Skip to main content


Connector Improvement: Google Analytics - Make "Date" dimension not mandatory

Not planned

Please sign in to leave a comment.



  • Official comment

    Hi Nicolas and Dan, Drew from the Product Team here!

    The Date field is a required field for Fivetran in order to allow for efficient export of data from the Google Analytics platform. Due to speed and volume limitations, we pull reporting based on date ranges.

    In the past we have evaluated strategies to pull data without the Date field, but it led to unreliability and other issues.

    We could potentially reevaluate this with the newer Data API for GA4, but for now I will be marking this as not planned.

    Completely agree. This is not required by the API so should not be required by Fivetran.

    I ran also into this major issue.
    For clients we would like to generate reports with specific metrics on a month based aggregation. Now FiveTran requires me to add the Date dimension (which is in YYYYMMDD).
    But including the Date dimension would skew the numbers of newUsers and Users metrics. Day granularity/aggregation does not simply add up to a month granularity/aggregation.
    That is the way how Google Analytics calculates new users.
    Example taken from the source Google Analytics: 1 nov 2021 - 31 dec 2021
    Day aggregation:
    If you add up newUsers per day nov 2021 the total will be: 33,245
    If you add up newUsers per day dec 2021 the total will be: 29,161
    Total of 62,406
    Month aggregation:
    nov 2021: 33,225
    nov 2021: 29,133
    Total: 62,358
    Clients require us to report also on weekly, monthly and yearly granularity.  In fact they will compare the results of a custom dashboard with what they see in Google Analytics UI.
    Other API/ingestion tools like SuperMetrics, Google Analytics AddIn for Google Sheets, etc do not require the Date dimension.
    Looking forward to a solution.
    Thank you.

    With kind regards,

    André Koot <>

    T. +3171 542 53 06 | M. +316 15 87 25 24

    Maxlead | Wilhelminapark 17 | 2342 AD Oegstgeest

    As an alternative suggestion, is there a way for fivetran to prepare a "Standard report" that aggregates the numbers by week or month already and that you can then pull by date? The report could be either a last 7 days report or a last month report where the numbers for the latter would be identical for each date but at least you had the correct number to pull and can then filter for only on date in the visualization later

    I can only agree that this feature is regularly requested by our clients and is delivered by other tools via the Analytics API

    Upvoting this feature for a reason that i've come across recently.

    The activeUsers dilemma which i am sure we are all familiar with - 

     In order to resolve the activeusers issues there is in fact a way to fetch the correct Active users count ( as seen on the GA UI) using the API. This can be done by not selecting a date dimension (or any dimension actually) and just adding metrics into your custom report. This effectively gives you a distinct count of users. This would then allow data synced from Fivetran to be more closely aligned to the GA UI, which users are accustomed to. One case i had recently ( for a month) there was a 70,000 difference between the activeUsers reported in the UI, and from the API.

    It would be great to utilise the date filter advanced functionality to say 1000daysAgo to 1daysAgo and have that be a rolling report, by date/day so that you end up with - 

    property, date (yyyymmdd), activeUsers