data layer

data-layer/infra/clickhouse/retention_cohort.sql.tmpl

@@ -0,0 +1,41 @@
+-- Retention Cohort -- of users whose first `initial_event` lands on day D,
+-- what share triggered `return_event` on day D+k for k in 1..Periods.
+--
+-- Required parameters (clickhouse.Named):
+--   workspace_id  : String
+--   from          : DateTime64(3,'UTC')
+--   to            : DateTime64(3,'UTC')
+--   initial_event : String
+--   return_event  : String
+--
+-- Template inputs:
+--   .Outer : []{ RIndex int; OffsetDay int; Last bool }
+--     One entry per follow-up day. RIndex is the position in the retention()
+--     output array; OffsetDay is the day delta from the cohort day.
+SELECT
+    cohort_day,
+    countIf(arrayElement(r, 1)) AS cohort_size,
+{{- range $p := .Outer }}
+    countIf(arrayElement(r, {{ $p.RIndex }})) AS retained_d{{ $p.OffsetDay }}{{ if not $p.Last }},{{ end }}
+{{- end }}
+FROM (
+    SELECT
+        user_id,
+        toDate(min(if(event = {initial_event:String}, timestamp, NULL))) AS cohort_day,
+        retention(
+            event = {initial_event:String} AND toDate(timestamp) = cohort_day,
+{{- range $p := .Outer }}
+            event = {return_event:String} AND toDate(timestamp) = addDays(cohort_day, {{ $p.OffsetDay }}){{ if not $p.Last }},{{ end }}
+{{- end }}
+        ) AS r
+    FROM events_track
+    WHERE workspace_id = {workspace_id:String}
+      AND received_at >= {from:DateTime64(3,'UTC')}
+      AND received_at <  {to:DateTime64(3,'UTC')}
+      AND user_id != ''
+      AND event IN ({initial_event:String}, {return_event:String})
+    GROUP BY user_id
+    HAVING cohort_day IS NOT NULL
+)
+GROUP BY cohort_day
+ORDER BY cohort_day