A messaging platform currently supports only one-to-one voice calls and is considering whether to launch group calling.
You have access to the same data sources used in the SQL exercise:
Table: call_events
-
call_id
BIGINT
-
event_time
TIMESTAMP
-
caller_user_id
BIGINT
-
recipient_user_id
BIGINT
-
country_code
STRING
-
Assume one row per completed call attempt
Table: daily_active_users
-
activity_date
DATE
-
user_id
BIGINT
-
country_code
STRING
The interviewer asks two product analytics questions:
-
How would you use these tables to determine whether users need a group calling feature?
-
If the company launches group calling, how would you determine whether the launch was successful?
Provide a structured answer that covers:
-
What user behaviors in the existing one-to-one call data could signal latent demand for group calls
-
How you would define candidate metrics, including tradeoffs between adoption, engagement, retention, and call quality
-
What success metrics and guardrail metrics you would use after launch
-
Likely confounders, selection bias, and interpretation risks
-
How you would design an experiment or quasi-experiment if a randomized launch is possible or not possible