Count feedback records

Filter by created_at >= created_since (ISO 8601, inclusive). created_at is when Hub stored the record; collected_at (see since) is when the feedback was given. They diverge on a historical re-import, so this is the parameter for "what did this import bring in". Must be between 1970-01-01 and 2080-12-31.

Filter by created_at <= created_until (ISO 8601, inclusive). See created_since for how created_at differs from collected_at. Must be between 1970-01-01 and 2080-12-31.

Filter by updated_at >= updated_since (ISO 8601, inclusive). updated_at is stamped on every write to a record — create, update, and every enrichment result (translation, sentiment, emotions) — so this selects records created OR changed since a point: the parameter for incremental extraction. Pair it with sort=updated_at&order=asc. For the next run, use the time the previous run started (its first response's Date header) minus an overlap, not the largest updated_at received. The overlap must outlast the longest write to a record; Hub does not bound that, so a few minutes is a default rather than a guarantee. Permanently deleted records are never returned: periodically list every record, upsert what it returns (which also repairs any change a slow write slipped past the overlap), and remove only local records that were not returned and were created before that listing started (its first response's Date header) minus the overlap. See "Incremental extraction" in the filtering and sorting guide.

Filter by updated_at <= updated_until (ISO 8601, inclusive). See updated_since for what updates updated_at.

Filter by value_number >= value_number_min (inclusive) — e.g. NPS promoters with value_number_min=9. Records whose value_number is NULL (every non-numeric answer) are excluded, so this is never a no-op filter.

Filter by value_number <= value_number_max (inclusive). Paired with value_number_min it selects a band: 9..10 is the NPS promoters, 0..6 the detractors. Records whose value_number is NULL are excluded. Supplying a max below the min is a 400, not an empty result.

Filter by value_date >= value_date_min (ISO 8601, inclusive) — bounds the answer to a date question, not when the feedback was collected. Records whose value_date is NULL are excluded.

Filter by value_date <= value_date_max (ISO 8601, inclusive). Records whose value_date is NULL are excluded.

Filter by sentiment_score >= sentiment_score_min (inclusive). The score is continuous where the label is bucketed, so this is the parameter for "the most negative feedback". Records that have not been enriched are excluded.

Filter by sentiment_score <= sentiment_score_max (inclusive). Records that have not been enriched are excluded.

Filter on whether sentiment enrichment has produced a label: true selects records where sentiment IS NOT NULL, false selects those where it IS NULL. Omit for no constraint.

Filter on whether the record carries emotion labels: true selects records where emotions IS NOT NULL, false selects those where it IS NULL. Note that false covers both "not yet classified" and "classified, no emotion detected" — the two are indistinguishable here. Omit for no constraint.

Filter on whether the record has been translated: true selects records where translation_lang_key IS NOT NULL, false selects those where it IS NULL. Omit for no constraint.