To address this, the team at Tredence used the following approach:

  • Developed a multi-touch attribution model using the the Markov chain concept to accurately identify the contribution of each channel in the conversion of a user.
  • Calculated the channel index using the Removal effect on Markov model. This attributed a channel index value to the channel after calculating the number of conversions made in the absence of that particular channel.
  • Established the relationship between SEM clicks vs impression from paid social, display and retargeting to reattribute channel index using the linear model. The new index was used to calculate the ROI of each channel.
  • Optimized spend by maximizing ROI of channels using Linear programming.


  • The client can maximize the overall ROI by allocating more budget to a channel which has higher returns.


  • With reallocated spend across each channel and ROI, we would achieve ~20% higher customer conversion when compared to the previous spend allocation.