Sample size for a smaller effect

You planned an A/B test to detect a 2% relative lift. Now the team wants to detect a 1% lift with the same power and significance level. Roughly how does the required sample size change?