On the placement expansion, from watching a campaign through it: average cost per tap did fall, and so did the value of the top slot, which used to be the only one.
Two practical effects. Your old top-position benchmarks are no longer comparable, so a lower cost per tap is not necessarily an improvement in outcomes. And because enrolment was automatic with no placement control, the change arrived in your numbers without appearing in your change log.
If you are comparing performance across March, the break in the series is that, not anything you did.