We are navigating in the dark, struggling with disjointed systems and manual processes. Even with all the data, making informed decisions is getting difficult.

CDO, Waste Management Firm

Client Situation

  • A leading specialty pharmaceutical company was looking to identify avenues for faster growth, and wanted to establish a data backed method of allocation of resources across direct and indirect marketing , and across several channels of communication
  • Our approach was to build an attribution model to establish the efficacy of each channel, and build a simulation environment for client to optimize the spends based on internal and external data to improve:
    • Budget Allocation: The client had a significant marketing budget spread across multiple channels including personal communications (e.g., field calls, email campaigns, virtual engagements), and non-personal communications (e.g., programmatic displays, websites).
    • Effectiveness Measurement: There was a need to understand the effectiveness of each marketing channel on sales to optimize future spending.
  • We also made recommendations on the the optimum journey of marketing touchpoints to enhance HCP journey efficiency.

How we added Value

  • Data quality and richness was a known problem. Hence we proposed Cognitive AI (advanced statistical techniques) to build the attribution model as below:
    • Bayesian Hierarchical Models: Employed Bayesian hierarchical modeling to analyze the impact of different marketing channels on sales. This approach allowed for the incorporation of interaction effects, handle spurious activations at specific geographic regions, seasonal events (speaker programs, conferences), and non-linear relationships.
    • Markov Chains: Created k-order Markov chains with enhanced visualization to understand the transition probabilities between different marketing touchpoints and their eventual impact on customer actions (enrolment or dropout of targeted drug)
  • We leveraged GenAI to enrich data from new sources , and fed to our Cognitive AI engine to develop attribution weights, and apply them to the simulation environment through a GenAI interface. The whole development of Cognitive model was also accelerated using custom opensource LLMs


~10% Reduction in marketing cost/revenue