I’m seeing more aggregated date data, but now I’m getting user configuration errors all over the place. Nothing has changed on my end. I’d give you the error messages but we all know Looker Studio error messages are completely useless.
Thanks for this! My dashboards appear to be updating properly, now.
I think any visualization I’m using which has a non COUNT / COUNT DISTINCT metric in it is broken. Specifically SUM and AVG of rows with just simple numbers in them is breaking where data source is an aggregated CSV
Looks like the fix for the merged CSVs in datasets seemed to overwrite the fact that some fields in my data were numbers, so I believe it was erroring out trying to do SUMs and AVGs on text. I had to manually go reassert the fields as numbers to rectify my issues.
How did you do this? When I try to manually specify a field as the date - i get the “can’t convert to date” error no matter what date format I try…The existing date format that used to work prior to the UI update is no longer working and any other methods of writing the date (e.g. 20250101) or formatting the date field is still throwing that error for me…
Olá!
Você conseguiu resolver esse problema? Aqui está ocorrendo a mesma coisa, e tenho uma base com mais de 350 arquivos CSV, e ao realizar o upload de novo arquivo está bagunçando toda a base.
Hello! I have a database with over 350 CSV files. Recently, when uploading a file, it’s causing the semantics to become corrupted, and the database won’t load anymore. It displays the message “this database has been incorrectly configured,” and I simply can’t fix it, even after manually changing the semantic settings. It only returns to normal when I delete the last CSV file and reconnect to the database. Is anyone else experiencing this? I don’t know how to solve it.
Hi Nucleo, this looks like a problem, a little more context of your database would definitely help, is it like a Bigquery Table or something else ?
My data source is based on CSV spreadsheets, all with the same column structure and formatting. I have been feeding this dataset continuously since 2023, using CSV files from multiple departments. Currently, the base consists of 357 CSV files, all following the same schema, which has always worked properly until recently.
All CSV files are saved using UTF-8 encoding and follow consistent formatting rules. Date fields have always been formatted using the MM-DD-YYYY date format, which has been correctly recognized by Looker Studio in the past.
After recent updates to Looker Studio, when I upload a new CSV file into the existing dataset, Looker Studio automatically changes the semantic configuration of some fields (especially numeric fields and date fields). As part of this issue, numeric fields are being converted into text fields. As a result, existing charts and calculated fields break, and the system shows the error message:
“This data source was configured incorrectly.”
To temporarily fix the dashboards, I need to delete the most recently uploaded CSV file (the one that triggered the semantic changes) and then manually reconfigure the affected fields in the data source. However, this manual configuration does not persist. As soon as a new CSV file is uploaded, the semantic configuration is changed again, and the same errors reappear.
Because of this behavior, I am currently unable to update the dataset with new CSV files, even though all files follow the same structure and formatting.
Important points:
- All CSV files have the same column names, order, and structure.
- CSV files are encoded in UTF-8.
- Date fields consistently use the MM-DD-YYYY format.
- The formatting and ingestion process have not changed.
- This workflow has been stable and functional since 2023.
- The issue started occurring only after recent updates to the Looker Studio user experience.
- The problem occurs specifically when adding a new CSV file to the existing dataset.
I would like to understand whether recent updates introduced changes in how CSV uploads are parsed or how semantic configurations are handled, and if there is any required procedure or best practice to ensure that semantic field configurations are preserved when uploading new CSV files.
At this moment, I am not aware of any changes in my process that would justify this behavior.