12.10 Chapter Review
12.10.1 Chapter summary
An effective analytical story is a selective but honest sequence that connects context, evidence, interpretation, and action. It begins with the audience and decision, uses a clear central message and evidence hierarchy, and maintains cohesion across text, visuals, recommendations, and limitations.
12.10.2 Common mistakes
- presenting results in the order produced;
- using a topic as the central message;
- including every graph from the EDA;
- changing key terms across sections;
- introducing recommendations not prepared by the findings;
- hiding uncertainty because it complicates the narrative;
- building slides before agreeing on the story; and
- using a memorable persona to disguise unsupported assumptions.
12.10.4 Exercises
Exercise 1. Convert the topic “NVRW membership cancellations” into a central message using this finding: cancellation is highest among new members with fewer than three visits in their first sixty days, but the data do not record reasons for leaving.
One Possible Answer
Low early engagement identifies a practical group for a retention pilot, although the available data cannot establish why individual members leave. This message states the pattern, its possible use, and the main limitation without claiming causation.
Exercise 2. Create a six-panel storyboard for the NVRW retention question. Each panel should contain a message sentence, the evidence needed, and the implication for the next panel.
One Possible Answer
A coherent sequence is: (1) retention is the decision priority; (2) retention differs meaningfully across member groups; (3) low early engagement identifies a practical risk group; (4) predictive evidence helps prioritize limited contact capacity; (5) observational data cannot establish the effect of an intervention; and (6) NVRW should run and evaluate a defined pilot. Each panel should use only the evidence required to support its message.
Exercise 3. Write two versions of the same finding: one for an analyst reviewing model performance and one for an executive deciding whether to authorize a pilot.
One Possible Answer
Analyst: The model improves lift among the 150 highest-risk eligible members, but calibration and subgroup stability require continued monitoring. Executive: The analysis can identify a manageable group with elevated non-renewal risk, supporting a limited pilot, but the pilot must measure whether contact actually changes retention.
Exercise 4. Review the NVRW synthesis appendix and identify one table or result that belongs in a technical appendix rather than the main narrative. Defend your choice.
One Possible Answer
A full model-diagnostics or validation table belongs in a technical appendix when the main audience needs only the baseline comparison, selected performance measures, and decision implication. Moving it preserves transparency for technical reviewers without interrupting the central story. A short main-text reference should explain why the method was considered adequate.