Ever since the UI update in Looker Studio, I’ve been having issues with the date field when uploading new CSV files as a data source. When I try to upload my file, no matter what date format I try (even as text formatted in ISO date format), I get the error message that Looker Studio “Can’t convert to date” and will only recognize my date field as “text”. It suggests using a parse formula but parse formula doesn’t seem to work if I want to maintain the existing formatting/structure of all my tables and charts for YoY analysis and I get errors within the chart themselves when going down this route. I get the “System Error” message with an Error ID of: df723a1e.
This seems to be a bug related to csv file uploads as this doesn’t seem to be an issue when using Google Sheets. However, I have too much data to fit into a single Google Sheet so need to continue with the route of uploading csv files as a merged data set.
This behavior stems from recent changes in Looker Studio’s CSV ingestion pipeline that now enforces stricter type detection for date fields. The upload parser expects dates to conform exactly to ISO 8601 (YYYY-MM-DD or YYYY-MM-DDTHH:MM:SSZ) without locale variations or embedded delimiters, and any mismatch causes the column to be treated as text. When this happens, the PARSE_DATE() or PARSE_DATETIME() functions can still be applied in calculated fields, but existing chart structures depending on a date dimension may break until the field type is redefined. To restore your original setup, re-upload the CSV ensuring the date column is explicitly formatted in ISO 8601 and set the field type manually in Resource > Manage added data sources > Edit Schema, changing the column type from Text to Date before applying it to your charts. If your dataset exceeds the Google Sheets size limit, consider storing it in BigQuery and connecting Looker Studio directly, as BigQuery preserves date types natively and avoids CSV parsing restrictions. If the issue persists even with ISO-compliant data, capture the Error ID and report it through Looker Studio feedback since this parsing bug is known under ongoing review in recent releases.
Thanks for the reply - I am using ISO-compliant data but not getting an Error ID. I just get the message that Looker studio “Can’t convert to date” and then when I use the parse formula, I get a “System Error” in my charts/graphs with an error ID. One example of an error ID is: df723a1e.
I have already also reported it through Looker Studio feedback.
Yes, you’re not alone, this does look like a Looker Studio CSV ingestion issue that started showing up after the recent UI update. I’ve seen the same behavior where valid date formats (even ISO) get locked as text and PARSE_DATE() either breaks existing charts or throws generic system errors. The fact that it works fine with Google Sheets but not CSVs strongly points to a bug in the CSV connector rather than your data.
If Sheets isn’t an option because of volume, one workaround is to avoid CSV uploads altogether and use a database or managed connector as the source. For example, some people move the data into BigQuery and connect Looker Studio from there, date types are preserved properly and YoY logic keeps working.
Another approach I’ve seen work is using a sync tool like Windsor.ai to load your data into Google Sheets or BigQuery instead of uploading CSVs manually. Since those connectors define the schema explicitly, Looker Studio receives the field as a real DATE type, not inferred text, which avoids the conversion issue entirely.
Until Google fixes the CSV parser, the most stable solution seems to be switching the source rather than fighting with parse formulas inside Looker Studio