[Discussion] What's a proper integrity test for `zendesk__sla_policies`?

[fivetran-jamie]

Is there an existing feature request for this?

• I have searched the existing issues

Describe the Feature

Context

Currently, when we run the sla_count_match integrity data validation test using our internal dataset, we get a handful of failures, specifically that the zendesk__sla_policies end model is missing some SLA records present in the source ticket_field_history table.

Case 1

Each of these missing records was a requester_wait_time SLA in which the SLA was either:

1. Applied to a ticket after the ticket was closed/solved, or
2. Applied to the ticket but there were no impacted status (where ticket_historical_status.status in ('new', 'open', 'on-hold', 'hold')) after the SLA was applied.

In both cases, the package filters out this sort of SLA record because it was never actually active. It's not even like the sla_elapsed_time = 0, it's really just all null.

Case 2

For tickets which have multiple requester wait times and agent work times, we currently don't bring them all in as we only focus on one of the latest slas. Thus the ticket_field_history source will have a row count that differs between source and end models.

Call to Action

So let this serve as a discussion thread --

For Case 1, would folks want these null records to persist to the zendesk__sla_policies end model? Are they helpful at all, or should we continue excluding them?

For Case 2, should we be applying requester_wait_time

How would you implement this feature?

TBD

Describe alternatives you've considered

No response

Are you interested in contributing this feature?

• Yes.
• Yes, but I will need assistance.
• No.

Anything else?

No response

[fivetran-jamie]

Discussed this with our internal engineering analysts, and they agree with the current implementation, in which inactive SLAs are excluded from zendesk__sla_policies

I will leave this ticket open for discussion in case other folks feel differently 🗣️