This article outlines recommendations and notes certain limitations imposed by the data source API, which you may need to consider when building your queries.
If you are connecting to this data source, please follow the instructions in our connection guide, and the available fields for this connector are listed in the metrics and dimensions documentation.
Request limits and recommended refresh rate
Request limit:
Tokens per property per day: 200,000 (Google Analytics 360: 2,000,000)
Tokens per property per hour: 40,000 (Google Analytics 360: 400,000)
Concurrent requests per property: 10 (Google Analytics 360: 50)
Recommended refresh window: Set to 12 days to allow for modelled conversions to be finalized.
Historical range
User-level data, including conversions, is deleted after 14 months, and age, gender, and interest data after 2 months. For all other event data, you may choose the length of retention to be 2 or 14 months. Learn more in Google’s documentation. If you’re using Google Analytics 360, you can set the data retention to 26, 38, or 50 months. Learn more in Google’s documentation.
Accounts in Google Analytics 4
In the Google Analytics 4 connector, each Property is considered a separate account within the connection.
Your Supermetrics subscription determines the number of data source accounts you can use simultaneously.
Things to consider
Dimension and metric limitations
Google Analytics 4’s API only allows a maximum of 9 dimensions and 10 metrics to be fetched in a single query.
Where a query has a date range of one day (today/yesterday/startdate = endate) date is removed as a dimension, and after the request requested date is added as a value.
Where a query has a date range of more than one day, date is counted as a dimension and therefore takes one of the 9 available dimension slots.
The Explore tab does not match API data
Data from the Explore tab in Google Analytics is often used to compare Supermetrics data, as it allows users to query metrics and dimensions as they wish. However, the data there is sometimes poor, and the Google Support team has warned that this is a Beta feature that shouldn’t be trusted 100%.
Also note, the total at the top of the table isn’t the aggregated values in the table. If the rows contain duplicated values, they won’t be counted several times in total. If the rows contain unique values, the aggregated total won’t match.
Support for segments
Currently, the Data API that Google Analytics 4 uses does not support segments. As a workaround, you can create a filter in place of a segment; however, there isn’t a way to map this 1-to-1. You may need to experiment with how best to filter your data to match your segment as closely as possible.
Discrepancies with Active users or total users are queried together with non-user related metrics
In the Google Analytics 4 user interface and the API Query Explorer, rows where active users or total users are equal to 0 will return no data for other metrics, even if these metrics contain data for these rows. When Google hides the user related metric, the whole row is hidden, even though some metrics are not subject to thresholding.
Attribution modeling
Google Analytics 4’s Data API does not appear to support functionality for attribution modeling, either getting which model was used or setting which attribution model to use in the query.
Historical limitations for advertiser metrics and date breakdowns
In early June 2026, Google will be making changes for their data retention policies for Google Ads. Due to these changes, advertiser ad metrics, which are sourced from Google Ads, can only be fetched for the last 37 months when using certain date breakdowns.
This change will affect the metrics and dimensions listed below.
Metrics:
Ads clicks
Ads cost
Ads cost per click
Cost per conversion
Ads impressions
Dimensions:
Date
Nth day
Week (Sun-Sat)
Week (Mon-Sun)
Day of week (Sun-Sat)
Day of week (Mon-Sun)
Year & week (Sun-Sat)
Year & week (Mon-Sun)
If a query contains one or more of the advertiser metrics, and one or more of the date dimensions, and has a start date longer than 37 months in the past, the query will fail with an error.
To address this, the query needs to be edited to either move the start date to within the last 37 months OR remove the date dimension.