Review Data Transfer sample queries

Unfilled impressions

Unfilled impressions from NetworkImpressions

To find the number of unfilled impressions for a day, query NetworkImpressions for entries where LineItemID is 0 . There are no unfilled impressions in NetworkBackfillImpressions .

SELECT
 COUNT(1) AS UnfilledImpressions
FROM
 NetworkImpressions
WHERE
 LineItemID = 0
 AND Time >= ‘2020-01-01’ AND Time < ‘2020-01–02’

Unfilled impressions from NetworkRequests

You can also find the number of unfilled impressions by querying NetworkRequests . Look for requests where IsFilledRequest is false. There are no unfilled impressions in NetworkBackfillRequests .

SELECT
 COUNT(1) AS UnfilledImpressions
FROM
 NetworkRequests
WHERE
 NOT IsFilledRequest 
 AND Time >= '2020-01-01' AND Time < '2020-01-02'

Unfilled impressions by URL

Ad Manager Reporting can show unfilled impressions by ad unit or requested size, but not by URL. Include RefererURL to help you find the top ten pages that generate unfilled impressions.

SELECT
 RefererURL, COUNT(1) AS UnfilledImpressions
FROM 
 NetworkImpressions
WHERE
 LineItemID = 0
 AND Time >= '2020-01-01' AND Time < '2020-01-02'
GROUP BY RefererURL
ORDER BY UnfilledImpressions DESC
LIMIT 10

Unfilled impressions by ad unit

Find the ad units that are responsible for the most unfilled impressions on a single page. If you use the BigQuery Connector , use the Ad Unit Match Table to find the name of the ad units. Because the match table contains ad unit data for every day, be sure to limit the match table data to one day.

SELECT
 AdUnitID, Name AS AdUnitName, COUNT(1) AS UnfilledImpressions
FROM
 NetworkImpressions AS NI
 INNER JOIN MatchTableAdUnit AS MTAU ON 
  AdUnitID = ID
  AND LineItemID = 0
  AND Time >= '2020-01-01' AND Time < '2020-01-02'
  AND RefererURL = 'https://example.com/'
  AND MTAU._DATA_DATE = '2020-01-01'
GROUP BY AdUnitID, AdUnitName
ORDER BY UnfilledImpressions DESC, AdUnitID
LIMIT 10

Impressions

Impressions by Product and DealType

Use the Product and DealType fields in Data Transfer to generate reports comparable to Ad Manager reports that use the "Demand channel," "Programmatic channel," and "Optimization type" dimensions. Select impressions from NetworkImpressions (where LineItemID is not zero) and NetworkBackfillImpressions .

SELECT
 Product, DealType, COUNT(1) AS Impressions
FROM
 NetworkImpressions
WHERE
 LineItemID != 0 
 AND Time >= '2020-01-01' AND Time < '2020-01-02'
GROUP BY Product, DealType
UNION ALL
SELECT
 Product, DealType, COUNT(1) AS Impressions
FROM
 NetworkBackfillImpressions
WHERE
 Time >= '2020-01-01' AND Time < '2020-01-02'
GROUP BY Product, DealType
ORDER BY Product, DealType

Revenue

Revenue for a CPM line item

The NetworkImpressions file does not contain revenue data, but if you use the BigQuery Connector , you can use the Line Item Match Table to find the CPM rate. Otherwise, use the Ad Manager API to find the rate of a line item. Find the revenue for a given CPM line item for a given date range by counting the impressions, multiplying by the rate, and dividing by 1,000. Because the match table contains ad unit data for every day, be sure to limit the match table data to one day.

WITH Impression_Data AS (
 SELECT
   LineItemID, COUNT(1) AS Impressions
 FROM
   NetworkImpressions
 WHERE
   LineItemID = 123456789
   AND Time >= '2020-01-01' AND Time < '2020-01-11'
 GROUP BY
   LineItemID
)
 
SELECT
 LineItemID, Impressions, CostPerUnitInNetworkCurrency AS Rate, CostType, ((Impressions * CostPerUnitInNetworkCurrency) / 1000) AS Revenue
FROM
 Impression_Data
 JOIN MatchTableLineItem ON LineItemID = ID
WHERE
 MatchTableLineItem._DATA_DATE = '2020-01-10'

Revenue for a CPD line item

As with CPM line items, you can use the Line Item Match Table or the Ad Manager API to find the CPD rate of a line item. Because the match table contains ad unit data for every day, be sure to limit the match table data to one day. To find the revenue for a given CPD line item, count the number of days in which impressions served and multiply by the rate. You may want to include the number of impressions served to find the average eCPM.

WITH Impression_Data AS (
 SELECT
   SUBSTR(Time, 0, 10) AS Date,
   LineItemID,
   CostPerUnitInNetworkCurrency AS Rate,
   CostType,
   COUNT(1) AS Impressions
 FROM
   NetworkImpressions
   JOIN MatchTableLineItem ON LineItemID = ID
 WHERE
   LineItemID = 123456789
   AND MatchTableLineItem._DATA_DATE = '2020-01-01'
 GROUP BY
   Date, LineItemID, Rate, CostType
)
SELECT
 LineItemID,
 COUNT(1) AS Days,
 CostType,
 Rate,
 (COUNT(1) * Rate) AS Revenue,
 SUM(Impressions) AS Impressions,
 ROUND((COUNT(1) * Rate) / SUM(Impressions) * 1000, 2) AS Average_eCPM
FROM
 Impression_Data
GROUP BY
 LineItemID, CostType, Rate

Revenue for a CPC line item

As with CPM line items, you can use the Line Item Match Table or the Ad Manager API to find the CPC rate of a line item. Because the match table contains ad unit data for every day, be sure to limit the match table data to one day. To find the revenue for a given CPC line item for a given date range, count the clicks and multiply by the rate. You may want to include the number of impressions served to find the average eCPM.

WITH Impression_Data AS (
 SELECT
   LineItemID,
   COUNT(1) AS Impressions
 FROM
   NetworkImpressions
 WHERE
   LineItemID = 123456789
 GROUP BY
   LineItemID
), Click_Data AS (
 SELECT
   LineItemID,
   CostPerUnitInNetworkCurrency AS Rate,
   CostType,
   COUNT(1) AS Clicks
 FROM
   NetworkClicks
   JOIN MatchTableLineItem ON LineItemID = ID
 WHERE
   LineItemID = 123456789
   AND MatchTableLineItem._DATA_DATE = '2020-01-01'
 GROUP BY
   LineItemID, Rate, CostType
)
 
SELECT
 LineItemID,
 CostType,
 Impressions,
 Clicks,
 ROUND(Clicks / Impressions * 100, 2) AS CTR,
 Rate,
 (Clicks * Rate) AS Revenue,
 ROUND((Clicks * Rate) / Impressions * 1000, 2) AS Average_eCPM
FROM
 Impression_Data
 JOIN Click_Data USING (LineItemID)

Revenue for a vCPM line item

As with CPM line items, you can use the Line Item Match Table or the Ad Manager API to find the vCPM rate of a line item. Because the match table contains ad unit data for every day, be sure to limit the match table data to one day. To find the revenue for a given vCPM line item, count the viewable impressions from NetworkActiveViews and multiply by the rate. You may want to include the number of impressions served to find the average eCPM.

WITH Active_View_Data AS (
 SELECT
   LineItemID, COUNT(1) AS ViewableImpressions
 FROM
   NetworkActiveViews
 WHERE
   LineItemID = 123456789
 GROUP BY LineItemID
), Impression_Data AS (
 SELECT
   LineItemID, COUNT(1) AS Impressions
 FROM
   NetworkImpressions
 WHERE
   LineItemID = 123456789
 GROUP BY LineItemID
)
SELECT
 Active_View_Data.LineItemID,
 CostType,
 Impressions,
 ViewableImpressions,
 CostPerUnitInNetworkCurrency AS Rate,
 (CostPerUnitInNetworkCurrency * ViewableImpressions / 1000) AS Revenue,
 ROUND((CostPerUnitInNetworkCurrency * ViewableImpressions / 1000) / Impressions * 1000, 2) AS Average_eCPM
FROM
 Impression_Data
 JOIN Active_View_Data USING (LineItemID)
 JOIN MatchTableLineItem ON Active_View_Data.LineItemID = ID
WHERE
 MatchTableLineItem._DATA_DATE = '2020-08-01'

Revenue for an advertiser

To find the revenue for a given advertiser for a given date range, count the impressions for each line item and multiply by the rate. Use the Line Item Match Table to find the rate and the Company Match Table to find the advertiser name.

WITH Impression_Data AS (
 SELECT
   AdvertiserID, LineItemID, COUNT(1) AS Impressions
 FROM
   NetworkImpressions
 WHERE
   AdvertiserID = 111222333
   AND Time >= '2020-01-01' AND Time < '2020-01-02'
 GROUP BY
   AdvertiserID, LineItemID
)
 
SELECT
 AdvertiserID,
 MTC.Name AS CompanyName,
 LineItemID, Impressions,
 CostPerUnitInNetworkCurrency AS Rate,
 CostType,
 ((Impressions * CostPerUnitInNetworkCurrency) / 1000) AS Revenue
FROM
 Impression_Data
 JOIN MatchTableLineItem AS MTLI ON LineItemID = MTLI.ID
 JOIN MatchTableCompany AS MTC ON AdvertiserID = MTC.ID
WHERE
 MTLI._DATA_DATE = '2020-01-01'
 AND MTC._DATA_DATE = '2020-01-01'

Code serves

For networks with fallback enabled, Data Transfer counts a code serve for every line item selected in the fallback chain while Ad Manager Reporting counts a code serve for only the first line item selected in the fallback chain. Data Transfer also counts a code serve for companion ads, while Ad Manager Reporting does not. If you are looking for your Data Transfer report to match your Ad Manager report as closely as possible, only count code serves where VideoFallbackPosition = 0 and where IsCompanion is false . Mediation code serves in Data Transfer may not match Mediation code serves in Ad Manager Reporting. Depending on your implementation, there may be other differences between code serve counts in Data Transfer and Ad Manager Reporting.

Code serves, impressions, and render rate by line item for a single advertiser

Find how often code serves turn into impressions for each line item of a direct advertiser. Because we are looking at a direct advertiser, these code serves will only be in NetworkCodeServes and the impressions will only be in NetworkImpressions .

WITH Code_Serve_Data AS (
 SELECT
   LineItemID, COUNT(1) AS CodeServes
 FROM
   NetworkCodeServes
 WHERE
   AdvertiserID = 12345678
   AND VideoFallbackPosition = 0
   AND IsCompanion IS FALSE
   AND Time >= '2020-01-01' AND Time < '2020-01-02'
 GROUP BY LineItemID
), Impression_Data AS (
 SELECT
   LineItemID, COUNT(1) AS Impressions
 FROM
   NetworkImpressions
 WHERE
   AdvertiserID = 12345678
   AND Time >= '2020-01-01' AND Time < '2020-01-02'
 GROUP BY LineItemID
)
SELECT
 LineItemID, 
 CodeServes, 
 Impressions, 
 ROUND((Impressions / CodeServes) * 100, 2) AS RenderRate
FROM
 Code_Serve_Data JOIN Impression_Data USING (LineItemID)
ORDER BY RenderRate DESC

Code serves, impressions, and render rate by device category and creative size delivered

Include the Device Category and the Creative Size Delivered to see how render rates vary for one order of one advertiser.

WITH Code_Serve_Data AS (
 SELECT
   LineItemID, CreativeSizeDelivered, DeviceCategory, COUNT(1) AS CodeServes
 FROM
   NetworkCodeServes
 WHERE
   AdvertiserID = 87654321
   AND OrderID = 1111111111
   AND VideoFallbackPosition = 0
   AND IsCompanion IS FALSE
 GROUP BY LineItemID, CreativeSizeDelivered, DeviceCategory
), Impression_Data AS (
 SELECT
   LineItemID, CreativeSizeDelivered, DeviceCategory, COUNT(1) AS Impressions
 FROM
   NetworkImpressions
 WHERE
   AdvertiserID = 87654321
   AND OrderID = 1111111111
 GROUP BY LineItemID, CreativeSizeDelivered, DeviceCategory
)
SELECT
 LineItemID, 
 DeviceCategory, 
 CreativeSizeDelivered, 
 CodeServes, 
 Impressions, 
 ROUND((Impressions / CodeServes) * 100, 2) AS RenderRate
FROM
 Code_Serve_Data
 JOIN Impression_Data USING (LineItemID, CreativeSizeDelivered, DeviceCategory)
ORDER BY LineItemID, CreativeSizeDelivered, DeviceCategory 

Viewability

Viewability data can be found in both the Impression and ActiveView files using the ActiveViewEligibleCount , ActiveViewMeasurableCount , and ActiveViewViewableCount fields. Read more about these fields and how their values are set .

Viewability for a time period can be calculated by totaling the values found in the files or by joining the individual events and then totalling the values. There can be slight differences between these two strategies, detailed below.

Eligible impressions, measurable impressions, and viewable impressions (in total)

Find the number of eligible, measurable, and viewable impressions for a given day. As noted above, the Impression files and the ActiveView files must be used.

Eligible impressions, measurable impressions, and viewable impressions (by joined impression)

Find the number of eligible, measurable, and viewable impressions for a given day by joining the individual events in the Impression files and the ActiveView files and then totaling the values. There are rare instances in which we received an active view viewable ping but not receive an impression ping. When joining ActiveView data to Impression data, you may find differences in measurable impressions and fewer viewable impressions relative to the first query, shown above.

DECLARE startdate STRING DEFAULT '2023-01-01 00:00:00';
DECLARE enddate STRING DEFAULT '2023-01-02 00:00:00';


WITH ActiveView_Data AS (
 SELECT
   CAST(substr(Time, 0, 10) as Date) AS Date,
   Product,
   KeyPart,
   TimeUsec2,
   SUM(ActiveViewMeasurableCount) AS avAVMC,
   SUM(ActiveViewViewableCount) AS ActiveViewViewableCount
 FROM
   NetworkActiveViews
 WHERE Time >= startdate AND Time < enddate
 GROUP BY Date, Product, KeyPart, TimeUsec2
 UNION ALL
 SELECT
   CAST(substr(Time, 0, 10) as Date) AS Date,
   Product,
   KeyPart,
   TimeUsec2,
   SUM(ActiveViewMeasurableCount) AS avAVMC,
   SUM(ActiveViewViewableCount) AS ActiveViewViewableCount
 FROM
   NetworkBackfillActiveViews
 WHERE Time >= startdate AND Time < enddate
 GROUP BY Date, Product, KeyPart, TimeUsec2 
), Impression_Data AS (
 SELECT
   CAST(substr(i.Time, 0, 10) as Date) AS Date,
   i.Product,
   i.TimeUsec2,
   i.KeyPart,
   i.ActiveViewEligibleCount AS ActiveViewEligibleCount,
   i.ActiveViewMeasurableCount AS iAVMC,
   av.avAVMC AS avAVMC,
   av.ActiveViewViewableCount     
 FROM
   NetworkImpressions i
   LEFT JOIN ActiveView_Data AS av USING (Product, Keypart, TimeUsec2)
 WHERE
   i.Time >= startdate AND i.Time < enddate
   AND i.LineItemID !=0
 UNION ALL
 SELECT
   CAST(substr(i.Time, 0, 10) as Date) AS Date,
   i.Product,
   i.TimeUsec2,
   i.KeyPart,
   i.ActiveViewEligibleCount AS ActiveViewEligibleCount,
   i.ActiveViewMeasurableCount AS iAVMC,
   av.avAVMC AS avAVMC,
   av.ActiveViewViewableCount     
 FROM
   NetworkBackfillImpressions i
   LEFT JOIN ActiveView_Data AS av USING (Product, Keypart, TimeUsec2)
 WHERE
   i.Time >= startdate AND i.Time < enddate
), Full_Data AS (
 SELECT
   Date,
   Product,
   TimeUsec2,
   KeyPart,
   ActiveViewEligibleCount,
   CASE WHEN ActiveViewViewableCount >=1 THEN 1 ELSE (IFNULL(iAVMC, 0) + IFNULL(avAVMC, 0)) END AS ActiveViewMeasurableCount,
   IFNULL(ActiveViewViewableCount, 0) AS ActiveViewViewableCount
 FROM
   Impression_Data
)


SELECT
 SUM(ActiveViewEligibleCount) AS ActiveViewEligibleCount,
 SUM(ActiveViewMeasurableCount) AS ActiveViewMeasurableCount,
 SUM(ActiveViewViewableCount) AS ActiveViewViewableCount
FROM
 Full_Data

Key values

Key usage

Find how often each of your keys appears in an ad request (appears in CustomTargeting ) and how often each key was used to serve a line item (appears in TargetedCustomCriteria ). Active keys that don’t appear in the results or that are infrequently used might be good candidates for archiving in order to stay under your key limit.

WITH Key_Value_Pairs AS (
 SELECT
   KVPair
 FROM
   NetworkImpressions CROSS JOIN UNNEST(SPLIT(CustomTargeting, ';')) AS KVPair
 WHERE
   CustomTargeting IS NOT NULL
 UNION ALL
 SELECT
   KVPair
 FROM
  NetworkBackfillImpressions CROSS JOIN UNNEST(SPLIT(CustomTargeting, ';')) AS KVPair
 WHERE
   CustomTargeting IS NOT NULL
), Targeted_Key_Value_Pairs AS (
 SELECT
   TargetedKVPair
 FROM
   NetworkImpressions CROSS JOIN UNNEST(SPLIT(TargetedCustomCriteria, ';')) AS TargetedKVPair
 WHERE
   TargetedCustomCriteria IS NOT NULL
 UNION ALL
 SELECT
   TargetedKVPair
 FROM
   NetworkBackfillImpressions CROSS JOIN UNNEST (SPLIT(TargetedCustomCriteria, ';')) AS TargetedKVPair
 WHERE
   TargetedCustomCriteria IS NOT NULL
), Key_Usage AS (
 SELECT
   REGEXP_REPLACE(KVPair, '=.+', '') AS Key,
   COUNT(1) AS KeyUsageCount
 FROM Key_Value_Pairs
 GROUP BY Key
), Key_Targeted_Usage AS (
 SELECT
   REGEXP_REPLACE(TargetedKVPair, '(!)*(=|~).+', '') AS Key,
   COUNT(1) AS KeyTargetedCount
 FROM Targeted_Key_Value_Pairs
 GROUP BY Key
)
 
SELECT
 CASE WHEN Key_Usage.Key IS NULL THEN Key_Targeted_Usage.Key ELSE Key_Usage.Key END AS Key,
 KeyUsageCount,
 KeyTargetedCount
FROM
 Key_Usage
 FULL JOIN Key_Targeted_Usage ON Key_Usage.Key = Key_Targeted_Usage.Key
ORDER BY Key

Bids by bidding partners

Find how often each of your partners bids by extracting the bids from CustomTargeting . The example below expects the name of each partner to begin “ bidder_prefix_ ” as in “ bidder_prefix_partnername ”, and it expects a bid for that partner to be in the format “ bidder_prefix_partnername=1.23 ”.

SELECT
 Bidder, COUNT(1) AS BidCount
FROM (
 SELECT
   Bidder
 FROM
   NetworkImpressions CROSS JOIN UNNEST(REGEXP_EXTRACT_ALL(CustomTargeting, '(bidder_prefix_[A-z]+)=[0-9]+\\.[0-9]*')) AS Bidder
 WHERE
   CustomTargeting LIKE '%bidder_prefix_%'
   AND Time >= '2020-01-01' AND Time < '2020-01-02'
 UNION ALL
 SELECT
   Bidder
 FROM
   NetworkBackfillImpressions CROSS JOIN UNNEST(REGEXP_EXTRACT_ALL(CustomTargeting, '(bidder_prefix_[A-z]+)=[0-9]+\\.[0-9]*')) AS Bidder
 WHERE
   CustomTargeting LIKE '%bidder_prefix_%'
   AND Time >= '2020-01-01' AND Time < '2020-01-02'
)
GROUP BY Bidder
ORDER BY BidCount 

Bid values and counts for a single bidding partner

For a single bidding partner, find the most common bid values and how often each bid was made. In the example below, select the 10 most common bids from the impressions tables for the partner named “ bidder_partner ” (impressions where CustomTargeting contains the key “ bidder_partner ” that is set to a bid price, such as “ 1.23 ”).

SELECT
 BidPrice, SUM(BidCount) AS BidCount
FROM (
 SELECT
    SAFE_CAST(REGEXP_EXTRACT(CustomTargeting, 'bidder_partner=([0-9]+\\.[0-9]*)') AS FLOAT64) AS BidPrice,
    COUNT(1) AS BidCount
 FROM
   NetworkImpressions
 WHERE
    CustomTargeting LIKE '%bidder_partner=%'
    AND Time >= '2020-01-01' AND Time < '2020-01-02'
 GROUP BY BidPrice
 UNION ALL
 SELECT
    SAFE_CAST(REGEXP_EXTRACT(CustomTargeting, 'bidder_partner=([0-9]+\\.[0-9]*)') AS FLOAT64) AS BidPrice,
    COUNT(1) AS BidCount
 FROM
   NetworkBackfillImpressions
 WHERE
    CustomTargeting LIKE '%bidder_partner=%'
    AND Time >= '2020-01-01' AND Time < '2020-01-02'
 GROUP BY BidPrice
)
GROUP BY BidPrice
ORDER BY BidCount DESC
LIMIT 10

Bid counts and average bids

Find the total number of bids and the average bid from the impressions tables for all bidding partners. The example below expects the name of each partner to begin “ bidder_prefix_ ” as in “ bidder_prefix_partnername ”, and it expects a bid for that partner to be in the format “ bidder_prefix_partnername=1.23 ”.

WITH Bid_Data AS (
 SELECT
   REGEXP_EXTRACT(Bid, '(bidder_prefix_[A-z]+)=[0-9]+\\.[0-9]*') AS Bidder,
   SAFE_CAST(REGEXP_EXTRACT(Bid, 'bidder_prefix_[A-z]+=([0-9]+\\.[0-9]*)') AS FLOAT64) AS BidPrice,
   COUNT(1) AS BidCount 
   FROM (
     SELECT Bid
     FROM NetworkImpressions CROSS JOIN UNNEST(REGEXP_EXTRACT_ALL(CustomTargeting, 'bidder_prefix_[A-z]+=[0-9]+\\.[0-9]*')) AS Bid
     WHERE
       CustomTargeting LIKE '%bidder_prefix_%'
       AND Time >= '2020-01-01' AND Time < '2020-01-02'
     UNION ALL
     SELECT Bid
     FROM NetworkBackfillImpressions CROSS JOIN UNNEST(REGEXP_EXTRACT_ALL(CustomTargeting, 'bidder_prefix_[A-z]+=[0-9]+\\.[0-9]*')) AS Bid
     WHERE
       CustomTargeting LIKE '%bidder_prefix_%'
       AND Time >= '2020-01-01' AND Time < '2020-01-02'
   )
   GROUP BY Bidder, BidPrice
), BidPrice_Totals AS (
 SELECT
   Bidder, SUM(BidValue) AS TotalBidValue
 FROM (
   SELECT Bidder, BidPrice * BidCount AS BidValue
   FROM Bid_Data
 )
 GROUP BY Bidder
), BidCount_Totals AS (
  SELECT
   Bidder, SUM(BidCount) AS TotalBidCount
  FROM
   Bid_Data
  GROUP BY Bidder
)
 
SELECT
 BidCount_Totals.Bidder,
 TotalBidCount,
 ROUND((TotalBidValue / TotalBidCount), 2) AS AverageBid
FROM
 BidCount_Totals
 INNER JOIN BidPrice_Totals ON BidCount_Totals.Bidder = BidPrice_Totals.Bidder
ORDER BY Bidder

DMP segment counts

Data Management Platforms often pass the segments to which a user belongs as key-value pairs. Find how often these segments appear in ad requests -- how many impressions were eligible to be targeted for each segment. Extract the segment IDs from CustomTargeting. The example below expects the name of the key to be “seg” and the values to be made up of letters and numbers.

SELECT
 Segment, COUNT(1) AS Count
FROM (
 SELECT
   Segment
 FROM
   NetworkImpressions CROSS JOIN UNNEST(REGEXP_EXTRACT_ALL(CustomTargeting, 'seg=([A-z0-9]+)')) AS Segment
 WHERE
   CustomTargeting LIKE '%seg=%'
   AND Time >= '2020-01-01' AND Time < '2020-01-02'
 UNION ALL
 SELECT
   Segment
 FROM
   NetworkBackfillImpressions CROSS JOIN UNNEST(REGEXP_EXTRACT_ALL(CustomTargeting, 'seg=([A-z0-9]+)')) AS Segment
 WHERE
   CustomTargeting LIKE '%seg=%'
   AND Time >= '2020-01-01' AND Time < '2020-01-02'
)
GROUP BY Segment
ORDER BY Count DESC

Video

Video errors by URL, ad unit ID, and position

To troubleshoot video line items with significant errors, you may need to find the page and/or ad slot on the page that is most responsible for the errors. Use NetworkVideoConversions to find errors by line item (where ActionName contains “ error ”). If you have more than one video player on a page, the players use the same ad unit, and you use a key like “ pos ” to distinguish between ad units on a page, extract that pos value from CustomTargeting . The example below expects the name of that key to be “ pos ” and shows the top five combinations of RefererURL , AdUnitID , and Position responsible for errors to a single video line item.

SELECT
 RefererURL, AdUnitID, REGEXP_EXTRACT(CustomTargeting, 'pos=([^;]+)') AS Position, COUNT(1) AS ErrorCount
FROM
 NetworkVideoConversions
WHERE
 LineItemID = 123456789
 AND ActionName LIKE '%error%'
 AND Time >= '2020-01-01' AND Time < '2020-01-02'
GROUP BY RefererURL, AdUnitID, Position
ORDER BY ErrorCount DESC
LIMIT 5