! There's no one-size-fits-all approach that can be taken here.
! The best thing is to try and focus on the clear actions that are taken as a result of the investigation
! Then try to relate those to a money figure.
+ Evaluate the technical expertise of your stakeholder’s team.
+ Establish a baseline model, a counterfactual for what they would have been able to create independently.
+ e.g. “Without Data Science support, the Growth team would have been able to build at
+ most a regularised (ridge/LASSO) model for their use-case”
+ Create your own model and compare the performance of it with the baseline model.
+ Calculate impact based on how much the Data Science model outperforms the baseline.
+ e.g. “Data Science’s model was able to predict XXX better than the Growth team could,
- Run a causal experiment to measure the incrementality of a particular marketing channel.
- Experiment shows that channel is more or less incremental than was believed.
- Decision to spend more (Y = incremental revenue) or spend less (Y = cost savings) on that channel