Airbnb's experimentation team is deciding whether to ship the checkout redesign — that means comparing conversion rates between the two variants. Using the ab_test_assignments table, return variant, total_users, conversions, and conversion_rate_pct (rounded to 2 decimal places), ordered by variant.
ab_test_assignments
| column | type |
|---|---|
| id | INTEGER |
| user_id | INTEGER |
| variant | TEXT |
| converted | BOOLEAN |
ab_test_assignments
| id | user_id | variant | converted |
|---|---|---|---|
| 1 | 1 | control | false |
| 2 | 2 | control | true |
| 3 | 3 | control | false |
| 4 | 4 | control | true |
| 5 | 5 | control | false |
| 6 | 6 | treatment | true |
| 7 | 7 | treatment | true |
| 8 | 8 | treatment | false |
| 9 | 9 | treatment | true |
| 10 | 10 | treatment | false |
| variant | total_users | conversions | conversion_rate_pct |
|---|---|---|---|
| control | 5 | 2 | 40.00 |
| treatment | 5 | 3 | 60.00 |
Control had 5 users with 2 conversions (rows 2 and 4), giving 40.00%. Treatment had 5 users with 3 conversions (rows 6, 7, and 9), giving 60.00%.