What is the correct way to use Azure AI Search to search for CSV data

Question

What is the correct way to use Azure AI Search to search for CSV data

Sukesh 0

Hello Everyone,

I have a .csv file with employee sales data which has records of sales made by employees every month. An employee can have multiple rows in that csv. I have uploaded this to Azure storage account. I want to Azure AI Search to answer queries like:

What is highest order made by John Doe
Who made the lowest sale order
What is the total revenue for June 2025 etc.
I want to test this first in the Azure Foundry chat playground.

Here are the steps I followed but that is not giving the expected responses.
Could you please tell me what am I doing wrong and what is the correct approach.

Uploaded csv file to Azure storage account.
Create a data source in AI Search service.
Created a new index.

Screenshot 2025-09-30 132000

Created a new indexer.

Screenshot 2025-09-30 132214

The documents count generated in the new index match the number of rows in the csv.

Screenshot 2025-09-30 132651

In the chat playground, added the index s data source and select "keyword" as search type.

Screenshot 2025-09-30 133228

The response is not correct. I'm expecting a response like $200.

Screenshot 2025-09-30 133325

Thank you

Praneeth Maddali 1,110 Reputation points Microsoft External Staff Moderator

2025-10-02T08:21:05.61+00:00

Hi @Sukesh
I'm just following up to see if your issue has been resolved or if you've had a chance to review my previous comment?

1 answer

Your answer

Praneeth Maddali 1,110 Reputation points Microsoft External Staff Moderator

2025-10-02T08:21:05.61+00:00

Hi @Sukesh
I'm just following up to see if your issue has been resolved or if you've had a chance to review my previous comment?

Answer 1

Praneeth Maddali 1,110 Microsoft External Staff Moderator

Hi @Sukesh

Thank you for reaching out regarding the issue with querying your CSV data in Azure AI Search.

Looks like The issue occurs because the CSV indexer does not have a delimiter setting, which leads to incomplete field mapping, such as missing the Total_Cost field. Additionally, the Keyword search type in the chat playground has difficulty with natural language queries and aggregations. While Azure AI Search can retrieve the data, it depends on the LLM to calculate sums or maximum values on the client side, which can result in incomplete or irrelevant outcomes. To resolve this, set the delimiter to , for accurate CSV parsing and use Semantic search for improved query performance.

To resolve this and receive the correct responses (such as "$200" for totals), please follow these steps:

Correct CSV Parsing in the Indexer:
- Go to Azure AI Search > Indexers > Edit index-csv-data.
- Set the Parsing mode to Delimited text, choose Delimiter character as ,, and ensure First line contains headers is checked.
- Save your changes, then Reset and Run the indexer. Check in Search explorer using Employee_Name:Yesenia Hernandez to confirm that Total_Cost values are displayed.
How to Update Search Type in Azure AI Studio:
- Go to the Chat playground and add index-csv-data as a data source.
- Set the Search type to Semantic, adjust Top to 10-20, and select GPT-4o-mini.
- Try queries such as "What is the total cost for Yesenia Hernandez?" to confirm it sums the Total_Cost field.

Reference :
https://free.blessedness.top/en-us/azure/search/search-how-to-index-csv-blobs
https://free.blessedness.top/en-us/azure/ai-foundry/openai/use-your-data-quickstart?tabs=keyless%2Ctypescript-keyless%2Cpython-new&pivots=ai-foundry-portal
https://free.blessedness.top/en-us/azure/search/semantic-search-overview

Kindly let us know if the above helps or you need further assistance on this issue.

Please "upvote" if the information helped you. This will help us and others in the community as well.

Sukesh 0 Reputation points

2025-10-24T03:03:26.4366667+00:00

Hello @Praneeth Maddali
Thank you for the reply.
Unfortunately that did not work.

The demiliter , was already specified in the indexer, so that is not the issue.

When I executed this command in Search Explorer of the Index, "Employee_Name: Will Smith", it displayed all employees having the name "Smith".

So, this could be one reason.

In the chat playground, I selected "Semantic" instead of "Keyword" and when I submit a question, this was the error:

400: An error occurred when calling Azure Cognitive Search: Azure Search Error: 400, message='Server responded with status 400. Error message: {"error":{"code":"","message":"This index must have valid semantic configurations defined before using the 'semanticConfiguration' query parameter.\r\nParameter name: semanticConfiguration"}}', url='https://gptkb-zuu5ns2scnky4.search.windows.net/indexes/index-csv-cdata/docs/search?api-version=2024-03-01-preview' Semantic search with Azure Search failed with configuration=default. Please ensure semantic search is enabled and you have semantic search quota available on your instance.

From the error I understand that Semantic Configuration needs to be done in the Index, but I was unable to figure out what exactly needs to be configured there, which field name to select and how would it behave.
Praneeth Maddali 1,110 Reputation points Microsoft External Staff Moderator

2025-10-24T11:44:59.5566667+00:00
Hi @Sukesh

Thank you for your update and for clarifying the steps you've already taken.

Regarding the error message— "This index must have valid semantic configurations defined before using the 'semanticConfiguration' query parameter."—your CSV was parsed correctly, but your search index is missing a semantic configuration. This setup is necessary to use semantic search features in Azure AI Search.

To resolve this, please follow these steps:

Add a Semantic Configuration to Your Index

In the Azure Portal, go to your Azure AI Search resource and select your index.

Under Semantic configurations, create a new configuration.

Assign your fields as follows:

Title field: A unique or key identifier (e.g., Employee_Name)

Content field: Fields containing main descriptive data (e.g., Order_Description or a summary field)

Keyword field: Any structured fields commonly used for filtering or facet operations (e.g., Order_ID, Month)

Save the semantic configuration.

Reference docs: Configure semantic ranker.

Enable Semantic Search

Ensure your Azure Search service is on a supported tier (Basic or higher).

In the playground, set the search mode to Semantic and use the configuration name you created.

Test Your Queries

Try structured questions like "What is the highest order made by John Doe?" or "What is the total revenue for June 2025?"

The semantic configuration will help the AI interpret and aggregate your results.

Reference:

https://free.blessedness.top/en-us/azure/search/semantic-search-overview

https://free.blessedness.top/en-us/azure/search/search-get-started-semantic?pivots=csharp

Kindly let us know if the above helps or you need further assistance on this issue.

Please "upvote" if the information helped you. This will help us and others in the community as well.

Share via

What is the correct way to use Azure AI Search to search for CSV data

1 answer

Your answer