Why Looker Studio cannot connect to your GA4 data set

Starting on November 7th, 2022, Looker Studio users of the GA4 connector have been seeing various error messages. The errors include "Looker Studio cannot connect to your data set", "Data Set Configuration Error". Another is "Failed to fetch data from the underlying data set".

The reason is that Google has turned on quota limiting for the Looker Studio GA4 connector, and in fact ALL clients of the Google Analytics Data API, as mentioned in their release notes and in the Looker Studio Help forum. It was also confirmed by the Product Management Lead for GA360 on Twitter. 

Most users are completely unaware of quota limiting because previously it was not being enforced.

The solution is to load up a historic dataset in batches that are within your quota. Then keep that dataset updated incrementally over a 3-5 day window, which uses far less quota. Push that data to a source that doesn't have quota in Looker, such as BigQuery or a Partner Data Connector that uses Looker Studio's Advanced Services, like Analytics Canvas.

This is the preferred method that Google wants people to use, rather than pounding at the GA4 API for each chart, table, scorecard and user who lands on the report.

More on the Analytics Canvas solution below, but first a detailed description of the error and what is causing it. Watch the video or continue reading below.

The first message you might see is "Looker Studio cannot connect to your Dataset"  When you click "see details", you may see one or more of the following: 

Quota is tracked at the GA4 property level, impacting all users of the property. There are both hourly and daily quota. This is why some users say that their reports work for a short time, or only at the start of an hour - because they haven't yet hit one of these two quotas. 

If you are seeing one or more of the messages above, your issue is related to GA4 API quotas being exceeded. 

The release notes only mention the concurrency limits, but the troubleshooting guide says that all quota limitations are enforced. What we have seen in our testing is that all the quota rules are in place.

The full details of the GA4 data API quota can be found here.

The key quotas are the concurrent requests, and the hourly and daily token limits. The concurrent limit is simply the number of requests at once- this is limited to 10.  Looker studio could exceed this if the combination of number of charts and users totals more than 10 at a time. 

Google's recommendation (not ideal) is “Consider sharing the report with fewer people”.

The other limit is based on the total number of tokens consumed in either an hour, or day per project. 

Every time you do a query, it consumes a certain number of tokens.  Once the limit of tokens is reached, the API returns an error.

For Looker studio you have 1250  1750 Tokens per hour, with not more than 25,000 tokens per day for all tools you use. 

This is important- if you are using other tools they may start to give you quota errors too.  This could be because Looker studio has consumed your daily quota.

The good news for 360 customers is that your quota is ten times this- 17,500 tokens per hour and 250,000 per day.

The cost of a query is affected by the following factors:

• Date range of the query
• Number of columns in the query
• Number of rows involved (how much traffic the site gets affects this)
• Complexity of any filters

Google says that most requests will be less than 10 tokens.  However, using a simple query of Date, Session Source/Medium and Sessions for a 1 year period on a low traffic site consumed 25 Tokens. In many cases a Looker studio query will consume much more than 10 tokens per query, particularly if the resulting data set is large. 

If you use a tool like ours (Analytics Canvas) you can see the quota used by each query easily.   Every query shows how many tokens were used. This will give you an idea of what the same query will cost when used in Looker Studio.

While it very much depends on the above factors, in general, let's use the figure of 5 tokens per query.  This means you have 1250/5, at most about 250 queries an hour.

In Looker Studio, many charts require 2 queries.  If we look at a fairly rich report that has a number of items on it (some charts, some score cards etc.) we might have 20 or 30 queries on a page.

It’s a rough estimate, but you might only be able to refresh this page 10 or 20  times an hour before running out of tokens.  If you have 5 users, then that means just 4 filter changes each!

Remember, there is caching, so sometimes that can avoid a query.  But if you have multiple users, and they are changing filters and date ranges, then you will quickly consume those tokens for a given property.  If you have a lot of data, and more complex queries you might exceed quota more quickly.  When the hourly quota is exceeded, you will get error messages on all reports for that property.

The hourly quota counters reset at the top of each hour (1:00, 2:00, 3:00 etc.)  the daily quota counters reset at midnight Pacific Standard Time (PST) for all Properties, regardless of your time-zone setting.  For this reason, you might notice that your Looker Studio reports tend to fail at roughly the same time each hour of the day.  Then start working again at the top of the next hour.  If you exceed 25,000 tokens, you are then blocked for the rest of the day PST.

Unfortunately, as a result of this change, the GA4 Looker Studio connector will be quite limited. The problem will also affect other Looker Studio connectors unless they buffer the data.

Google suggests:

1. Reducing the traffic to the report by sharing your reports with fewer users, however obviously, this doesn’t work if your users need access to their data!
2. Reducing the number of charts on each page (you can also reduce the time period in the charts, which will consume less quota). 
3. Using Extracts for faster performance, however this only works if the extracts don't consume all of your quota!

The best solution is to copy the data into a table in a data source that is not affected by quota, such as BigQuery or the Analytics Canvas partner data connector, and create your Looker Studio report from that.  You must then update that table, ideally incrementally (just reload the new data, not all the historical data) to avoid running out of quota when refreshing.

While we are obviously biased, we think that Analytics Canvas is the best way to do this.  You can get your data from GA4 into Looker Studio without having to have your own BigQuery account, and without having to write any code.

Canvas allows you to easily create the queries you need and automatically refreshes them incrementally, consuming far less quota. Canvas also has a fast Looker Studio partner connector to host your data and make it available to all your users, no matter how many you have. 

GA4 queries might take 40 seconds or more to refresh with the Looker Studio connector, and will fail repeatedly due to quota limits.  Whereas with Analytics Canvas, they will refresh in seconds, and never exceed quota since it is not hitting the GA4 API directly. 

Canvas doesn’t use up the API quota because it loads the data regularly so it is up to date, but not every time someone changes a filter or opens a Looker Studio report.  This lets you manage your quota and have as many Looker Studio users making as many ad-hoc reports as they want.

Analytics Canvas provides a powerful way to query GA4 data and access it directly in Looker studio with the Analytics Canvas partner connector. To learn more about our GA4 capabilities, check out this post on our blog, connect with us for a demo and discussion, or jump right in to a free trial.