Novelty effect over four weeks
Shopee launched a new home-feed layout and the first week looked spectacular. A novelty effect is when users engage with something because it's *new*, not because it's *better* — the lift decays as the novelty wears off, so a test stopped after week one ships a change that does nothing.
From weekly_ab, return each week with lift — the treatment conversion rate minus the control conversion rate — rounded to 4 decimal places. Order by week.
Schema
| week | INT |
| variant | TEXT |
| users | INT |
| conversions | INT |
Example input
Stuck? Ask the coach for a nudge that won't give the answer away, or reveal the reference solution and a short walkthrough of why it works.