feat(queries): add BOUNCE_THRESHOLD and enhance bounce calculation logic

Updated bounce calculations to account for visits with custom events
(`EVENT_TYPE.customEvent`). A visit counts as a bounce only when its
custom event count is below `BOUNCE_THRESHOLD`, so engaged single
pageview visits are no longer misreported.

Since #3614 was opened v3 moved `getInsights` into `getBreakdown` and
copied the old formula into the channel, pageview and session
breakdowns, so those are updated too.

src/queries/sql/events/getEventExpandedMetrics.ts

@@ -1,5 +1,5 @@
 import clickhouse from '@/lib/clickhouse';
-import { EVENT_TYPE, FILTER_COLUMNS, SESSION_COLUMNS } from '@/lib/constants';
+import { BOUNCE_THRESHOLD, EVENT_TYPE, FILTER_COLUMNS, SESSION_COLUMNS } from '@/lib/constants';
 import { CLICKHOUSE, PRISMA, runQuery } from '@/lib/db';
 import prisma from '@/lib/prisma';
 import type { QueryFilters } from '@/lib/types';
@@ -54,7 +54,7 @@ async function relationalQuery(
       sum(t.c) as "pageviews",
       count(distinct t.session_id) as "visitors",
       count(distinct t.visit_id) as "visits",
-      sum(case when t.c = 1 then 1 else 0 end) as "bounces",
+      sum(case when t.c = 1 and (t.events_count - t.c) < ${BOUNCE_THRESHOLD} then 1 else 0 end) as "bounces",
       sum(${getTimestampDiffSQL('t.min_time', 't.max_time')}) as "totaltime"
     from (
       select
@@ -63,10 +63,19 @@ async function relationalQuery(
         website_event.visit_id,
         count(*) as "c",
         min(website_event.created_at) as "min_time",
-        max(website_event.created_at) as "max_time"
+        max(website_event.created_at) as "max_time",
+        max((
+          select count(*)
+          from website_event we2
+          where we2.website_id = website_event.website_id
+            and we2.session_id = website_event.session_id
+            and we2.visit_id = website_event.visit_id
+            and we2.created_at between {{startDate}} and {{endDate}}
+            and we2.event_type = ${EVENT_TYPE.customEvent}
+        )) as "events_count"
       from website_event
       ${cohortQuery}
-      ${joinSessionQuery}  
+      ${joinSessionQuery}
       where website_event.website_id = {{websiteId::uuid}}
         and website_event.created_at between {{startDate}} and {{endDate}}
         ${filterQuery}
@@ -104,7 +113,7 @@ async function clickhouseQuery(
       sum(t.c) as "pageviews",
       uniq(t.session_id) as "visitors",
       uniq(t.visit_id) as "visits",
-      sum(if(t.c = 1, 1, 0)) as "bounces",
+      sumIf(1, t.c = 1 and (toInt64(ifNull(e.events_count, 0)) - toInt64(t.c)) < ${BOUNCE_THRESHOLD}) as "bounces",
       sum(max_time-min_time) as "totaltime"
     from (
       select
@@ -122,7 +131,15 @@ async function clickhouseQuery(
         ${filterQuery}
       group by name, session_id, visit_id
     ) as t
-    group by name 
+    left join (
+      select session_id, visit_id, toUInt32(count()) as events_count
+      from website_event
+      where website_id = {websiteId:UUID}
+        and created_at between {startDate:DateTime64} and {endDate:DateTime64}
+        and event_type = ${EVENT_TYPE.customEvent}
+      group by session_id, visit_id
+    ) as e using (session_id, visit_id)
+    group by name
     order by visitors desc, visits desc
     limit ${limit}
     offset ${offset}

src/queries/sql/getChannelExpandedMetrics.ts

@@ -1,6 +1,8 @@
 import clickhouse from '@/lib/clickhouse';
 import {
+  BOUNCE_THRESHOLD,
   EMAIL_DOMAINS,
+  EVENT_TYPE,
   LLM_DOMAINS,
   PAID_AD_PARAMS,
   SEARCH_DOMAINS,
@@ -120,20 +122,34 @@ async function relationalQuery(
           min(created_at) as min_time,
           max(created_at) as max_time
         from prefix
+        group by session_id, visit_id),
+
+      visit_events as (
+        select
+          session_id,
+          visit_id,
+          count(*) as events_count
+        from website_event
+        where website_id = {{websiteId::uuid}}
+          and created_at between {{startDate}} and {{endDate}}
+          and event_type = ${EVENT_TYPE.customEvent}
         group by session_id, visit_id)
-  
+
       select
         visit_channels.name,
         sum(visit_stats.c) as "pageviews",
         count(distinct visit_stats.session_id) as "visitors",
         count(distinct visit_stats.visit_id) as "visits",
-        sum(case when visit_stats.c = 1 then 1 else 0 end) as "bounces",
+        sum(case when visit_stats.c = 1 and coalesce(visit_events.events_count, 0) < ${BOUNCE_THRESHOLD} then 1 else 0 end) as "bounces",
         sum(${getTimestampDiffSQL('visit_stats.min_time', 'visit_stats.max_time')}) as "totaltime"
       from visit_stats
       join visit_channels
         on visit_channels.session_id = visit_stats.session_id
         and visit_channels.visit_id = visit_stats.visit_id
-      group by visit_channels.name 
+      left join visit_events
+        on visit_events.session_id = visit_stats.session_id
+        and visit_events.visit_id = visit_stats.visit_id
+      group by visit_channels.name
       order by visitors desc, visits desc
       `,
     queryParams,
@@ -158,7 +174,7 @@ async function clickhouseQuery(
       sum(t.c) as "pageviews",
       uniq(t.session_id) as "visitors",
       uniq(t.visit_id) as "visits",
-      sum(if(t.c = 1, 1, 0)) as "bounces",
+      sumIf(1, t.c = 1 and ifNull(e.events_count, 0) < ${BOUNCE_THRESHOLD}) as "bounces",
       sum(max_time-min_time) as "totaltime"
     from (
       select
@@ -213,7 +229,15 @@ async function clickhouseQuery(
       )
       group by session_id, visit_id
     ) as t
-    group by name 
+    left join (
+      select session_id, visit_id, toUInt32(count()) as events_count
+      from website_event
+      where website_id = {websiteId:UUID}
+        and created_at between {startDate:DateTime64} and {endDate:DateTime64}
+        and event_type = ${EVENT_TYPE.customEvent}
+      group by session_id, visit_id
+    ) as e using (session_id, visit_id)
+    group by name
     order by visitors desc, visits desc;
     `,
     queryParams,

src/queries/sql/getWebsiteStats.ts

@@ -1,5 +1,5 @@
 import clickhouse from '@/lib/clickhouse';
-import { EVENT_COLUMNS } from '@/lib/constants';
+import { BOUNCE_THRESHOLD, EVENT_COLUMNS, EVENT_TYPE } from '@/lib/constants';
 import { CLICKHOUSE, PRISMA, runQuery } from '@/lib/db';
 import prisma from '@/lib/prisma';
 import type { QueryFilters } from '@/lib/types';
@@ -35,7 +35,9 @@ async function relationalQuery(
     });
 
   const { excludeBounce } = filters;
-  const bounceQuery = excludeBounce ? '0' : 'coalesce(sum(case when t.c = 1 then 1 else 0 end), 0)';
+  const bounceQuery = excludeBounce
+    ? '0'
+    : `coalesce(sum(case when t.c = 1 and t.events_count < ${BOUNCE_THRESHOLD} then 1 else 0 end), 0)`;
 
   return rawQuery(
     `
@@ -51,11 +53,20 @@ async function relationalQuery(
         website_event.visit_id,
         count(*) as "c",
         min(website_event.created_at) as "min_time",
-        max(website_event.created_at) as "max_time"
+        max(website_event.created_at) as "max_time",
+        max((
+          select count(*)
+          from website_event we2
+          where we2.website_id = website_event.website_id
+            and we2.session_id = website_event.session_id
+            and we2.visit_id = website_event.visit_id
+            and we2.created_at between {{startDate}} and {{endDate}}
+            and we2.event_type = ${EVENT_TYPE.customEvent}
+        )) as "events_count"
       from website_event
       ${cohortQuery}
       ${excludeBounceQuery}
-      ${joinSessionQuery}  
+      ${joinSessionQuery}
       where website_event.website_id = {{websiteId::uuid}}
         and website_event.created_at between {{startDate}} and {{endDate}}
         and website_event.event_type NOT IN (2, 5)
@@ -81,7 +92,9 @@ async function clickhouseQuery(
 
   let sql = '';
   const { excludeBounce } = filters;
-  const bounceQuery = excludeBounce ? '0' : 'sumIf(1, t.c = 1)';
+  const bounceQuery = excludeBounce
+    ? '0'
+    : `sumIf(1, t.c = 1 and ifNull(e.events_count, 0) < ${BOUNCE_THRESHOLD})`;
 
   if (EVENT_COLUMNS.some(item => Object.keys(filters).includes(item))) {
     sql = `
@@ -106,7 +119,15 @@ async function clickhouseQuery(
         and event_type NOT IN (2, 5)
         ${filterQuery}
       group by session_id, visit_id
-    ) as t;
+    ) as t
+    left join (
+      select session_id, visit_id, toUInt32(count()) as events_count
+      from website_event
+      where website_id = {websiteId:UUID}
+        and created_at between {startDate:DateTime64} and {endDate:DateTime64}
+        and event_type = ${EVENT_TYPE.customEvent}
+      group by session_id, visit_id
+    ) as e using (session_id, visit_id);
     `;
   } else {
     sql = `
@@ -130,7 +151,14 @@ async function clickhouseQuery(
       and event_type NOT IN (2, 5)
       ${filterQuery}
       group by session_id, visit_id
-    ) as t;
+    ) as t
+    left join (
+      select session_id, visit_id, toUInt32(sumIf(views, event_type = ${EVENT_TYPE.customEvent})) as events_count
+      from website_event_stats_hourly
+      where website_id = {websiteId:UUID}
+        and created_at between {startDate:DateTime64} and {endDate:DateTime64}
+      group by session_id, visit_id
+    ) as e using (session_id, visit_id);
     `;
   }
 

src/queries/sql/pageviews/getPageviewExpandedMetrics.ts

@@ -1,5 +1,5 @@
 import clickhouse from '@/lib/clickhouse';
-import { FILTER_COLUMNS, GROUPED_DOMAINS, SESSION_COLUMNS } from '@/lib/constants';
+import { BOUNCE_THRESHOLD, EVENT_TYPE, FILTER_COLUMNS, GROUPED_DOMAINS, SESSION_COLUMNS } from '@/lib/constants';
 import { CLICKHOUSE, PRISMA, runQuery } from '@/lib/db';
 import prisma from '@/lib/prisma';
 import type { QueryFilters } from '@/lib/types';
@@ -96,7 +96,7 @@ async function relationalQuery(
       sum(t.c) as "pageviews",
       count(distinct t.session_id) as "visitors",
       count(distinct t.visit_id) as "visits",
-      sum(case when t.c = 1 then 1 else 0 end) as "bounces",
+      sum(case when t.c = 1 and t.events_count < ${BOUNCE_THRESHOLD} then 1 else 0 end) as "bounces",
       sum(${getTimestampDiffSQL('t.min_time', 't.max_time')}) as "totaltime"
     from (
       select
@@ -105,12 +105,21 @@ async function relationalQuery(
         website_event.visit_id,
         count(*) as "c",
         min(website_event.created_at) as "min_time",
-        max(website_event.created_at) as "max_time"
+        max(website_event.created_at) as "max_time",
+        max((
+          select count(*)
+          from website_event we2
+          where we2.website_id = website_event.website_id
+            and we2.session_id = website_event.session_id
+            and we2.visit_id = website_event.visit_id
+            and we2.created_at between {{startDate}} and {{endDate}}
+            and we2.event_type = ${EVENT_TYPE.customEvent}
+        )) as "events_count"
       from website_event
       ${cohortQuery}
       ${excludeBounceQuery}
-      ${joinSessionQuery} 
-      ${entryExitQuery} 
+      ${joinSessionQuery}
+      ${entryExitQuery}
       where website_event.website_id = {{websiteId::uuid}}
       and website_event.created_at between {{startDate}} and {{endDate}}
       and website_event.event_type NOT IN (2, 5)
@@ -188,7 +197,7 @@ async function clickhouseQuery(
       sum(t.c) as "pageviews",
       uniq(t.session_id) as "visitors",
       uniq(t.visit_id) as "visits",
-      sum(if(t.c = 1, 1, 0)) as "bounces",
+      sumIf(1, t.c = 1 and ifNull(e.events_count, 0) < ${BOUNCE_THRESHOLD}) as "bounces",
       sum(max_time-min_time) as "totaltime"
     from (
       select
@@ -211,6 +220,14 @@ async function clickhouseQuery(
         ${filterQuery}
       group by name, session_id, visit_id
     ) as t
+    left join (
+      select session_id, visit_id, toUInt32(count()) as events_count
+      from website_event
+      where website_id = {websiteId:UUID}
+        and created_at between {startDate:DateTime64} and {endDate:DateTime64}
+        and event_type = ${EVENT_TYPE.customEvent}
+      group by session_id, visit_id
+    ) as e using (session_id, visit_id)
     group by name
     order by visitors desc, visits desc
     limit ${limit}

src/queries/sql/reports/getBreakdown.ts

@@ -1,5 +1,5 @@
 import clickhouse from '@/lib/clickhouse';
-import { EVENT_TYPE, FILTER_COLUMNS, SESSION_COLUMNS } from '@/lib/constants';
+import { BOUNCE_THRESHOLD, EVENT_TYPE, FILTER_COLUMNS, SESSION_COLUMNS } from '@/lib/constants';
 import { CLICKHOUSE, PRISMA, runQuery } from '@/lib/db';
 import prisma from '@/lib/prisma';
 import type { QueryFilters } from '@/lib/types';
@@ -50,7 +50,7 @@ async function relationalQuery(
       sum(t.c) as "views",
       count(distinct t.session_id) as "visitors",
       count(distinct t.visit_id) as "visits",
-      sum(case when t.c = 1 then 1 else 0 end) as "bounces",
+      sum(case when t.c = 1 and t.events_count < ${BOUNCE_THRESHOLD} then 1 else 0 end) as "bounces",
       sum(${getTimestampDiffSQL('t.min_time', 't.max_time')}) as "totaltime",
       ${parseFieldsByName(fields)}
     from (
@@ -60,14 +60,23 @@ async function relationalQuery(
         website_event.visit_id,
         count(*) as "c",
         min(website_event.created_at) as "min_time",
-        max(website_event.created_at) as "max_time"
+        max(website_event.created_at) as "max_time",
+        max((
+          select count(*)
+          from website_event we2
+          where we2.website_id = website_event.website_id
+            and we2.session_id = website_event.session_id
+            and we2.visit_id = website_event.visit_id
+            and we2.created_at between {{startDate}} and {{endDate}}
+            and we2.event_type = ${EVENT_TYPE.customEvent}
+        )) as "events_count"
       from website_event
-      ${cohortQuery}  
+      ${cohortQuery}
       ${joinSessionQuery}
       where website_event.website_id = {{websiteId::uuid}}
         and website_event.created_at between {{startDate}} and {{endDate}}
         ${filterQuery}
-      group by ${parseFieldsByName(fields)}, 
+      group by ${parseFieldsByName(fields)},
         website_event.session_id, website_event.visit_id
     ) as t
     group by ${parseFieldsByName(fields)}
@@ -99,7 +108,7 @@ async function clickhouseQuery(
       sum(t.c) as "views",
       count(distinct t.session_id) as "visitors",
       count(distinct t.visit_id) as "visits",
-      sum(if(t.c = 1, 1, 0)) as "bounces",
+      sumIf(1, t.c = 1 and ifNull(e.events_count, 0) < ${BOUNCE_THRESHOLD}) as "bounces",
       sum(max_time-min_time) as "totaltime",
       ${parseFieldsByName(fields)}
     from (
@@ -115,9 +124,17 @@ async function clickhouseQuery(
       where website_id = {websiteId:UUID}
         and created_at between {startDate:DateTime64} and {endDate:DateTime64}
         ${filterQuery}
-      group by ${parseFieldsByName(fields)}, 
+      group by ${parseFieldsByName(fields)},
         session_id, visit_id
     ) as t
+    left join (
+      select session_id, visit_id, toUInt32(count()) as events_count
+      from website_event
+      where website_id = {websiteId:UUID}
+        and created_at between {startDate:DateTime64} and {endDate:DateTime64}
+        and event_type = ${EVENT_TYPE.customEvent}
+      group by session_id, visit_id
+    ) as e using (session_id, visit_id)
     group by ${parseFieldsByName(fields)}
     order by 2 desc, 1 desc
     limit 500

src/queries/sql/sessions/getSessionExpandedMetrics.ts

@@ -1,5 +1,5 @@
 import clickhouse from '@/lib/clickhouse';
-import { FILTER_COLUMNS, SESSION_COLUMNS } from '@/lib/constants';
+import { BOUNCE_THRESHOLD, EVENT_TYPE, FILTER_COLUMNS, SESSION_COLUMNS } from '@/lib/constants';
 import { CLICKHOUSE, PRISMA, runQuery } from '@/lib/db';
 import prisma from '@/lib/prisma';
 import type { QueryFilters } from '@/lib/types';
@@ -62,7 +62,7 @@ async function relationalQuery(
       sum(t.c) as "pageviews",
       count(distinct t.session_id) as "visitors",
       count(distinct t.visit_id) as "visits",
-      sum(case when t.c = 1 then 1 else 0 end) as "bounces",
+      sum(case when t.c = 1 and t.events_count < ${BOUNCE_THRESHOLD} then 1 else 0 end) as "bounces",
       sum(${getTimestampDiffSQL('t.min_time', 't.max_time')}) as "totaltime"
     from (
       select
@@ -72,11 +72,20 @@ async function relationalQuery(
         website_event.visit_id,
         count(*) as "c",
         min(website_event.created_at) as "min_time",
-        max(website_event.created_at) as "max_time"
+        max(website_event.created_at) as "max_time",
+        max((
+          select count(*)
+          from website_event we2
+          where we2.website_id = website_event.website_id
+            and we2.session_id = website_event.session_id
+            and we2.visit_id = website_event.visit_id
+            and we2.created_at between {{startDate}} and {{endDate}}
+            and we2.event_type = ${EVENT_TYPE.customEvent}
+        )) as "events_count"
       from website_event
       ${cohortQuery}
       ${excludeBounceQuery}
-      ${joinSessionQuery}  
+      ${joinSessionQuery}
       where website_event.website_id = {{websiteId::uuid}}
         and website_event.created_at between {{startDate}} and {{endDate}}
         and website_event.event_type NOT IN (2, 5)
@@ -122,7 +131,7 @@ async function clickhouseQuery(
       sum(t.c) as "pageviews",
       uniq(t.session_id) as "visitors",
       uniq(t.visit_id) as "visits",
-      sum(if(t.c = 1, 1, 0)) as "bounces",
+      sumIf(1, t.c = 1 and ifNull(e.events_count, 0) < ${BOUNCE_THRESHOLD}) as "bounces",
       sum(max_time-min_time) as "totaltime"
     from (
       select
@@ -144,7 +153,15 @@ async function clickhouseQuery(
       group by name, session_id, visit_id
       ${includeCountry ? ', country' : ''}
     ) as t
-    group by name 
+    left join (
+      select session_id, visit_id, toUInt32(count()) as events_count
+      from website_event
+      where website_id = {websiteId:UUID}
+        and created_at between {startDate:DateTime64} and {endDate:DateTime64}
+        and event_type = ${EVENT_TYPE.customEvent}
+      group by session_id, visit_id
+    ) as e using (session_id, visit_id)
+    group by name
     ${includeCountry ? ', country' : ''}
     order by visitors desc, visits desc
     limit ${limit}