Data Sharing SQL Calculations
Discover SQL snippets for crafting common queries to fetch data from JWP Data Sharing
The following tables provide SQL snippet examples that can help with creating common queries for retrieving data from JWP Data Sharing.
Metric Calculations Reference
The following tables provide SQL calculations for common, useful metrics.
Ads
Metric | SQL Query |
---|---|
Ad Clicks | SUM(AD_CLICKS) |
Ad Completes | SUM(AD_COMPLETES) |
Ad Skips | SUM(AD_SKIPS) |
Engagement
Metric | SQL Query |
---|---|
25% Completes | SUM(CASE WHEN NVL(WATCHED_PCT, 0) >= 25 THEN 1 ELSE 0 END) |
50% Completes | SUM(CASE WHEN NVL(WATCHED_PCT, 0) >= 50 THEN 1 ELSE 0 END) |
75% Completes | SUM(CASE WHEN NVL(WATCHED_PCT, 0) >= 75 THEN 1 ELSE 0 END) |
Complete Rate | SUM(CASE WHEN COMPLETES > 0 THEN 1 ELSE 0 END) / SUM(CASE WHEN PLAYS > 0 THEN 1 ELSE 0 END) as complete_rate |
Video Seeks | SUM(VIDEO_SEEKS) |
Video Shares | SUM(VIDEO_SHARES) |
Watched Duration Viewable | SUM(WATCHED_DURATION_VIEWABLE) |
Watched Percentage | SUM(watched_duration_viewable) / SUM(watched_duration) AS percent_viewable_watch_time |
Performance
Metric | SQL Query |
---|---|
Ad Impressions | SUM(AD_IMPRESSIONS) |
Ad Viewability Rate | SUM(AD_VIEWABLE_IMPRESSIONS) / SUM(AD_IMPRESSIONS) |
Ad Viewable Impressions | SUM(AD_VIEWABLE_IMPRESSIONS) |
Completes | SUM(CASE WHEN COMPLETES > 0 THEN 1 ELSE 0 END) |
Count Playback Stalls | SUM(COUNT_PLAYBACK_STALLS) |
Embeds | SUM(CASE WHEN EMBEDS > 0 THEN 1 ELSE 0 END) or COUNT(DISTINCT EMBED_ID) |
First Frame | AVG(FIRST_FRAME) |
Play Attempts | SUM(CASE WHEN PLAY_ATTEMPTS > 0 THEN 1 ELSE 0 END) |
Plays | SUM(CASE WHEN PLAYS > 0 THEN 1 ELSE 0 END) |
Time Watched | SUM(WATCHED_DURATION) |
Viewers
Metric | SQL Query |
---|---|
Ad Impressions per Viewer | SUM(AD_IMPRESSIONS) / count(distinct viewer_id) as ads_per_viewer |
Plays per Viewer | SUM(CASE WHEN PLAYS > 0 THEN 1 ELSE 0 END) / count(distinct viewer_id) as plays_per_viewer |
Time Watched per Viewer | sum(watched_duration) / count(distinct viewer_id) as time_watched_per_viewer |
Unique Viewers | COUNT(DISTINCT VIEWER_ID) |
Dimension Calculations Reference
The following tables provide SQL calculations for common, useful dimensions.
_Content
Dimension | SQL Query |
---|---|
Ad Client | case AD_CLIENT when 0 then 'VAST' when 1 then 'Google IMA' when 2 then 'Freewheel' when 3 then 'DAI' ELSE 'Unknown' END as AD_CLIENT, |
Feed ID | WHERE FEED_ID = '' , GROUP BY FEED_ID |
Is Autostart | case is_autostart when 0 then 'Click Start' when 1 then 'Autostart' when 2 then 'Autostart on viewable' END as start_setting |
Is Click to Play | WHERE is_click_to_play = 1 |
Media Title | GROUP BY MEDIA_TITLE |
Media Type | WHERE MEDIA_TYPE = '' or GROUP BY MEDIA_TYPE |
Number of Ad Breaks | SUM(NUM_AD_BREAKS) or WHERE NUM_AD_BREAKS > 0 |
Play Sequence | case ps.play_seq when 1 then 'Seed' else 'Incremental' END as is_seed_play, |
Site ID | WHERE SITE_ID = '' or GROUP BY SITE_ID |
Date/Time
Dimension | SQL Query |
---|---|
Start Time | date_trunc('hour', convert_timezone('UTC', 'America/New_York', sap.START_TIME)) as EASTERN_START_HR |
Viewer Timezone Offset | DATEADD('hour', VIEWER_TZ_OFFSET, START_TIME) as local_start_time |
Device
Dimension | SQL Query |
---|---|
App User ID (Apps) | count(distinct ott_app_user_id) or HLL(ott_app_user_id) |
Average Player Height | AVG(AVG_PLAYER_HEIGHT) |
Average Player Width | AVG(AVG_PLAYER_WIDTH) |
Device Model | where ott_device_model = '' or group by ott_device_model |
Operating System Language (Apps) | where ott_os_language = '' |
Operating System | WHERE OPERATING_SYSTEM IN/= '' or GROUP BY OPERATING_SYSTEM |
Operating System (Apps) | where ott_device_firmware_version = '' or group by ott_device_firmware_version |
Geography
Dimension | SQL Query |
---|---|
Region | WHERE REGION = '' or GROUP BY REGION |
Placement
Dimension | SQL Query |
---|---|
Domain | WHERE PAGE_DOMAIN = '' or GROUP BY PAGE_DOMAIN |
Dynamic Strategy Rules Queries
These SQL queries demonstrate how to retrieve data from the Dynamic Strategy Rules Data Sharing datasets.
Daily Placement Metrics View
The following SQL query creates a compiled view that rolls up daily metrics from JW_DATA.PUBLIC.PLAY_SESSIONS_<jw_account_id>
and JW_DATA.PUBLIC.PLACEMENT_SESSIONS_<jw_account_id>
. This view is grouped by placement_embed_id
, site_id
, placement_id
, and eastern_date
.
To customize this query, uncomment the joins related to strategy outcomes (
strategy_outcome_id
) and placement evaluations (placement_eval_id
).
with placement_play_sessions as (
select
eastern_date
, site_id
, placement_id
, placement_embed_id
-- , strategy_outcome_id
-- , placement_eval_id
, sum(case when embeds > 0 then 1 else 0 end) as embeds
, sum(case when play_attempts > 0 then 1 else 0 end) as play_attempts
, sum(case when plays > 0 then 1 else 0 end) as plays
, sum(case when completes > 0 then 1 else 0 end) as completes
, sum(ad_requests) as ad_requests
, sum(ad_impressions) as ad_impressions
, sum(ad_viewable_impressions) as ad_viewable_impressions
, sum(ad_completes) as ad_completes
, sum(ad_clicks) as ad_clicks
, sum(ad_skips) as ad_skips
-- , count(distinct viewer_id) as player_viewers
from prd_analytics.public.play_sessions
where placement_id is not null
group by
eastern_date
, site_id
, placement_id
, placement_embed_id
-- ,strategy_outcome_id
-- ,placement_eval_id
)
select
pcms.site_id
, pcms.eastern_date
, pcms.placement_id
, pcms.strategy_outcome_id
-- placement_sessions metrics
, sum(pcms.placement_embed_attempts) as placement_embed_attempts
, sum(pcms.placement_embeds) as placement_embeds
, sum(pcms.strategy_evaluations) as strategy_evaluations
, sum(pcms.strategy_evaluation_errors) as strategy_evaluation_errors
-- , count(distinct pcms.viewer_id) as placement_viewers
-- play_sessions metrics
, sum(ps.embeds) as embeds
, sum(ps.play_attempts) as play_attempts
, sum(ps.plays) as plays
, sum(ps.completes) as completes
, sum(ps.ad_requests) as ad_requests
, sum(ps.ad_impressions) as ad_impressions
, sum(ps.ad_viewable_impressions) as ad_viewable_impressions
, sum(ps.ad_completes) as ad_completes
, sum(ps.ad_clicks) as ad_clicks
, sum(ps.ad_skips) as ad_skips
-- , sum(ps.player_viewers) as player_viewers
from prd_analytics.public.placement_sessions pcms
left join placement_play_sessions ps
on pcms.eastern_date = ps.eastern_date
and pcms.site_id = ps.site_id
and pcms.placement_id = ps.placement_id
and pcms.placement_embed_id = ps.placement_embed_id
-- and pcms.strategy_outcome_id = ps.strategy_outcome_id
-- and pcms.placement_eval_id = ps.placement_evaluation_id
where pcms.eastern_date = current_date - 1
group by
pcms.site_id
, pcms.eastern_date
, pcms.placement_id
, pcms.strategy_outcome_id;
Placement Performance Metrics
The following SQL query creates a placement performance report with the following filter conditions:
- Placements must be enabled and deliverable.
- Date range is the last 7 complete days.
- Data are grouped by
device_type
, a detailed breakdown not typically included in roll-up views such asstrategy_outcome_daily_metrics
orplacement_daily_metrics
. - Metrics include total plays split by US and international viewers.
To customize this query, uncomment the joins related to strategy outcomes (
strategy_outcome_id
) and placement evaluations (placement_eval_id
).
with placement_play_sessions as (
select
eastern_date
, site_id
, placement_id
, placement_embed_id
-- , strategy_outcome_id
-- , placement_eval_id
, sum(case when embeds > 0 then 1 else 0 end) as embeds
, sum(case when play_attempts > 0 then 1 else 0 end) as play_attempts
, sum(case when plays > 0 then 1 else 0 end) as plays
-- example of how metrics can be split by a dimension
, sum(case when plays > 0 and country_code = 'US' then 1 else 0 end) as us_plays
, sum(case when plays > 0 and country_code <> 'US' then 1 else 0 end) as international_plays
--
, sum(case when completes > 0 then 1 else 0 end) as completes
, sum(ad_impressions) as ad_impressions
, sum(ad_viewable_impressions) as ad_viewable_impressions
, sum(ad_completes) as ad_completes
, sum(ad_clicks) as ad_clicks
, sum(ad_skips) as ad_skips
, count(distinct viewer_id) as player_viewers
from play_sessions_view
where placement_id is not null
group by
eastern_date
, site_id
, placement_id
, placement_embed_id
)
select
pcmd.id as placement_id
, pcmd.site_id
, pcmd.strategy_id
, pcmd.name as placement_name
, pcmd.description as placement_description
, pcms.device_type
-- placement_sessions metrics
, sum(pcms.placement_loads) as placement_loads
, sum(pcms.placement_embed_attempts) as placement_embed_attempts
, sum(pcms.placement_embeds) as placement_embeds
, sum(pcms.strategy_evaluations) as strategy_evaluations
, sum(pcms.strategy_evaluation_errors) as strategy_evaluation_errors
, count(distinct pcms.viewer_id) as placement_viewers
-- play_sessions metrics
, sum(ps.embeds) as embeds
, sum(ps.play_attempts) as play_attempts
, sum(ps.plays) as plays
-- example of how metrics can be split by a dimension
, sum(us_plays) as us_plays
, sum(international_plays) as international_plays
--
, sum(ps.completes) as completes
, sum(ps.ad_impressions) as ad_impressions
, sum(ps.ad_viewable_impressions) as ad_viewable_impressions
, sum(ps.ad_completes) as ad_completes
, sum(ps.ad_clicks) as ad_clicks
, sum(ps.ad_skips) as ad_skips
, sum(player_viewers) as player_viewers
from placement_definitions_view pcmd
left join placement_sessions_view pcms
on pcms.placement_id = pcmd.id
left join placement_play_sessions ps
on ps.site_id = pcms.site_id
and ps.eastern_date = pcms.eastern_date
and ps.placement_id = pcms.placement_id
and ps.placement_embed_id = pcms.placement_embed_id
-- and ps.strategy_outcome_id = pcms.strategy_outcome_id
-- and ps.placement_evaluation_id = pcms.placement_eval_id
where
pcmd.enabled
and pcmd.is_deliverable
and not pcmd.is_deleted
and pcms.eastern_date >= current_date - 7 and pcms.eastern_date < current_date
group by
pcmd.id
, pcmd.site_id
, pcmd.strategy_id
, pcmd.name
, pcmd.description
, pcms.device_type
Updated about 2 months ago