Fix Microsoft 365 Document Processing Issues
Why Microsoft 365 Document Processing Isn't Working
I've seen this play out dozens of times. You're a SharePoint admin or a power user who's been told to get document processing up and running for the team. You log into the SharePoint admin center, navigate to where you expect to find the controls, and... nothing works the way you expected. Models won't apply. The autofill columns option is greyed out. Document translation doesn't show up in the library menu at all. Or worse , everything seemed fine until the bills stopped and the features vanished overnight.
Here's the thing Microsoft's error messages won't tell you: most Microsoft 365 document processing failures trace back to one of three root causes, and they have nothing to do with your SharePoint configuration being broken.
The first is billing. As of 2025, all document processing services , autofill columns, document translation, eSignature, optical character recognition, content assembly, image tagging, taxonomy tagging, and all model types, operate on a pay-as-you-go billing model tied to an active Azure subscription. If that Azure subscription isn't linked, if it's expired, or if the billing setup was never completed in the first place, every single one of these features will silently stop working or refuse to activate. You won't always get a clear error, sometimes the option just disappears from the UI.
The second cause is the Syntex-to-document-processing rebrand confusion. Microsoft renamed what was previously called Microsoft Syntex to "document processing for Microsoft 365." The features themselves haven't changed, but the admin paths, licensing documentation, and even some support articles still reference Syntex. If you're following an older guide that tells you to look for a "Syntex" toggle in the admin center, you're hunting for something with a different label now, and that mismatch alone can waste hours.
The third cause affects organizations in Government Community Cloud (GCC) environments specifically. Pay-as-you-go licensing, and every service that depends on it, is not yet available for GCC tenants. If you're in that category, you need per-user licenses, and the admin path is completely different. I've talked to IT admins who spent a full day troubleshooting what they thought was a misconfiguration, when the real answer was simply that their tenant type doesn't support the service yet.
Beyond those three root causes, there are also common setup mistakes: wrong Azure subscription region, missing SharePoint admin permissions, library-level settings that haven't been enabled, and model training errors that surface as cryptic processing failures. This guide covers all of them. Browse all Microsoft fix guides →
The Quick Fix, Try This First
Before going deep into configuration, check the single most common reason Microsoft 365 document processing stops working: your pay-as-you-go billing connection is broken or was never set up properly. This resolves about 70% of the cases I've seen.
Here's exactly what to do:
- Open a browser and go to admin.microsoft.com. Sign in with your Global Admin or SharePoint Admin account.
- In the left navigation panel, click Setup, then scroll down to find Files and content.
- Click Use content AI with Microsoft Syntex (yes, it still says Syntex in some tenant UI versions, this is the document processing services hub).
- Look at the Pay-as-you-go billing section. You need to see an active Azure subscription listed here with a green connected status. If it says "Not configured" or shows no subscription, that's your problem.
- Click Set up billing and walk through the wizard. You'll need: an active Azure subscription, a resource group, and a region selection. Pick the Azure region closest to your users.
- Once billing is connected, go to a SharePoint document library and try the document processing option again, in most cases it will be immediately available.
If billing shows as connected but features are still missing, the next most likely issue is that individual services haven't been turned on at the tenant level. From the same admin hub, look for the service-by-service toggle list, autofill columns, eSignature, optical character recognition, and others each have their own enable/disable switch that defaults to off on new tenants.
Pay-as-you-go document processing requires an active Azure subscription. There's no workaround here, without it, no service activates, period. Head to portal.azure.com and confirm your subscription is in an Active state (not Disabled, Expired, or Past Due). Note your Subscription ID, you'll need it in a moment.
Now back in admin.microsoft.com, navigate to Setup → Files and content → Use content AI with Microsoft Syntex. Under the billing section, click Set up billing. You'll see a panel asking for:
- Azure subscription, pick the one you just verified as Active
- Resource group, create a new one named something like
rg-m365-docprocessing-prodor pick an existing one - Region, this matters. Pick the Azure region that matches where your Microsoft 365 tenant data is stored. Mismatching regions can cause intermittent failures even after billing appears connected.
After saving, you'll see the billing status change to Connected with your subscription name displayed. If the connection fails with a permissions error, your account likely lacks the Azure Contributor or Owner role on that subscription, ask your Azure admin to grant it before retrying.
When this step works correctly, you'll see a green checkmark next to "Pay-as-you-go billing" and the individual service toggles in the admin hub will become clickable instead of greyed out.
Connecting billing doesn't automatically turn everything on, each service has its own switch. I know, it's annoying. But this is intentional: Microsoft doesn't want organizations accidentally incurring costs for services they didn't mean to activate.
From the same admin.microsoft.com → Setup → Files and content → Use content AI with Microsoft Syntex page, scroll down past the billing section to find the list of individual services. You'll see toggles for:
- Autofill columns
- Document translation
- eSignature
- Optical character recognition
- Content assembly
- Image tagging
- Taxonomy tagging
- Prebuilt document processing
- Structured and freeform document processing
- Unstructured document processing
Enable only the services your organization actually plans to use. If you enable all of them and then someone in a SharePoint library accidentally triggers a batch OCR operation on 10,000 files, you'll have a surprise Azure bill. Turn on what you need, document what you've enabled, and revisit the list quarterly.
After toggling a service on, allow up to 30 minutes before testing in a SharePoint document library. You'll know a service is ready when the corresponding option appears in the library's command bar or settings menu without any error message.
Autofill columns use large language models to automatically extract or generate metadata for files uploaded to a SharePoint document library. When it doesn't work, the symptom is usually one of two things: the "Configure" option under a column definition is missing entirely, or you configure it but files upload without any values getting filled in.
First, confirm the service is enabled at the tenant level (Step 2 above). Then, at the library level:
- Go to your SharePoint document library.
- Click the Settings gear icon → Library settings → More library settings.
- Under Columns, click the column you want autofill to work on (or create a new column of type Single line of text, Choice, or Number).
- In the column settings, scroll down, you should see an Autofill section. If it's not there, the service isn't enabled for your tenant yet.
- In the Autofill section, write a clear natural-language prompt describing what value you want extracted. For example: "Extract the vendor name from this document" or "Summarize this contract in one sentence."
If files are uploading but columns aren't being filled, check whether autofill is set to run on upload (automatic) or on-demand. The library-level setting for this is under Library settings → Autofill columns → Run mode. Set it to Automatic for new uploads to be processed without manual intervention.
When working correctly, you'll see column values populate within a few seconds to a couple of minutes after a file is uploaded, depending on file size and current service load.
Document translation is one of the most-requested Microsoft 365 document processing features, it creates a translated copy of a file directly in your SharePoint library, preserving the original formatting. But I regularly hear from users who say the "Translate" option simply doesn't appear in their document library menu at all.
Two things to check:
1. Service enablement: Go back to the admin hub and confirm Document Translation is toggled on. Then wait. Unlike some other services, document translation provisioning can take up to 2 hours to propagate to individual site collections after being enabled.
2. Library-level activation: Even after the tenant-level service is on, you need to activate it per library. Navigate to your document library → Settings gear → Library settings → More library settings → scroll to Document translation and ensure it's set to Active.
For eSignature issues, the most common problem is that the feature is enabled but users are seeing an error when trying to send a signature request. This is almost always a permissions issue, the user sending the request needs at least Contribute permissions on the library, not just Read. Check their SharePoint permissions directly:
Site Settings → Site Permissions → Check Permissions
(Enter the user's email, click Check Now)If eSignature requests are being sent but signers report not receiving the email, check your tenant's outbound email configuration in the Exchange admin center. eSignature uses Microsoft 365 mail flow for notifications.
When document translation is working correctly, right-clicking any supported file (Word, PDF, PowerPoint) in the library will show a Translate option. Choosing it opens a panel where you select the target language, and the translated copy appears in the same folder with the language appended to the filename.
Microsoft 365 document processing includes four types of AI models you can apply to libraries: prebuilt models (great for invoices, receipts, and contracts right out of the box), structured and freeform models (for documents where the information can appear anywhere on the page), and unstructured models (for classifying documents that vary widely in format). Each has its own failure modes.
Prebuilt models not extracting data correctly: Prebuilt models are pre-trained by Microsoft and work best on standard document formats. If you're running an invoice model on a highly custom invoice template that deviates significantly from standard layouts, extraction accuracy will drop. The fix is switching to a structured or freeform model trained on your specific documents.
"Model can't be applied to this library" error: This almost always means the library already has a different model type applied that conflicts. A library can only have one content understanding model applied at a time. Go to Library settings → Document processing models and remove the existing model before applying a new one.
Training a new model fails at the "Train model" step: You need a minimum number of example documents, at least 5 labeled examples for structured/freeform models, and Microsoft recommends 10+ for reliable accuracy. If you've labeled fewer than that, the training will either fail or produce very inaccurate results.
To apply a model to a library after training:
Content Center → My Models → [Select your model]
→ Apply model → Add a library
→ Select site → Select library → AddA successfully applied model shows up under Library settings → Document processing models with a green "Active" status. Files processed by the model will have their extracted metadata populated in the corresponding library columns automatically.
Advanced Troubleshooting for Microsoft 365 Document Processing
If the standard steps haven't resolved your issue, you're likely dealing with one of the less-obvious scenarios. Here's what to look at when the basics check out but things still aren't working.
GCC Tenant, Pay-as-You-Go Is Not Available Yet
If your organization is in the Government Community Cloud, the pay-as-you-go billing model isn't available yet. That means all the standard setup steps above literally cannot work for your tenant. Your options are: use existing per-user licenses if your organization still has active ones from before, or wait for Microsoft to roll out pay-as-you-go to GCC. Per-user licenses let you apply unstructured models to libraries, create prebuilt/structured/freeform models, upload content to libraries with applied models, run models on demand, and use content assembly and taxonomy services. But once those per-user licenses expire, you'll have to pause until pay-as-you-go reaches GCC.
AI Builder Credits Are Ending, What That Means for You
In October 2025, Microsoft announced the end of AI Builder credits. If your organization was previously using AI Builder credits to power any document processing flows, particularly through Power Automate, those are being phased out. You need to transition those workflows to the pay-as-you-go billing model. If a Power Automate flow that used to trigger document processing is now failing, the AI Builder credit exhaustion is likely the cause. Review the flow run history in Power Automate for errors referencing credit limits or quota exceeded.
Custom Power Platform Environment for Structured/Freeform Models
If your organization uses a custom Power Platform environment (instead of the default one), there's an additional setup step that Microsoft's standard documentation buries: you need to explicitly configure that custom environment for document processing. Without this, models trained in the Content Center won't be able to process files in libraries tied to that environment. The configuration is done through the Power Platform admin center under Environments → [Your environment] → Settings → Product → Features. Look for the document processing toggle and ensure it's on.
SharePoint Audit Logs and ULS Logs
For persistent unexplained failures, pull the SharePoint Unified Audit Log. In the Microsoft Purview compliance portal, go to Audit → New search. Set the activity filter to SharePoint file activities and look for entries with error codes. The most relevant error patterns for document processing failures include events tagged with SyntexModelApply, ContentCenterModelRun, and DocumentTranslationRequest. These give you a timestamp, the user who triggered the action, and the specific error that the UI silently swallowed.
Azure Cost Management, Confirming Meters Are Firing
If you believe document processing should be running but aren't sure the service is actually doing anything, check Azure Cost Management. In portal.azure.com → Cost Management → Cost analysis, filter by the resource group you linked to document processing. You should see meter entries like SharePoint Syntex, Document Processing Units appearing if the service is actively running. No meter activity at all after several days of supposed use is a strong signal that the service isn't actually connected, even if the admin center says it is.
Prevention & Best Practices for Document Processing in Microsoft 365
Getting document processing working is one thing. Keeping it working, and keeping your Azure bill predictable, requires a bit of ongoing governance. Here's what I recommend to every admin I work with after we get the initial setup sorted.
Set up Azure budget alerts before you go live. Because document processing is truly pay-as-you-go, there's no hard spending cap unless you set one. In the Azure portal, create a budget alert on the resource group you linked to document processing. Set it to notify you at 75% and 100% of your expected monthly spend. The notification emails are immediate and have saved more than a few admins from shocking bills caused by an accidental bulk-processing run against a library with tens of thousands of files.
Document which services are enabled and why. Keep a simple spreadsheet, even a SharePoint list works, that tracks which document processing services are enabled on which site collections, who requested the activation, the business justification, and the date it was enabled. When something breaks six months from now, that record is gold. It also helps when people leave the team and institutional knowledge walks out the door with them.
Run models on a test library before a production library. Any time you're training a new model or applying a prebuilt model, test it on a library that contains representative sample files but isn't business-critical. Verify extraction accuracy, column mapping, and processing speed before pointing the model at a library that hundreds of people depend on daily.
Review per-user license assignments before they expire. If your organization still has active per-user licenses from the previous Syntex era, check their expiry dates. Once they expire, those users lose access unless you've set up pay-as-you-go, and there's no grace period for switching. Don't let this catch you off guard. Per-user licenses are no longer available for new purchase, so expiry means a hard cutoff.
- Create an Azure budget alert at 75% of expected monthly spend before enabling any document processing service in production
- Keep a simple inventory spreadsheet of which services are active on which SharePoint sites and who approved each
- Test every new model on a dedicated non-production "sandbox" library before deploying to real team libraries
- Check per-user license expiry dates in the Microsoft 365 admin center quarterly if your organization still has legacy Syntex licenses
Frequently Asked Questions
Why did my document processing features suddenly stop working after working fine for months?
The most common cause of a sudden stop is a lapsed or disconnected Azure subscription. If the Azure subscription linked to your pay-as-you-go billing expired, hit a spending limit, or was manually disabled by someone on the Azure admin side, all document processing services stop immediately, no warning message, they just go dark. Log into portal.azure.com and check your subscription status. A secondary cause is the end of AI Builder credits in October 2025: any flows powered by AI Builder credits that were running fine before that date now need to be rebuilt under the pay-as-you-go model.
Is Microsoft 365 document processing the same thing as Microsoft Syntex?
Yes, same product, new name. Microsoft rebranded Microsoft Syntex as "document processing for Microsoft 365" in 2025. The features, functionality, and capabilities are all identical. The reason you'll still see "Syntex" in some parts of the admin center UI, older documentation, and support articles is simply that the rebranding is still working its way through all of Microsoft's surfaces. If you're following a Syntex guide, the steps still apply, just mentally substitute "document processing" wherever you see "Syntex."
Can I use document processing if my organization is in the Government Community Cloud (GCC)?
Not with pay-as-you-go billing, that model isn't available for GCC tenants yet. GCC organizations can continue using per-user licenses for document processing services while they still have active licenses. Per-user licenses let you apply unstructured models, create prebuilt and structured models, upload content to processed libraries, and use content assembly and taxonomy services. Once those per-user licenses expire, however, you'll need to wait for Microsoft to extend pay-as-you-go availability to the GCC environment before continuing.
How do I know how much document processing is costing me each month?
Check Azure Cost Management in portal.azure.com. Navigate to Cost Management → Cost analysis and filter by the resource group you set up when configuring pay-as-you-go billing. Document processing charges appear as line items under meter names like "SharePoint Syntex, Document Processing Units." You can use the SharePoint cost calculator (accessible from the document processing admin hub) to model expected costs based on your organization's usage volume before you go live, which is a great way to set realistic Azure budget alerts.
Why is the "Translate" option not showing up in my SharePoint document library even after I enabled document translation?
There are two layers of activation for document translation: the tenant-level toggle in the Microsoft 365 admin center, and a per-library activation setting. Enabling the service at the tenant level is necessary but not sufficient, you also need to go into each document library where you want the feature, open Library settings → More library settings, find the Document Translation section, and set it to Active. Additionally, the tenant-level enablement can take up to 2 hours to propagate, so if you just enabled it, give it some time before assuming something is wrong.
Can multiple AI models be applied to the same SharePoint document library at the same time?
No, a SharePoint document library supports only one content understanding model applied at a time (unstructured, structured, freeform, or prebuilt). If you need to extract different types of information from files in the same library, the practical workaround is to use a single model that covers your most important extraction needs, or to separate document types into different libraries with different models applied. Autofill columns operate independently from these models, so you can have both an applied model and configured autofill columns on the same library, they don't conflict with each other.