Choose an Attribution Window for a Re-Engagement Campaign

A product team runs a re-engagement email campaign targeting dormant users. They need to decide the attribution window: how long after an email is sent a subsequent app open is credited to that email versus treated as unrelated organic return.

1. Explain the trade-off between a short window (e.g. 2 hours) and a long window (e.g. 14 days) for this specific use case.
2. Describe an experiment you could run to empirically estimate the right window, rather than picking one arbitrarily.
3. Once the attribution window is chosen, explain how it should concretely change how the training data for a "should we send this re-engagement email" model gets constructed, referencing the maturity/censoring discipline from this subject.