How the interviewer opens it
“Design a social media analytics platform for creators and publishers. It should report reach, engagement, and audience statistics for their content.”
The prompt is vague on purpose. The first minutes are for you. Reduce the prompt to the core features and agree on the numbers before you draw.
What to clarify first
- Metric definitions
- Report scope and audience
- Event source and identity
- Event scale
- Read and write mix
- Latency and freshness
- Accuracy and availability
- Late events and duplicate events
- Durability and retention
- Access and exclusions
The interviewer has the answers. In a SysRound round, Alex gives these answers, as a real interviewer does.
Scope of a strong answer
Core requirements
- Report impressions, reach, and engagement counts for a creator or content item.
- Report audience reach by country or age band for a creator.
- Support daily charts and totals across a selected date range.
- Correct reports when late events arrive without counting duplicate events.
Usually out of scope
- Collect data from external social networks.
- Build a social feed or recommendation service.
- Report follower growth.
- Support arbitrary audience filters.
What interviewers ask about
- Reach and audience totals
- The same viewer can appear on many dates, content items, and audience segments. The sum of daily reach values overstates range reach.
- Late events and duplicate events
- Retries and failures can repeat accepted events. A worker delay can hide accepted events after the deadline.
- A viral creator
- One creator can receive a large share of events and report requests. Many small content buckets also require substantial state.
- Historical report correction
- A correction must replace incorrect totals while new events continue. Expired raw records can make some historical errors impossible to correct.
Common mistakes
- The design adds daily reach values to calculate reach across a date range.
- The candidate treats event deduplication and distinct viewer counts as the same operation.
- The design expires duplicate IDs before the service closes the related event dates.
- The design uses receipt time for reports and places late events on the wrong date.
- The candidate trusts a broker guarantee without explaining state recovery and report writes.
- The candidate ignores the state cost of content buckets and audience segments.
The bar by level
- Senior (L5/E5)
- The candidate defines the late-event contract and finality rules. The candidate quantifies deduplication and sketch state. The design connects checkpoints to safe sink writes and handles hot creators.
- Staff (L6+/E6+)
- The candidate budgets state across content and audience dimensions. The design defines completeness during failures and supports versioned corrections. The candidate explains retention limits, recovery capacity, and accuracy costs.