Repository navigation
Migrate combined_video_engagements_counts_report to use dimensional layer (blocked by mart migration) #2112
Description
Epic: #2072
File: src/ol_dbt/models/reporting/combined_video_engagements_counts_report.sql
Problem: References marts__combined_video_engagements instead of afact_video_engagement / tfact_video_events. Blocked by migration of that mart.
Note: This issue is blocked by the migration of the upstream mart model(s) it references. It should be addressed after those mart migrations are complete.
Acceptance criteria:
- Model references only
dimensional/, mart, or reporting models (all of which use dimensional) - No
{{ ref('int__*') }}or{{ ref('stg__*') }}calls remain in this model or its upstream mart dependencies - Model compiles and passes dbt tests
Activity
blarghmatey commentedon Jul 6, 2026
2026-07 warehouse audit — additional finding (epic #2375)
Beyond the layer migration, the audit found a metric-semantics bug in the canonical video-count model:
combined_video_engagements_counts_report.sql:5-14computestotal_video_countascount(distinct video_title)from engagement events, while_reporting__models.yml:1155-1157describes it as "total number of distinct videos in this chapter". Videos never played by anyone are excluded, understating denominators in "watched X of Y" metrics.Fix: recompute totals from course structure (
dim_video/int__combined__course_videos), or correct the descriptions to match the events-derived reality. Apply to the canonical model only once the two duplicate video-count models are deleted (see the dead-model work under #2072).Tracked as witan task
tk-fix-video-total-count-semantics-events-derived-v-d026fb.