Multi Touch Attribution solution enables in flight marketing campaign optimization resulting in 11% higher campaign effectiveness

Sigmoid built an ML model based on Multi Touch Attribution to measure and optimize marketing effectiveness across various brands, delivering over 70% prediction accuracy on campaign performance.

Business Challenges

The client is a multinational consumer goods company with multiple brands in the home and personal care products. To analyze the performance of their marketing campaigns, they received half-yearly MTA (multi-touch attribution) and MMM (marketing mix modeling) reports from a third-party attribution partner. They wanted to analyze the ROI of their online marketing campaigns and design an adaptive digital marketing mix and analytics solution. Their existing approach was limited to only reactive analysis and was not built for in-flight campaign changes.

Sigmoid Solution

Sigmoid developed a self-adapting ML model that assessed the effect of marketing activity on sales. The weekly auto-calibrated model, based on Multi-Touch Attribution, analyzed each marketing touchpoint's impact on conversion. Automated data pipelines were built to ingest data from multiple sources. Model training and model prediction stages were integrated into the pipeline to help make accurate predictions and determine the ROI for different campaigns. The workflow was implemented on Google Cloud Platform. This proprietary solution used quasi-experimentation to overcome the challenges related to absence of user-level data. It evaluated campaigns for different combinations of creatives, strategies, audience segments, zip codes, and product categories.

Download case study

Business Impact

The customized MTA solution transformed decision-making for marketing teams and delivered 11% ROI on marketing investment. While the previous approach provided campaign analysis only after 12 weeks, Sigmoid offered weekly insights that could be fed back into the dynamic system on the go. The solution was also scalable across various brands, delivering over 70% prediction accuracy on campaign performance.