Mirela

The reason this survives everywhere is that the dashboard rewards it. Long onboarding raises completion and raises trial starts, and both of those are on somebody's chart. Refund rate and review sentiment are on nobody's chart, so the test looks like a clean win and ships.

Waiting a month is exactly the cost people will not pay, though. That is the whole reason the bad metric wins.

Careful with the natural experiment framing, because two platforms is also two audiences and two sets of pricing norms. People who pay up front for software self select. Some of that rating gap is the model and some of it is who is doing the buying, and this data cannot separate them.

That is a much better argument than the star numbers and I would lead with it. The complaint text is the finding, the rating gap is decoration.