First page of document not recognized / no text extracted in both Content Understanding and Document Intelligence

Question

First page of document not recognized / no text extracted in both Content Understanding and Document Intelligence

Achraf El Meryny 20

We are experiencing an issue where the first page of our documents is not being recognized, and no text is extracted from it. This happens consistently across multiple modules, including both Azure AI Document Intelligence (prebuilt and/or custom models) and Azure AI Content Understanding.

Observed behavior:

Text extraction starts correctly from page 2 onward.
Page 1 returns empty/no content, as if it were not processed at all.

0 comments

1 answer

Your answer

Answer 1

Hello Achraf El Meryny,

Greetings! Thanks for raising this question in Q&A forum.

From your description, this usually happens when the first page is not properly readable by the OCR engine (for example due to format, image quality, or PDF structure issues), so the service skips it and starts extracting from page 2.

Please try the below steps which typically fix this issue:

Validate the first page content

Open the PDF and try to select text on page 1
- If you cannot select text, it means it is an image and OCR must process it manually

Improve document quality

Ensure the first page is clear (no blur, skew, or rotation)
Use at least 300 DPI if it is scanned
Avoid shadows, glare, or cropped edges

Re-save or re-generate the file

Save the PDF again (Print → Save as PDF)

Or split the first page and test it alone

This helps fix hidden PDF structure issues

Try a different model

- Use **prebuilt-read** or **prebuilt-layout** instead of custom model

   - Layout model often handles page structure better for OCR scenarios

   **Test first page separately**

      - Upload only page 1 to check if extraction works

         - If it still returns empty, the issue is with that specific page content

         **Check service limits / format**

            - Make sure file format and size meet requirements

               - Corrupted or unsupported pages can be skipped by the service

If after these checks the first page is still not processed, this may require Microsoft backend investigation, because consistent skipping across multiple services (as you mentioned) is not expected behavior.

In that case:

Raise an Azure support ticket with sample document
The product team will review logs and confirm if it's a service-side issue

If this answer helps you kindly accept the answer which will help others who have similar questions

Best Regards,

Jerald Felix.