You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
"List short links and existing folders for the website's organization. Use to enumerate all links before referencing one or choosing where a new link should go.",
969
+
"List the newest short links and existing folders for the website's organization. The count covers the full catalog; use search_links to find a specific older link.",
- Geographic data (country, region, city) exists only on analytics.events, NOT on web_vitals_spans or error_spans. Join via session_id if needed.
192
193
- All timestamps are in UTC.
193
194
- analytics.custom_events.properties contains JSON strings — use JSONExtractString(properties, 'key') to parse.
194
-
- Identity: \`anonymous_id\` is a per-device id; \`profile_id\` is the customer-assigned user id ('' when anonymous, on analytics.events, analytics.custom_events, and analytics.revenue). To count people, dedupe identified users across devices with \`uniq(${EVENTS_VISITOR_KEY})\` (on custom_events/revenue wrap the fallback: \`if(profile_id != '', profile_id, ifNull(anonymous_id, ''))\` since anonymous_id is Nullable there). To count only identified users: \`uniqIf(profile_id, profile_id != '')\`. error_spans/web_vitals_spans/outgoing_links have no profile_id — resolve an identified user's rows there via their anonymous_ids from analytics.events.
195
+
- Identity: \`anonymous_id\` is a per-device id; \`profile_id\` is the customer-assigned user id ('' when anonymous, on analytics.events, analytics.custom_events, and analytics.revenue). To count people, dedupe identified users across devices with \`uniq(${EVENTS_VISITOR_KEY})\`; on custom_events/revenue use \`uniq(${CUSTOM_EVENTS_VISITOR_KEY})\` because anonymous_id is Nullable and rows missing both identifiers must not count as a person. To count only identified users: \`uniqIf(profile_id, profile_id != '')\`. error_spans/web_vitals_spans/outgoing_links have no profile_id — resolve an identified user's rows there via their anonymous_ids from analytics.events.
195
196
196
197
## Aggregate function preferences
197
198
- Percentiles: use \`quantileTDigest(p)(col)\` for p50/p75/p95/p99. Plain \`quantile(p)\` uses reservoir sampling and is noisy at the tails (~10% error at p99). \`quantileTDigest\` is within 0.1% of exact at the same memory cost.
0 commit comments