Sample Ratio Mismatch (SRM) Checker

Validate your experiment’s traffic split with a chi-square test. If SRM is present, pause and fix the cause before interpreting uplift.

1
Enter Traffic Counts

Control visitors

Variant visitors

Expected split (Control %)

%Use 50 for a standard A/B split

Verdict

No SRM detected (traffic split looks plausible).

p-value thresholds: p < 0.01 (strong SRM), p < 0.05 (warning)

Observed control

50.51%

Observed variant

49.49%

Chi-square (df=1)

2.020

p-value

0.1552

Split looks plausible

No SRM detected at conventional thresholds.

p-value + confidence interval

Plan your experiment

How long to run