant incoming data inconsistency and, subsequently, its effect on the model accuracy, Unicsoft initiated setting up of support team to continuously monitor and fine-tune model and maintain high accuracy.


The initial effect of the model became noticeable and appreciable within first four months. In addition to that, due to the complex analytical model for new activities prediction and doing business considering its “advice”, Large tobacco company managed to have a solid increase through customer response rate and, consequently, a revenue increase.

Large tobacco company pointed out a remarkable level of Delivery as well as high level of overall performance provided by deployed solution.

TECHNOLOGY & TOOLS: PythonRTableauMachine LearningArtificial intelligence

Platform: Web