Storage transfer service - Google Ads geo data is incorrect

[ashalitkin]

Hi there,

I've set up a Storage Transfer Service to pull data from the Google Ads to BigQuery which serves as a DWH system for us. The issue is that we are interested in GEO and as suggested by Google Docs I tried to query GeoStats views.
The issue is that the numbers (e.g. number of clicks) don't much.
E.g.

SELECT

    _DATA_DATE, sum(clicks)

FROM `_project_._dataset_.GeoStats_XXX`

where LocationType='LOCATION_OF_PRESENCE'

and IsTargetingLocation IN (true, false)
group by 1

It constantly shows fewer clicks than the Google Ads platform or a similar query based on the CampaignBasicStats table.
Is it any solution to it or is it a bug in the Transfer Service?
I've also noticed that there are very few rows with the IsTargetingLocation=false for the LOCATION_OF_PRESENCE LocationType.

[Snoshone07]

Hi @ashalitkin It sounds like you're encountering a discrepancy between the clicks reported in the GeoStats table and what you're seeing directly in Google Ads or when querying the CampaignBasicStats table.

There are a few things you might want to consider to troubleshoot this:

1. Location Type and Filters: The LOCATION_OF_PRESENCE and IsTargetingLocation filters can have a significant impact on the results. The GeoStats table specifically tracks the geographical location of users (their presence), whereas other stats tables like CampaignBasicStats may aggregate data across other types of location targeting.

   • Make sure that the filters you're applying (e.g., IsTargetingLocation being either true or false) align with how you expect the data to behave.
   • It's also worth reviewing whether Google Ads uses other location targeting types (such as location of interest) that may not be captured in your current query.

2. Sampling/Delays: There could also be some discrepancies due to sampling or data freshness. Ensure that the data from your transfers is up-to-date and matches the reporting period you're comparing against in the Google Ads platform.

3. Data Granularity: The GeoStats table may offer more granular breakdowns of the data, potentially resulting in fewer clicks depending on how the location data is processed (i.e., removing some data points for locations that don't meet certain criteria).

4. IsTargetingLocation=false: The fact that there are very few rows where IsTargetingLocation=false for LOCATION_OF_PRESENCE could indicate that users' physical presence is being tracked mostly when they are in targeted locations, which might not capture all the possible clicks that could be attributed elsewhere (like LOCATION_OF_INTEREST or other campaign-specific factors).

5. Alternative Approach: If this discrepancy continues, you might want to explore other data aggregation approaches using a different API or connector. For instance, using a third-party connector like Windsor.ai can give you more flexibility in handling and harmonizing data from Google Ads and BigQuery, particularly when it comes to location-based insights.

Hope this helps you!