You observe that weekly engagement fell after a product change launched at the start of week 9. You have 16 weeks of engagement rate data for this year and the matching 16 weeks from last year. In weeks 1-8, this year averaged 0.401 engagement versus 0.399 last year; in weeks 9-16, this year averaged 0.356 versus 0.389 last year. The standard deviation of the weekly year-over-year differences is 0.006 in weeks 1-8 and 0.008 in weeks 9-16.
How would you test whether the observed drop is more consistent with normal seasonality or with an incremental effect from the product change?