Empowering the client with the benefits of customer happiness index prediction - Beinex

Empowering the client with the benefits of customer happiness index prediction

With the implementation of the predictive model, the client was able to accurately observe the gaps in their customer service model and deploy solutions to ameliorate the overall as well as categorical index performance.
Industry Industry



Region Region


Middle East



Alteryx, Tableau

Client Client


Our client is a regulatory body that oversees the telecom sector and the licensees in a Middle Eastern nation in accordance with the Federal Laws applicable. It is an independent body, and its duties include ensuring telecom services are made available to all regions of the country, assuring that the licensed operators fully follow established rules and regulations to ensure sustainability, competitiveness and transparency among the service providers, customers and shareholders.

Requirement Requirement


The customer service model of the authority was plagued by recurring issues which resulted in negative impact on the customer satisfaction. This, in turn, caused issues in enforcement, compliance and monitoring.

A machine learning model for dynamic customer happiness index prediction was identified as the solution to effectively monitor and plug the gaps in service delivery.

Challenges Challenges


  1. Lack of well-defined variables to identify and build the required ML model.
  2. The available data was unstructured with little or no warehousing.
  3. Challenges in improving the accuracy of the prediction model.
Process Process


The solution was modelled as a multi-pronged approach which involved:

  1. Alteryx Designer – For developing the ETL workflow and deploying the ML model.

  2. Alteryx Server – For scaling the solution across organization as well as scheduling the same for dynamicity.
  3. Python (programming language) – The scripting language of the ML model.
  4. Tableau Desktop – For dynamic & interactive visualizations to help the business users in decision-making.

The data architecture available at the client was unstructured. Also, it didn’t have any kind of data warehousing methodologies deployed, making it difficult to structure the variables as well as coming up with an ETL strategy.

This bottleneck was overcome with the use of Alteryx Designer in the data preparation process by creating analytic workflows easily with the readily available no-code, low-code automation building blocks. The analytic flows were created in an incremental fashion with lucid and detailed documentation at each step, to future-proof the solution for end user engagement.

The analytic flows considerably reduced the manual effort (≈ 90%) involved in the variable selection while completely eliminating the effort involved in the variable calculation.

The Machine learning model is an essential part of the solution because it can analyze bigger and more complex data while delivering faster, more accurate results at larger scales. The identified model for this solution was the Decision Tree algorithm. Decision Trees are a supervised learning technique that predicts values of responses by learning decision rules derived from features. The model was designed using python and deployed in the Alteryx workflow making use of its in-built integration feature.

The Alteryx server was used to scale and deploy the functionality across organization as well as to schedule the same for recurring results corresponding to updates in data over pre-defined periods. In this case, a weekly periodicity was implemented.

The decision tree algorithm was used to find out the underlying parameters that contribute to customers’ happiness. The model provides the best combination of services that are to be constantly monitored and delivered to ensure customer satisfaction.

The data so generated have been visualized using the Tableau Desktop using its market-leading capabilities for businesses to discover and share insights. The backend data have been dynamically integrated with the Alteryx workflow to reflect the updated data from each periodic run.

Result Result


  1. The authority has a reliable and scalable solution to dynamically monitor customer satisfaction.
  2. The project also inspired the team in following the best practices in Data governance and Data warehousing.
  3. The variables found during the analysis have found cross-functional utilities within the organization.
Key Key

Key Takeaway

Today, the model can predict the satisfaction variance of customer with an accuracy of 94%.

Client Client Requirement Requirement Challenges Challenges Process Process Result Result Key Takeaway Key Takeaway