Quicksight - Setting Date Format from an S3 / Spiced csv file not working - direct upload working

Hi - The option to set the date format from a S3/SPICE csv file appears to have stopped working. If I add the CSV file directly as a new dataset in Quicksight then it is possible to set the correct date format.

I’m not sure how to best raise this as an issue - if there are any quicksight admins reading this then maybe this could be assigned to the correct team.

Hi @john.bridge ,
what are the steps you tried?

I’ve tried creating a fake csv like

date,temperature,humidity,location
19/09/2025 22:00,18.5,65,New York
19/09/2025 23:00,17.8,68,New York
20/09/2025 00:00,17.2,70,New York
20/09/2025 01:00,16.9,72,New York
20/09/2025 02:00,16.5,73,New York

After loading it, the date field is recognized as a string field.
I’ve then changed the format and added dd/MM/yyyy HH:mm as date format, then clicked on Validate and obtained this:

Have you performed these steps, right?
Maybe was a temporary error (I’ve just tested it and it seems to be working), so first of all I encourage you to retry, please :smiley:

If it’s still not working, can you paste here in plane a sample of your CSV data? It seems quite straight what you have done, maybe there are some hidden strange characters. Just trying to understand possible issues :slight_smile:

Andrea

Hi Andrea, looks like you just imported as a csv file directly into QS - as mention this part is ok.

The problem is all my files are on S3 therefore i use a manifest file for all my datasets. These have worked every time until now.

As a test i have used your example csv, drop this into S3, using the below manifest in my eval area. Loaded this into Quicksight and now i am not able to set the date format again same error.

Manifest file used..

{
“fileLocations”: [
{
“URIPrefixes”: [
https://analytics-evaluation-dev.s3-eu-west-1.amazonaws.com/tasks/”
]
}

],
"globalUploadSettings" : {
    "format": "CSV"
}

}

Your examlpe csv on S3

Now you not able to set date format

That’s quite a strange behavior, and I can confirm I can replicate it using a manifest.
I’ll try to go deeper on this with the team.

In the meanwhile, a workaround can be adding a calculated field like parseDate({date}, 'dd/MM/yyyy HH:mm').

Not the ideal solution, but it works :upside_down_face:

Andrea

Thanks Andra, that not a option for us due to other dependency’s within our set analyses. Plus it will break our standard onboarding templates when importing the datasets into the prebuilt analyses. Please could you keep me updated. Thanks for you help

1 Like

Hi @john.bridge ,
just want to let you know we have raised the issue and we’re working on it!

Andrea

1 Like

Thanks for the update @andrepgn and your support on getting this issue fixed.

1 Like

Hi @john.bridge,

Hope everything is well with you! Do you have any other questions/concerns that pertain to this topic? If you do not and/or we do not receive a response within the next 3 business days, I’ll go ahead and archive this topic.

Thank you!

Hi @WLS-Luis ,

This is being tracked under AWS Support Case 175879802900311 and is still under investigation, so no fix has been released yet.

While there’s no public-facing bug tracker for QuickSight, keeping this thread active may help others who run into the same problem. However If you need to archive this topic go ahead

1 Like

Hi @john.bridge,

Thanks for letting me know about the current support case and that is completely understandable! Will keep this topic open then for a bit to allow time for closure on the support case.