Design Creator Analytics Dashboard - System Design | WinJob

Problem Statement and Creator Analytics Context

Creators need trustworthy engagement metrics (views, reach, watch time, follower growth) with clear freshness tiers and auditability—similar to Instagram Insights, YouTube Studio, TikTok Analytics.

Key mechanisms

Speed layer serves "last hour" cards with explicit staleness.
Batch layer produces auditable daily finals and powers exports.
Cubes bound cardinality: creator × content × grain × day.
Governance versions metrics and enforces privacy thresholds.

Deep dive

Interviewers probe whether you can defend numbers and correctness, not just box diagrams. For Problem Statement and Creator Analytics Context, tie every component to a metric: ingest lag minutes, cube freshness hours, dashboard p95 milliseconds, reconciliation delta percent.

javaOne Dark Pro

1 public final class CreatorMetricEvent {
2   private final String creatorId;
3   private final String contentId;
4   private final String metricName;
5   private final long value;
6   private final long eventTimeMs;
7  
8   public String dedupeKey() {
9     return creatorId + ":" + contentId + ":" + metricName + ":" + eventTimeMs;
10   }
11 }

pythonOne Dark Pro

1 from dataclasses import dataclass
2  
3 @dataclass(frozen=True)
4 class CreatorMetricEvent:
5     creator_id: str
6     content_id: str
7     metric: str
8     value: int
9     event_time_ms: int
10  
11     def dedupe_key(self) -> str:
12         return f"{self.creator_id}:{self.content_id}:{self.metric}:{self.event_time_ms}"

typescriptOne Dark Pro

1 interface CreatorMetricEvent {
2   creatorId: string;
3   contentId: string;
4   metric: "view" | "like" | "share" | "watch_ms";
5   value: number;
6   eventTimeMs: number;
7 }
8  
9 export function dedupeKey(evt: CreatorMetricEvent): string {
10   return `${evt.creatorId}:${evt.contentId}:${evt.metric}:${evt.eventTimeMs}`;
11 }

Operational notes

Section 1 note 1: Partition facts by dt and creator_id before discussing Problem Statement and Creator Analytics Context.
Section 1 note 2: Pin metric_definition_version on every aggregate written in section 1.
Section 1 note 3: Publish cube completeness SLI after batch job for Problem Statement and Creator Analytics Context.
Section 1 note 4: Use idempotent event_id dedupe at ingestion for creator 1.
Section 1 note 5: Cache keys must include creator scope—never global keys in Problem Statement and Creator Analytics Context.
Section 1 note 6: Reconciliation delta target 0.1% between speed and batch for Problem Statement and Creator Analytics Context.
Section 1 note 7: Suppress demographics when reach <100 in Problem Statement and Creator Analytics Context.
Section 1 note 8: Webhook fan-out only for milestones—not every minute tick.
Section 1 note 9: Backfill replays single dt partition without rewriting history.
Section 1 note 10: GraphQL DataLoader batches per creator in Problem Statement and Creator Analytics Context.
Section 1 note 11: Export jobs async with signed URLs expiring in 15 minutes.
Section 1 note 12: Deletion propagates to lake and cubes within 30 days.
Section 1 note 13: Staleness badge when minute layer lags >2 minutes.
Section 1 note 14: Cost guard aborts OLAP queries scanning >50GB per creator.

Interview checkpoint

Explain one failure mode for Problem Statement and Creator Analytics Context—for example duplicate events inflating views—and how dedupe plus reconciliation fixes it without silent data loss.

Key Highlights

•Lambda architecture: speed freshness + batch accuracy
•Versioned metric definitions with reconciliation SLIs
•Creator-scoped isolation across cache and OLAP

1	public final class CreatorMetricEvent {
2	private final String creatorId;
3	private final String contentId;
4	private final String metricName;
5	private final long value;
6	private final long eventTimeMs;
7
8	public String dedupeKey() {
9	return creatorId + ":" + contentId + ":" + metricName + ":" + eventTimeMs;
10	}
11	}

1	from dataclasses import dataclass
2
3	@dataclass(frozen=True)
4	class CreatorMetricEvent:
5	creator_id: str
6	content_id: str
7	metric: str
8	value: int
9	event_time_ms: int
10
11	def dedupe_key(self) -> str:
12	return f"{self.creator_id}:{self.content_id}:{self.metric}:{self.event_time_ms}"