What is sample ratio mismatch (SRM), and what should you do if you see it?
PICTURE THIS: 1, 2, 2, 8
Simple meaning
SRM is a statistically large deviation from the planned traffic split, detected with a goodness-of-fit test on assignment counts.
WHY — A/B Testing instead of guessing?
Why interviewers care about A/B Testing:
people who only read docs from people who shipped.
and tied to Data Science work.
Name the idea, why it exists, then one short example.
End with when you use it and one common pitfall.
STEPS — What happens step by step?
Before you speak the answer, walk the interviewer through these steps:
- 1SRM is a statistically
large deviation from the planned traffic split, detected with a goodness-of-fit test on assignment counts.
- 2It often means logging
bugs, redirected bots, or assignment that depends on the variant.
- 3Do not trust lift
estimates until the assignment pipeline is fixed
- 4analyzing only the remaining
users can still be biased.
- 5Common mistake
What juniors usually get wrong.
- 6Close
When you pick this over the alternative.
EXAMPLE — See it in action
Here's a short line you can speak, broken into clear beats:
Note: Adapt this scaffold to your own project — keep it under 60–90 seconds.
Key takeaway
SRM is a statistically large deviation from the planned traffic split, detected with a goodness-of-fit test on assignment counts. It often means logging bugs, redirected bots, or assignment that depends on the variant.