Boosting growth with a Markov Chains Stochastic attribution model

SUMMARY

Traditional attribution models (e.g. Last-Click, Time Decay, Position-Based, etc.) have limitations in properly estimating the true impact of each marketing channel and keyword in the conversion path. This is especially relevant when dealing with large B2B operations with $XM quarterly budgets, where thousands of keywords come into play. The resources needed to succeed with attribution are often underestimated as attribution is frequently perceived as a one-time effort and not an iterative process.

Despite such obstacles, proper attribution, in this specific case, was a pivotal point in unlocking additional growth for the client. An inadequate evaluation of upper funnel efforts represented a crucial roadblock to accelerating incremental advertising investments.

We leveraged technology to help our client overcome the limitations of heuristic models and bring us closer to the truth of which advertising channels were really aiding our customer. We used a stochastic process to understand how much each individual keyword increased the chance of a user eventually converting further down the line.

THE CHALLENGE:

Properly estimate the true impact of each marketing channel and keyword in the conversion path.

THE SOLUTION:

The Markov Chains attribution model demonstrated the value of upper funnel keywords on Google Search
advertising, which are usually overlooked in a last-click world. We shifted the business strategy towards a more upper funnel oriented approach, unlocking further investments in advertising.