Non-technical, but informative introduction to impact evaluation for assessing the effect of interventions (e.g. discounts) on business outcomes (e.g. sales) for decision support
Different evaluation designs: 1) experiments (A/B testing); 2) “instrumental variable” designs for fixing “broken” experiments; 3) “selection-on-observables” designs based on groups with and without intervention that are similar in observed characteristics; 4) “difference-in-differences” designs based on groups with comparable time trends of business outcomes; “regression discontinuity” designs based on indices (e.g. customer score) which determine the receipt of an intervention (e.g. fidelity card)
Machine learning-based impact evaluation for detecting and optimally targeting customer segments for which interventions are particularly effective (e.g. loyal customers)
Business cases and practical examples with real data using graphical interfaces in web applications or the no-code software “BigML” – no programming required!
