Rate limits
Every endpoint allows 60 requests per minute. Over the limit the API answers429 Too Many Requests. Spread bulk work over time and cache stable lookups such as trackers, tags, segments and profiles.
Timeouts
A request that takes longer than 30 seconds stops and answers504 Gateway Timeout. This mostly happens on long date windows combined with page-view filters or score sorts. Narrow the window or the filters and try again, rather than retrying the same request.
Date windows
Tracker endpoints takestart_date and end_date as YYYY-MM-DD. Both default to today, so a call without dates describes today. A window can span at most 730 days, and an end_date before start_date is rejected with 422.
Two things are computed independently of the window you ask for:
- Engagement and the unified score always look at the 90 days up to
end_date, whateverstart_dateis. - Page-view rules in company filters (
url,page_title) and Intent Builder profiles are limited to windows of 365 days.
Page sizes
Response envelopes
Lists come back as{ "data": [...], "meta": {...} }.
Bounded lists (trackers, tags, segments, personas, analytics buckets, pages, contacts, opportunities) carry meta.returned only.
Paginated feeds (sessions, visitors, submissions) carry meta.total, meta.has_more and meta.returned. total is null unless you pass include_total=true, because counting roughly doubles the cost of the request; page with limit and offset where available and stop when has_more is false.
The companies endpoints add meta.segment_id, meta.unsupported_rules (segment rules the API could not translate) and meta.scoring (the profiles and weights the scores were computed with). See Pagination for iteration examples.