9. And they don't clean data; they only build the best models
8. Not sure if your Marketing guys can appreciate what we do
7. We typically don't work with your IT folks
6. Do you maintain robust no-mail Controls?
5. We need atleast 50,000 observations
4. Our Model documentation is the best; typically 186 pages long
3. How are your demographic variables bucketed?
2. Our last response model had a 82% lift over random in the top 3 deciles
1. Unfortunately two of the most significant model variables were not available at implementation
The challenge I see is that most analytics-focused consultants get lost in technicals; failing to understand that in almost 100% of the cases, understanding the context, communicating the real-life impact of analytics to the business and ensuring that outcomes can be robustly implemented are much more important that getting that incremental 2% lift in model performance or using a slew of alternative statistical techniques. And that businesses rarely have the benefit of perfect information and foresight that most analytics-focused folks typically assume.