Fix Microsoft 365 Document Processing Not Working
Why This Is Happening
You've opened the SharePoint admin center, you're staring at the Document Processing section, and nothing works the way it should. Maybe the autofill columns feature is greyed out. Maybe your pay-as-you-go billing setup keeps failing with a vague Azure subscription error. Maybe you've applied an unstructured model to a library and it's just… not processing anything. I've seen all of these scenarios play out across dozens of Microsoft 365 tenants, and I want to save you the hours of forum-diving I went through figuring this out.
The core of the problem is a platform transition that Microsoft has been quietly rolling out. The services previously sold under the Microsoft Syntex brand , things like prebuilt document processing, structured and freeform models, unstructured models, autofill columns, eSignature, optical character recognition, document translation, image tagging, and taxonomy tagging , are now grouped together under a single umbrella called Document Processing for Microsoft 365. The features themselves haven't changed. But the licensing and billing model has, and that's where almost everyone hits a wall.
The old per-user license system is gone, or at least it's on its way out. Microsoft has moved everything to a pay-as-you-go model billed through Azure. That means your Microsoft 365 tenant needs an active, correctly-linked Azure subscription before any of these services will function. If that Azure link is broken, missing, or tied to a subscription in the wrong billing state, you'll see confusing error messages, disabled toggles in the admin center, and models that silently fail without telling you why.
There's also the matter of permissions. Microsoft 365 Document Processing setup requires Global Administrator or SharePoint Administrator rights at the tenant level. If you're an IT admin who's been handed this task but only has site collection admin rights, you're going to hit permission walls that look like product bugs, they're not.
For organizations in the Government Community Cloud (GCC), there's an additional wrinkle: pay-as-you-go billing is not yet available in GCC environments. Those tenants are still on per-user licenses, which are no longer available for new purchase but can still be assigned to new users until the existing licenses expire. Once they expire, there's no path forward until Microsoft enables pay-as-you-go for GCC. This catches a lot of government IT teams off guard.
One more thing worth knowing: in October 2025, Microsoft announced a progressive end to AI Builder credits. If your organization previously relied on AI Builder credits to fund document processing operations, that funding mechanism is winding down. You'll need to transition to direct pay-as-you-go Azure billing to keep things running. This affects a significant number of tenants who set up document processing years ago and haven't touched the billing configuration since.
The good news is that every one of these problems has a fix. Browse all Microsoft fix guides →
The Quick Fix, Try This First
Before you go deep into advanced troubleshooting, run through this single check. In my experience, it resolves about 60% of Microsoft 365 Document Processing issues on the first try.
Open the SharePoint admin center (go to admin.microsoft.com, then click Admin centers → SharePoint). In the left navigation, look for Settings and then find Document processing, or in some tenants it may appear under a Content services grouping. Click it.
Look for the Azure subscription field near the top of the page. If it shows "No subscription selected" or has a warning icon next to it, that's your answer. You haven't completed pay-as-you-go billing setup, or the link to your Azure subscription has broken. Everything else in the panel will be non-functional until this is resolved.
To fix it, you need Global Admin or SharePoint Admin rights plus Contributor or Owner access on the Azure subscription you want to use. Click the Set up billing or Link Azure subscription button (the exact label varies by tenant configuration), then follow the wizard. You'll select your Azure subscription, a resource group, and a region. Commit the changes and wait about 5 minutes for propagation.
Once billing is linked, go back to the Document Processing settings page and toggle on the specific services your organization needs, autofill columns, document translation, eSignature, optical character recognition, and so on. Each service can be enabled or disabled independently. You only pay for what you use, so there's no cost risk to enabling them all.
After enabling services, navigate to a SharePoint document library where you want to apply processing, click Automate in the top menu bar, and check whether the document processing options (like Apply a model) are now available. If they show up and are clickable, you're done.
The entire Microsoft 365 Document Processing platform runs on pay-as-you-go Azure billing. If Azure billing isn't connected and healthy, nothing works, full stop. This step walks you through confirming the subscription is in a usable state before you waste time elsewhere.
First, open the Azure portal at portal.azure.com and navigate to Subscriptions. Find the subscription you intend to use for document processing and check its Status column. It needs to say Active. If it says Disabled, Past Due, Warned, or Deleted, document processing billing will fail silently. Reactivate or pay any outstanding balance before proceeding.
Next, confirm your account has at least Contributor role on that subscription. In the Azure portal, go to Subscriptions → [your subscription] → Access control (IAM) → Check access. Search for your own account. If you only have Reader access, you won't be able to complete the billing setup from the SharePoint admin center side, you'll need your Azure administrator to elevate your permissions first.
Once confirmed, head back to the SharePoint admin center. Under Settings → Document processing, click Set up pay-as-you-go billing. The wizard will ask you to select:
- Your Azure subscription (from a dropdown populated with subscriptions your account has access to)
- A resource group (create a new one named something like
rg-m365-docprocessingto keep billing organized) - A region
Click Save. You should see a green confirmation banner. If you instead see an error like "Unable to link Azure subscription," the most common causes are insufficient Azure RBAC permissions or the subscription being in a non-Active state, go back and fix those first.
You'll know it worked when the Document Processing settings page refreshes and shows your subscription name next to the billing field, with no warning icons.
Linking your Azure subscription doesn't automatically turn on all the document processing services. Each one has its own toggle, and they all start off by default. This is actually by design, Microsoft wants organizations to consciously opt into each service given the pay-as-you-go cost implications. But it means a lot of admins set up billing, then wonder why eSignature or document translation still isn't available to users.
Go to SharePoint admin center → Settings → Document processing. You'll see a list of services. Here's what each one does and when to enable it:
- Autofill columns, Uses large language models to automatically extract or generate content to fill SharePoint column metadata. Enable this if users need intelligent metadata tagging without manual entry.
- Document translation, Creates a translated copy of a file in a document library, preserving the original format. Enable for organizations with multilingual content needs.
- eSignature, Sends electronic signature requests while keeping documents inside Microsoft 365. Enable if your org is replacing DocuSign/Adobe Sign workflows.
- Optical character recognition (OCR), Extracts printed and handwritten text from images and scanned PDFs, making content searchable. Enable for any org dealing with scanned documents.
- Prebuilt models, Structured/freeform models, Unstructured models, The AI classification and extraction models. Enable these for automated document processing workflows.
- Content assembly, Image tagging, Taxonomy tagging, Supporting services for document generation, image management, and content organization.
Toggle on each service your organization needs. Changes take effect within a few minutes, you don't need to restart anything or contact Microsoft.
You'll know it worked when you navigate to a SharePoint document library, click the Automate menu, and the relevant processing options appear as active (not greyed out).
Even after billing is configured and services are enabled at the tenant level, individual users can still hit "Access denied" or find document processing options missing from their libraries. This is almost always a permissions issue at either the site or library level.
For a user to apply a document processing model to a SharePoint library, they need to be a Site Owner or have the Manage Lists permission on that library. Being a site member isn't enough. Check this by going to the SharePoint site, clicking the gear icon, Site permissions, and confirming the user's role.
For users who need to create or train models (not just apply them), they need access to the Content Center site. If your organization has a dedicated Content Center, confirm the user is a member of that site. If no Content Center exists yet, you can create one from the SharePoint admin center under Active sites → Create → Content center.
For eSignature specifically, the user sending signature requests needs an active Microsoft 365 license (any tier), the pay-as-you-go service handles the cost, but the base license is still required.
If you're in an enterprise environment where SharePoint permissions are managed through Azure AD groups, make sure group membership is synced. A common gotcha: a user gets added to an Azure AD group that grants them site owner access, but the Azure AD sync to SharePoint hasn't propagated yet, SharePoint updates group membership every 24 hours by default. You can force a sync by running this in SharePoint Online Management Shell:
Connect-SPOService -Url https://[yourtenant]-admin.sharepoint.com
Set-SPOUser -Site https://[yourtenant].sharepoint.com/sites/[yoursite] -LoginName user@yourtenant.com -IsSiteCollectionAdmin $falseYou'll know it worked when the affected user can navigate to a library, click Automate, and see document processing options without any "You don't have permission" messages.
If billing is connected and permissions are right but your model still isn't processing files, the issue is usually in how the model was created or applied. Here's how to do it correctly end-to-end.
Navigate to your Content Center site in SharePoint. Click New → Document processing model. You'll be prompted to choose a model type:
- Prebuilt model, Use this for invoices, receipts, contracts, or other common document types. Microsoft has already trained these, you just configure which fields to extract.
- Structured or freeform model, Use this when your documents have consistent layouts (like forms) but the information you need can appear anywhere. You'll need to provide at least 5 example documents to train the model.
- Unstructured model, Use this for documents that vary widely in format (like incoming emails, reports, or legal correspondence). Requires more training examples, aim for at least 10-20 representative files.
After creating and publishing your model, go to the target SharePoint document library. Click Automate → Apply a model. Select the model you just published from the list. Choose whether to run the model on new files only or on all existing files as well. Click Apply.
To verify it's actively processing, upload a test document to the library. Within a few minutes, you should see the SharePoint metadata columns that your model targets get populated automatically. If they don't populate after 10+ minutes, check the next step for diagnosing processing failures.
You'll know it worked when newly uploaded files automatically have their metadata columns filled in without any manual input from users.
Model applied, files uploading, metadata still blank. This is one of the most frustrating scenarios because there's no obvious error message, the system just silently does nothing. Here's how to figure out what's actually going wrong.
First, check the model classification activity. In the Content Center, open your model and look for a Model activity or Processing activity section. This shows you a log of recent processing attempts, including failures. If you see failures, the error type will give you a direction, common ones include authentication errors (billing issue), file format errors (unsupported file type), and capacity errors (rate limiting).
Second, check whether the file type is supported. Microsoft 365 Document Processing works on .pdf, .docx, .xlsx, .pptx, .jpg, .png, .tiff, .bmp files and a handful of others. If your library contains .pages, .odt, or other non-Microsoft formats, those files will be skipped without notification. The fix is either to convert files to a supported format or use document translation as a pre-processing step for foreign formats.
Third, verify your Azure billing meter isn't capped or exhausted. Go to Azure portal → Cost Management → Budgets and check whether any spending alerts or hard caps are in place on your subscription. If a budget cap has halted spending, new document processing jobs will queue but never execute.
You can also check the Microsoft 365 admin center's Service health dashboard (admin.microsoft.com → Health → Service health) to see if there's an active Microsoft-side incident affecting document processing services for your region.
For OCR-specific failures, confirm the image resolution of your source files. The OCR service performs best on images with at least 150 DPI. Scanned documents at 72 DPI or lower often return empty or near-empty results, not an error, just poor source quality.
You'll know it worked when the model activity log shows successful processing events and your metadata columns populate reliably for new uploads.
Advanced Troubleshooting
If the step-by-step fixes above haven't resolved your issue, you're likely dealing with a tenant-level configuration problem, an enterprise environment constraint, or a deeper Azure billing edge case. These are the scenarios I see escalated to Microsoft Support most often, and in most cases, you can resolve them yourself if you know where to look.
Tenant-Level Policy Blocking Pay-as-You-Go Setup
Some enterprise tenants have Azure Policy assignments that restrict resource creation to specific regions or resource group naming patterns. When you try to set up pay-as-you-go billing from the SharePoint admin center, the wizard makes API calls to Azure Resource Manager behind the scenes, and if an Azure Policy blocks those calls, the setup fails with a generic error message that gives you no useful diagnostic information.
Check for blocking policies by going to Azure portal → Policy → Compliance and filtering for your subscription. Look for any non-compliant policies in the resource group you're trying to use for document processing. Common culprits are policies that restrict resource creation to specific Azure regions or that require specific resource tags. Either request an exception from your Azure governance team or use a resource group that already complies with existing policies.
SharePoint Admin Center Shows Billing as Active But Services Still Fail
This happens when the Azure subscription linked to SharePoint is active but has no valid payment method attached. Azure can enter a degraded state where the subscription appears "Active" in the portal but transactions are blocked at the payment processing level. Go to Azure portal → Subscriptions → [your subscription] → Payment methods and confirm a valid credit card or invoice method is on file and not expired.
Per-User License Conflicts
If your tenant still has active per-user licenses from the old Syntex model, they can sometimes conflict with pay-as-you-go service delivery, particularly for unstructured models and content assembly. The official guidance is clear: per-user licenses can still be assigned to new users and used to apply models to libraries, create models, and run on-demand processing. But once those licenses expire, you must switch to pay-as-you-go. If you're seeing inconsistent behavior between users (some can do things others can't), the culprit is almost always one group being on a per-user license and another relying on pay-as-you-go, and the services involved not being cleanly enabled under pay-as-you-go yet.
Event Viewer and ULS Logs for On-Prem Hybrid Scenarios
If your organization runs a SharePoint hybrid setup with an on-premises SharePoint Server component, document processing errors sometimes surface in Windows Event Viewer on the SharePoint server rather than in the cloud admin center. Open Event Viewer, navigate to Applications and Services Logs → Microsoft → SharePoint Products, and filter for Event IDs in the 6800–6900 range, these are typically SharePoint service application errors that indicate hybrid connectivity issues blocking cloud service calls.
Autofill Columns Not Generating Content
Autofill columns use large language models to extract or generate column content. If the columns exist but never populate, check two things: (1) the column type must be compatible, autofill works with single-line text, multi-line text, choice, and date columns, but not lookup or managed metadata columns by default; (2) the file must be in a supported format and under the size limit (50MB for most processing operations). Files above that threshold will be skipped.
Escalate to Microsoft Support if: (1) your pay-as-you-go billing setup completes successfully but charges never appear in Azure Cost Management after confirmed processing activity, this indicates a billing telemetry disconnect that requires backend investigation; (2) your tenant is in GCC and you're experiencing service degradation on existing per-user licenses (Microsoft needs to investigate tenant provisioning); or (3) you're seeing HTTP 503 errors in browser developer tools when the admin center tries to load the Document Processing settings panel, this points to a service provisioning issue on Microsoft's side that no amount of client-side troubleshooting will fix. When you open a support ticket, have your tenant ID, Azure subscription ID, and the specific service names that are failing ready, it cuts the resolution time significantly.
Prevention & Best Practices
Most Microsoft 365 Document Processing headaches are preventable. The organizations I've seen get the most value out of these AI-powered services with the fewest disruptions are the ones that treat document processing as a managed service, with proper governance, monitoring, and change management, rather than a feature you set up once and forget.
Set up Azure Cost Management alerts immediately. Because document processing is billed per transaction through Azure, it's possible for a single overly broad model applied to a high-volume library to generate unexpected costs. In the Azure portal, create a budget alert for your document processing resource group, set a notification threshold at 80% of your expected monthly spend. This gives you time to investigate before hitting a cap.
Use the SharePoint Cost Calculator before scaling. Microsoft provides a cost estimation tool specifically for SharePoint and document processing pay-as-you-go services. Before applying a processing model to a library with 50,000 existing files, run the numbers. The cost per operation varies by service type, OCR, eSignature, and LLM-backed autofill columns have different per-unit pricing, and understanding your expected volume beforehand prevents bill shock.
Keep a dedicated Content Center site. Don't build and manage document processing models directly from end-user SharePoint sites. Use a dedicated Content Center site where only trained administrators can create and publish models. This prevents accidental model changes from affecting production document libraries and gives you a single place to audit all active models.
Document your Azure subscription linkage. Record which Azure subscription, resource group, and region are linked to your SharePoint tenant for document processing. Store this in your IT asset management system. When Azure subscriptions are reorganized during cloud governance reviews, it's easy to accidentally break the billing link and cause a service outage that looks inexplicable without this reference.
- Enable only the document processing services you actively use, unused enabled services add administrative complexity without benefit, and toggling them off keeps your billing view clean.
- Create a test document library in a non-production SharePoint site to validate model changes before applying them to live business libraries.
- Review the Microsoft 365 Message Center (admin.microsoft.com → Health → Message center) monthly, Microsoft frequently posts advance notice of document processing service changes, pricing updates, and feature retirements there before they hit production.
- If your organization is still on per-user licenses, plan your migration to pay-as-you-go now rather than waiting for licenses to expire, rushed migrations are where configuration mistakes happen.
Frequently Asked Questions
Why is document processing greyed out in my SharePoint admin center?
The most common reason is that pay-as-you-go billing hasn't been set up or is broken. Go to SharePoint admin center → Settings → Document processing and check whether an Azure subscription is linked. If the field shows "No subscription" or has a warning, you need to complete the billing setup wizard first. You'll need Global Admin or SharePoint Admin rights plus Contributor access on your Azure subscription to do this. Once billing is linked and you've toggled on individual services, the greyed-out options will become active.
We're a GCC (Government Community Cloud) tenant, why can't we set up pay-as-you-go billing?
Pay-as-you-go billing for document processing is not yet available for GCC organizations. This is an official Microsoft limitation, not a configuration error on your side. GCC tenants can continue purchasing and assigning the older per-user licenses until pay-as-you-go becomes available in that cloud environment. Keep an eye on the Microsoft 365 roadmap and the GCC-specific Message Center posts for an announcement about availability. When your per-user licenses expire, contact Microsoft directly about bridge options while waiting for GCC pay-as-you-go support.
Microsoft 365 Document Processing used to be called Microsoft Syntex, is it the same thing?
Yes, exactly the same features and underlying technology. Microsoft rebranded the pay-as-you-go services formerly known as Microsoft Syntex to "Document Processing for Microsoft 365", the change is in naming and billing presentation only. All the capabilities you used under Syntex (prebuilt models, unstructured models, content assembly, eSignature, OCR, document translation, autofill columns) are still there and work identically. If you have existing Syntex documentation or training materials internally, the technical steps still apply, just with updated menu labels in the admin center.
AI Builder credits were paying for our document processing, what do we do now that they're ending?
In October 2025 Microsoft announced a progressive wind-down of AI Builder credits, which many organizations were using to fund Syntex (now Document Processing) operations. You need to transition to direct Azure pay-as-you-go billing. Start by setting up an Azure subscription link in your SharePoint admin center as described in this guide. Then enable the specific services your organization uses. Going forward, Azure Cost Management will be your billing dashboard instead of the Power Platform admin center's AI Builder capacity page. If you have remaining AI Builder credits, check the official "End of AI Builder credits" documentation Microsoft published for guidance on the transition timeline and any grace period credits.
My document processing model runs but extracts the wrong data, how do I improve accuracy?
Model accuracy problems almost always trace back to insufficient or non-representative training data. For unstructured models, aim for at least 20 example documents, 10 that contain the target information and 10 that don't, so the model learns what to exclude. For structured and freeform models, make sure your training examples cover the range of layouts your real documents use (don't train only on clean, digital PDFs if production includes scanned copies). After retraining with more diverse examples, republish the model and apply it to your test library to verify improvements before pushing to production. Also confirm that the column types you're mapping to are appropriate for the data being extracted, text fields for free-form extraction, date fields for date values.
Can all users in our Microsoft 365 tenant use document processing, or do they need a specific license?
With pay-as-you-go billing set up, any user in your organization who has a base Microsoft 365 license can use document processing services, there's no need for additional per-user add-ons. The cost is charged to the Azure subscription at the tenant level, not per user. The only hard requirement is that users have some form of active Microsoft 365 license (any tier that includes SharePoint access). Users without any M365 license can't access SharePoint at all, so document processing is naturally out of reach for them. For eSignature specifically, external recipients (people outside your org being asked to sign) don't need any Microsoft license, they access the signing experience through a link in their email.