$5 million in annual cost savings with algorithmic retargeting


  • Using rules to build advertising segments made Lenovo’s audiences heuristic, subjective, and difficult to scale
  • Conflicting rules created uncertainty about which audience a person would be included in
  • Each customer has a different intent and Lenovo’s first-party behavioral data has the signals they need to target more effectively
  • Building rules was a very manual and laborious process, and it was difficult to predict audience sizes


To scale audience sizes and retarget web visitors more effectively, Lenovo tested their rules-based retargeting audiences against Syntasa’s algorithmic retargeting model for a period of three months, using data for US visitors on They evaluated hundreds of variables and coefficients, created confidence intervals, and ran an experiment to compare the performance of rules-based and algorithmic audiences.

Use Cases

Algorithmic Retargeting
Algorithmic Attribution


  • Cost savings of $5 million in display advertising annually
  • Can now predict the purchasing likelihood for web visitors
  • Simple to adjust audience sizes by creating high-value and low-value segments
  • Scalable propensity algorithm can be leveraged across global marketing programs

Slider image

“Syntasa has been really invaluable in speeding up our time to value by architecting our Adobe Analytics data and productionizing data science and machine learning modeling at scale, and in such a way that we can confidently pass that into production systems to drive the user experience.”

Slider image

“Syntasa’s platform puts the business critical problem at the center. The pipeline, scheduling, and predictive capabilities are fit for real marketing use cases, but are flexible enough to prep difficult data and apply special transformations. With Syntasa, I am able to run experiments and focus on the insights, without spending hours coding a solution.”