ok

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

@@ -1,41 +1,43 @@
 -- 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):
+-- We compute the cohort day in a CTE first, then LEFT JOIN events_track and
+-- count distinct returners per (cohort_day, day_offset). Doing it this way
+-- avoids ClickHouse's "aggregate inside another aggregate" restriction that
+-- the older retention()-based form ran into.
+--
+-- Required parameters (clickhouse.Named, string-valued):
 --   workspace_id  : String
---   from          : DateTime64(3,'UTC')
+--   from          : DateTime64(3,'UTC')  (formatted as 'YYYY-MM-DD HH:MM:SS.mmm')
 --   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 (
+--   .Outer : []{ OffsetDay int; Last bool }
+WITH cohorts AS (
     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
+        toDate(min(timestamp)) AS cohort_day
     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})
+      AND event = {initial_event:String}
     GROUP BY user_id
-    HAVING cohort_day IS NOT NULL
 )
-GROUP BY cohort_day
-ORDER BY cohort_day
+SELECT
+    c.cohort_day                                                                                            AS cohort_day,
+    uniqExact(c.user_id)                                                                                    AS cohort_size,
+{{- range $p := .Outer }}
+    uniqExactIf(c.user_id, e.event = {return_event:String} AND toDate(e.timestamp) = addDays(c.cohort_day, {{ $p.OffsetDay }})) AS retained_d{{ $p.OffsetDay }}{{ if not $p.Last }},{{ end }}
+{{- end }}
+FROM cohorts AS c
+LEFT JOIN events_track AS e
+       ON e.workspace_id = {workspace_id:String}
+      AND e.user_id      = c.user_id
+      AND e.received_at  >= {from:DateTime64(3,'UTC')}
+      AND e.received_at  <  {to:DateTime64(3,'UTC')}
+GROUP BY c.cohort_day
+ORDER BY c.cohort_day