Results and Execution
Understand execution settings and the data behind results.
107. How is the longest transition identified?
Transition timing measures the gap from one accepted record to the next. Result inspection can highlight the largest gap.
For A at 10:00, B at 10:03, and C at 10:08, B → C is the longest transition at five minutes. A final step has no following transition. Transition details and aggregation differ by result view, so do not assume every Duration field represents this maximum.
108. Does displayed duration mean total time or one transition?
It depends on the field. Sequence duration is elapsed time from sequence start; transition duration is the gap to the next accepted step.
Check the label and view. A business-sequence duration is also different from query execution time. In incomplete results, timing generally ends at the last accepted milestone rather than the end of all subsequent skipped activity.
109. Why can result groups show the same event path?
The path text is not necessarily the complete grouping key. Groups can differ by completion status, matched query path, or other aggregation dimensions.
For Funnel’s pattern aggregation, the longest-transition position is also a grouping dimension. Inspect status and grouping columns rather than assuming matching labels imply duplicate data.
110. What if source data changes after a query completes?
Existing result tables are stored outputs; they are not live views that rematch against changed source data.
A newly submitted run can produce different results. Retention expiry can remove the previous output, so preserve any evidence needed for comparison before expiry.
111. Are failed and cancelled attempts included in cost tracking?
Cost registration tracks submitted processing attempts by their Snowflake query IDs, not only successful matches. Failed or cancelled work can therefore incur and report attributed credits when registration and Snowflake history are available.
Do not infer zero cost from a failed or cancelled status. Retried attempts can have separate cost records.
112. Does execution time include waiting in zinzu’s queue?
Execution time measures processing, rather than the full time from submission to completion. Time waiting in zinzu’s queue is separate.
Query execution time is also different from the duration of the actions in your source data.
113. Does refreshing results rerun the original query?
Refreshing or opening a results view reads stored result objects; it does not itself submit the sequence query again.
These reads can still use Snowflake compute. A refresh in the application is not a guarantee of zero cost.
114. What remains after result retention expires?
Expired query results and associated saved-query details are removed by cleanup processing. Your original source records remain unchanged.
Save any results or settings you need before expiry. Changing retention after deletion does not restore the removed results.