Aggregate Viewer Data

MEDKit aggregation APIs collect many viewer submissions without storing a custom object for every viewer in channel state.

Need	Submit	Read	Guide
Preserve arbitrary timestamped payloads	accumulate(id, data)	getAccumulateData(id, start)	Accumulation
Count matching text choices	rank(id, value)	getRankData(id)	Ranking
Collect numeric choices and statistics	vote(id, value)	getVoteData(id)	Voting

Accumulate viewer input

Any authorized viewer can submit JSON-serializable data. Reading the accumulated records is broadcaster-only.

await medkit.accumulate("round-feedback", {
  round: 4,
  reaction: "excited",
});

const oneMinuteAgo = Date.now() - 60_000;
const result = await medkit.getAccumulateData(
  "round-feedback",
  oneMinuteAgo,
);

for (const entry of result.data) {
  console.log(entry.observed, entry.data);
}

Use stable identifiers for a continuous stream and include a round or schema version in each payload. Validate payload size and shape before submission; the SDK's generic types do not enforce your application schema.

Choose the smallest result shape

• Use ranking when the useful result is a text value plus its score.
• Use voting when choices are numeric and you need count, mean, sum, or standard deviation.
• Use accumulation only when you need each submitted payload and its observation metadata.

Do not put aggregate results into state on every viewer submission. If viewers need a durable snapshot, update state at a bounded interval and use events only to prompt open clients to refresh.