Answers to frequent problems on the Docupath platform. Use this article to find the cause of an issue and the steps that correct it.

Find the section for the feature that shows the problem, then read the question that matches your problem, and follow the steps in the answer in the given sequence. Some answers contain a colored note. A Caution or Warning note tells you about an action you cannot undo. An Info note gives useful context. If the steps do not correct the problem, send a support request (see the Help and Support section below).

Problems that occur when you send documents to Docupath by drag & drop, email, API, or the mobile app.

Q: Which file types can I upload?

Docupath accepts these file types:

Docupath rejects video files, audio files, compressed files (ZIP, RAR, 7Z), executable files, and database files at upload. Searchable PDF is the recommended format. It gives the highest accuracy and the fastest processing.

Q: The platform rejected my file at upload. Why?

The platform rejects a file when the file breaks a limit or has an unsupported format. Do these checks:

Info: If one document in a batch breaks a limit, the full batch upload can fail. Remove that document and upload the batch again. A batch can contain a maximum of 100 documents.

Q: I sent documents by email, but they did not arrive. What do I check?

Info: The subject line and the email body are not used for routing. Only the document content decides the routing.

Q: A document shows "No sub organization detected". What does this mean?

The system could not find exactly one sub-organization for the document. For an email to a Main Organization address, Docupath applies these rules in sequence. It stops at the first rule that finds one sub-organization:

If no rule finds one sub-organization, or more than one sub-organization matches, the document stays in the Main Organization inbox. To correct it:

Info: A change to an override trading party applies only to future emails. Documents that are already routed keep their assignment.

Q: I sent the same email two times. Now I have two documents.

Docupath does not remove duplicate emails at ingestion. Each email creates new documents. Delete the extra document manually. To catch duplicates automatically, configure Duplicate Document Handling in System Settings (see the System Settings section below).

Q: A document named "Email" appeared. Where did it come from?

The email had no valid business-document attachment. Docupath then processed the email body as the document. This occurs only on the "Single Primary Document with Optional Attachments" address. Attachments with little or no readable text, such as logos and signature images, are dropped.

Q: My API upload failed. What do the error codes mean?

Caution: Docupath does not retry failed API uploads automatically. Add retry logic to your external system.

Problems that occur while the AI Model Garden extracts data from your documents in the upload queue.

Q: How long does processing usually take?

At peak load, processing times can increase by 50 to 100 percent. Documents marked Urgent are processed with priority. Admins and Managers can see the queue depth on the Organization Dashboard.

Q: My document stays in "Processing" status. What do I do?

Wait a maximum of 1 hour. If a document stays in "Processing" for more than 1 hour, the system automatically sets its status to "Failed". The document then becomes eligible for reprocessing. A manual step is not necessary before that.

Q: My document has "Failed" status. What are the causes?

Common causes are:

Export the file again from the source application, or scan it again with better quality. Then upload it again, or reprocess it from the Pending tab.

Info: The platform shows limited error detail to end users. Contact support for a detailed failure analysis.

Q: Extraction accuracy is low on my scanned documents. How do I improve it?

Info: Handwritten text gives approximately 70 to 80 percent accuracy. Tables with merged cells or nested tables, and multi-column layouts, can extract incorrectly.

Q: Only a part of my Excel data was captured. Why?

Docupath processes only the first (active) worksheet of a spreadsheet. It extracts a maximum of 500 rows as line items. Extra rows are cut off. Formulas are calculated, and only the result values are captured. The formula text is not kept.

Q: Do I need templates for new document layouts?

No. The AI Model Garden selects the correct models for each document based on its type, complexity, language, and quality. Your Instruction Builds are applied at processing time, without model retraining.

Info: A change to an Instruction Build takes effect on the next document that is processed. Reprocess older documents to apply the change to them.

Problems in the Document Queue and the Dual-Pane Review Screen, where you check and approve extracted data.

Q: I cannot find a document in the Document Queue.

Q: My manual corrections disappeared from a document.

The document was reprocessed. Reprocessing resets all manual edits. This includes corrected field values and manually assigned trading parties.

Caution: Reprocessing has no undo. Download or export important documents before you reprocess them, especially with bulk reprocess.

Q: A document opens as read-only.

A different user has the document open. A document lock prevents edits by two users at the same time. Wait until the other user closes the document, or agree on review assignments per document.

Q: I cannot reject a document.

You need the Reviewer or Reviewer+Validator role. Ask an administrator to assign the role to you. Each rejection needs a documented reason. The reason is kept for the audit trail.

Q: Can I approve a rejected document again?

No. Rejection is a terminal state. You cannot reprocess a document directly from the Rejected tab. Upload the document again and let it go through the full lifecycle.

Q: The Validated tab is missing.

The Validated tab shows only when at least one user has the Validator role. A sub-organization may not see the tab when validators exist only at the parent organization.

Q: The URG or DUP label does not show on documents.

These labels are add-on features. Make sure the add-on is active for your tenant.

Q: A document shows "Sent" but not "Acknowledged". What does this mean?

Docupath sent the document to the downstream system, but that system did not confirm receipt in the expected time. The document shows "Not Acknowledged". Examine the connection and the logs of the downstream system.

Q: What are the limits for bulk actions?

Caution: Bulk delete is permanent. Bulk approvals and bulk rejections cannot be undone. Read the confirmation dialog before you continue.

Q: A field covers several pages, but the highlight shows only the first page.

This is a known limitation. The bounding box shows only the first occurrence of the field. Check the other pages manually. On very large documents (more than 100 pages), navigation and highlighting can also become slow.

Problems with the two-level structure of parent organizations and sub-organizations, and with document visibility.

Q: A user cannot see documents that other users can see.

Self-service access is probably enabled. With self-service access:

Map the user to the correct organization in User Management. Then test the access with a sample document.

Q: I cannot change the email address of a sub-organization.

This is by design. The email address is set at creation and cannot be edited afterwards. The front part of the address can contain only letters and numbers, without dashes. Choose the front part carefully before you save. If the address must change, create a new sub-organization.

Q: Can I create a sub-organization inside a sub-organization?

No. Docupath supports two levels only: a parent organization and its sub-organizations. Plan your structure with one parent level and one child level. The limits are 100 parent organizations per tenant and 100 sub-organizations per parent.

Q: All documents from an organization are marked urgent. Why?

Urgent marking is enabled at the parent organization. Sub-organizations inherit this setting. To use urgency selectively, disable the setting at the parent. Then enable it per sub-organization, or mark single documents as urgent at upload.

Q: I deleted a sub-organization. What happens to its documents and settings?

Deletion is a soft delete. The sub-organization becomes inactive, and its documents stay accessible. New documents can no longer route to it.

Caution: A deleted sub-organization cannot be enabled again. It is fully removed from the tenant only after its trading parties, instructions, and rules are also deleted.

Q: Where do I create organization-scoped rules?

Transformation rules and rejection rules with the "Organization specific" scope are created at the sub-organization level. Each sub-organization has its own set. Only Instruction Builds are scoped at the organization level. Users often search at the wrong level; document this clearly for your team.

Problems with the master records for suppliers, buyers, and counterparties, and with party matching on documents.

Q: A document shows "No Trading Parties Found".

Docupath could not match the extracted text to a known trading party. Do these steps in the Review Screen:

To prevent this in the future, add alternate names (aliases) to the party profile. Each party supports a maximum of 20 alternate names.

Q: Docupath matched the wrong trading party.

Two parties probably share an overlapping alternate name. Matching uses the alternate names first, then the party name. Small differences in case and spacing are handled automatically.

Q: Party data does not enrich my documents.

The trading party must exist under the sub-organization that receives the document. To enrich the party automatically at upload, create the party under Trading Parties. Then link it with the Override Trading Party option of the sub-organization.

Q: The override trading party replaced a party that the document names. Is this correct?

Yes. An override trading party applies to every document routed to that sub-organization, without conditions. It replaces the detected primary party and its extracted values with the override party's values. Only one override party per sub-organization is possible. Remove the override if this behavior is not wanted.

Q: I cannot create a party because a duplicate exists. But I need it in another organization.

Duplicates are prevented inside the same sub-organization only. To use a party in other organizations, copy it. On the Trading Parties page, click the "Copy trading party" icon in the Actions column. Then select the target organizations or sub-organizations and click Add.

Q: What happens when I delete a trading party?

Deletion is a soft delete. Historical documents keep their references to the party. The party cannot be assigned to new documents and does not show in the override selection.

Problems with document flags: the master list in Settings, flag application through rules and instructions, and flag visibility.

Q: The Flags column or the flag filter is not visible.

Two causes are possible:

Q: A document shows "N/A" in the Flags column.

The document is in PROCESSING, RETRYING, FAILED, or SPLITTING status. Flags do not show for these states. Wait until processing completes.

Q: I cannot edit or delete a flag.

The flag is in use by a rule or an instruction. The edit modal then opens with the Update button disabled and a red warning, and deletion is blocked.

Q: The platform rejected my flag name.

Flag names follow these rules:

Q: I cannot save an instruction build that adds a flag.

The instruction references a flag that does not exist in the master list. Create the flag first in Settings > Flagging, or remove the reference from the instruction text. Then save again.

Q: Flags disappeared after I reprocessed a document.

This is expected. On reprocessing, all applied flags are cleared and calculated again. The result reflects the current state of your rules and instructions. Flags never carry stale data from earlier runs.

Q: My CSV flag rule does not match documents.

Info: Flag values from a CSV are applied directly from the file. They do not need to exist in Settings > Flagging. To update the mappings, remove the old Supporting File and upload the new version. In-place editing is not supported.

Q: The flag filter shows more documents than I expected.

The filter uses OR logic. When you select several flags, the queue shows documents that carry at least one of them, not all of them. Select fewer flags, or combine the flag filter with other filters.

Problems with the natural-language instructions that guide AI extraction. Instruction Builds run first, before Transformations and Rejection Rules.

Q: My instruction gives different results on similar documents.

The instruction is probably not specific enough. Vague text such as "make data complete" is not testable. Do these steps:

Q: An instruction changes documents that it must not change.

The rule scope is too broad. A Global or Country scope applies to many document flows. Set the scope as narrow as possible: use the Trading Party or Organization scope. Consolidate rules that conflict inside the same scope, or set a clear precedence between them.

Q: My instruction did not seem to apply. How do I check what happened?

Also check the scope order. Rules run from the most specific scope to the most general scope: Trading Party Pair, Trading Party, Organization, Country, then Global. See the Transformations section below for how overlapping rules interact.

Q: Can an instruction build read a Supporting File?

No. Instruction Builds cannot reference Supporting Files. Only Transformations and Rejection Rules can. To validate or enrich a field against a reference file, configure a Transformation or a Rejection Rule instead.

Q: When do I use an instruction build, and when a transformation?

Keep extraction logic in Instruction Builds. Keep formatting and reshaping in Transformations. Mixed logic is hard to maintain.

Q: Who can create instruction builds, and what are the limits?

Q: I changed an instruction, but old documents did not change.

This is expected. A change takes effect on the next document that is processed. Reprocess the older documents to apply the new instruction to them. Remember that reprocessing removes manual edits (see the Document Review and Validation section above).

Problems with the rules that reshape extracted data after Instruction Builds and before Rejection Rules.

Q: An arithmetic action (INCREASE, DECREASE, MULTIPLY, DIVIDE) did not run.

The target field is not numeric. Arithmetic actions are skipped on non-numeric fields. Use a SET action first to give the field a numeric default value. Then apply the arithmetic action.

Q: A general rule overwrote my more specific rule. Why?

For transformations, scope precedence is an execution order, not a filter. All matching rules run, from the most specific to the most general: Trading Party Pair, Trading Party, Organization, Country, then Global. A SET action overwrites the field. So the rule that runs last (the more general one) wins the value, if its condition still matches.

To make a specific rule win:

Info: The platform does not warn you about conflicting rules. Check the result in Activity > Document Processing, where the before and after values show the sequence of changes.

Q: Can I mix AND and OR conditions in one rule?

No. One rule uses one logic connector: all-AND or all-OR. To combine both, split the logic into more than one rule. Smaller, single-purpose rules are also easier to maintain and debug.

Q: The NOT_CONTAINS operator does not work on my older rule.

Rules built before the operator was released keep their old behavior. Regenerate the rule with "Generate with AI" so that it picks up the operator. Then reprocess the affected documents.

Q: Two rules in the same scope use CONTAINS on the same field. Which one wins?

Inside one scope, the rule that matches the longer literal value runs last. So the more specific text wins. This tie-break applies only inside a single scope, not across scopes.

Q: How do I round a value in a transformation?

Rounding and number formatting are not native actions. Apply rounding in a subsequent rule, or handle the precision in the destination system. Also note: percentage actions use the field value at the moment of execution. When you chain percentage adjustments, the order of the actions changes the result. Actions run from top to bottom.

Q: A transformation did not correct a wrong extraction.

This is by design. Transformations reshape data; they do not correct extraction errors. Use Instruction Builds for extraction accuracy. Then let transformations format the correct values.

Problems with the guardrail rules that automatically reject documents. Rejection Rules run last, on the final transformed values.

Q: Correct documents are rejected automatically. Why?

Two frequent causes:

Q: Which rejection rule rejected my document?

Rules are evaluated in scope order, from the most specific to the most general. The first rule that matches rejects the document with its own reason, and evaluation stops. Read the rejection reason on the Rejected tab. For more detail, check the Activity Logs.

Q: Can a more specific rule cancel a rejection from a general rule?

No. One matching rejection rule at any scope is sufficient to reject the document. A more specific rule can only reject earlier, with its own reason. It cannot prevent the rejection.

Q: How do I reject documents with values that are not on an approved list?

Q: What happens to a document after an automatic rejection?

The document moves to the Rejected tab with the logged rule reason. It leaves the approval flow. You cannot reprocess it directly; upload it again if processing is required. If global rejection notifications are configured, an email goes to the recipients (see the System Settings section below).

Q: A complex rejection rule is hard to troubleshoot.

Split compound conditions into more than one rule. Each rule then rejects with its own precise reason, which makes triage clear. Remember: one rule cannot mix AND and OR logic.

Problems with the CSV reference files that Transformations and Rejection Rules use for lookups and validation.

Q: What are the limits for Supporting Files?

Q: My upload failed, or the data was not recognized.

Q: Can I change the columns when I replace a Supporting File?

You can add columns, but you cannot remove or rename an existing one. The replacement must contain every column name that the current version has. A renamed column counts as a removal, because the original name is no longer present.

When a column is missing, the replacement fails during processing and the last successful version stays active. The upload reports success first, because the column check runs during processing rather than at upload time.

Column order does not matter, and rows can be reordered, added, or removed. Only the column names are checked. If more than one row can match the same lookup value, the first matching row in the file is used, so reordering rows can change which one applies.

A column you add becomes required from then on. The check compares the replacement against the most recent version that processed successfully, not against the file you first uploaded. Once a replacement that adds a column is promoted, every later replacement must include that column as well. A replacement that fails does not change what is required, because a failed version never records its columns.

The Upload Guidelines panel in the Re-upload supporting file dialog states these rules at the point of upload.

Info: This rule applies only to replacements. A first-time upload has no previous version to compare against, so any column structure is accepted within the limits above.

Q: I replaced a Supporting File. The replacement failed, but the file still shows "Processed" with a yellow indicator.

This is the intended safety behavior. The platform does not promote a replacement until it processes successfully. When a replacement fails (for example, missing or renamed columns, a corrupted header, or an empty file), the last successful version stays active and continues to serve lookups. The failed attempt is recorded as "Failed" in the version history.

Read the tooltip on the yellow indicator for the cause. Correct the file and replace it again.

Q: My first upload of a Supporting File shows a red "Failed" status.

A first-time upload has no previous version to fall back to. Correct the file and upload it again. Check the delimiter, the header row, and the limits.

Q: How do I update the reference data in a Supporting File?

Replace the file with an updated CSV. The rules that link to it then use the new data; you do not edit the rules. For CSV flag mappings, remove the old file and upload the new version. In-place editing of Supporting Files is not supported.

Caution: Old reference data causes silent validation failures. Keep cost centers, price lists, and supplier lists current.

Q: Which features can use Supporting Files?

Transformations and Rejection Rules can link Supporting Files. Both use the CONTAINS and NOT_CONTAINS operators for list matching. CSV flag rules can also apply flags from a mapped column. Instruction Builds cannot use Supporting Files (see the Instruction Builds section above).

Problems with automatic approval of repeat documents from trusted trading pairs. Auto Review works on Invoice and Purchase Order documents.

Q: No Auto Review icons show on my documents.

Q: The second document from a pair shows no green lightning icon.

Your role cannot trust pairs. Only the Admin and Manager roles, and Custom roles with the Auto Review module, see the green lightning icon. Ask a user with one of these roles to trust the pair. Approvals by Reviewers and Validators also do not start or complete the trust-building process.

Q: A document has a red lightning icon. What do I do?

Auto Review found something different and left the document in the Pending queue for a person.

Info: The Add action needs the Admin or Manager role, or a Custom role with the Auto Review module.

Q: I added a trusted value, but similar documents are still flagged.

A different field or check also failed on those documents. Open each flagged document and review the remaining Unverified rows and reasons. Adding a value applies from that point on; it does not change documents that were approved earlier.

Q: The Auto Review menu is missing from Settings.

Your role does not include Auto Review access. Only Admin and Manager roles, and Custom roles with the Auto Review module, can manage Auto Review, the trusted pairs, and the trust ledger.

Q: What happens when I switch Auto Review off, or stop trusting a pair?

Nothing is lost. The trusted pairs and their learned values are kept. When you switch Auto Review on again, they become active again. When you stop trusting a pair, its documents return to manual review, but the learned values stay. A green lightning icon appears on the pair's next document, and one click trusts it again.

Q: A pair auto-approves without a prompt. Is this an error?

No. The pair was trusted before, and its learned values are still in place. This is the expected behavior. To stop it, delete the pair on the Settings > Auto Review page, or turn the pair off on its Override page.

Q: How do I adjust the checks for one trading pair?

A per-pair override applies only to that pair, on top of the defaults. To stop trusting one learned value, open the Trust Ledger (shield) icon and delete the value. The next document with that value is flagged.

Problems with the operational dashboard for volumes, accuracy, and reviewer throughput.

Q: The Documents Accuracy card shows "N/A".

Accuracy is calculated from human corrections against the AI output. It stays "N/A" until documents have been reviewed. Review some documents; the value then appears. The value can also change with document complexity, volume, and reviewer consistency.

Q: The numbers for a date look wrong.

Check the "Filter by" dropdown in the Detailed document stats section. It has two bases:

Use Activity Date to analyze operational trends and reviewer productivity. Note that all dates are in UTC, and that delayed reviews shift activity metrics to later dates.

Q: Can I approve or reject documents from Usage Insights?

No. Usage Insights is a monitoring and reporting module only. Use the Review Screen for document actions.

Q: Data for an organization is missing from the dashboard.

The organization filter shows only the organizations you can access. Self-service and role-based visibility rules apply. Ask an administrator to check your organization mapping.

Q: The dashboard is slow.

A very broad date range returns a large data set. Start with a narrow range, for example one week. Widen the range only when necessary.

Q: How do I export the data?

Switch to the List View. Use the search box to filter the rows. Then click the Download button to export the data for external use.

Problems with the outbound data format and with the API credentials that external systems use to connect to Docupath.

Q: I lost my Client Secret.

The Client Secret is shown one time only, at generation. It cannot be retrieved later. Delete the credential and generate a new pair in Settings > Destination Format & APIs. Then update your external systems with the new values.

Info: Store the Client ID and Client Secret in a secrets manager. Do not put them in source code, plain-text files, or email.

Q: My API requests fail with an authentication error.

Docupath uses OAuth 2.0. Do not send the Client ID and Secret with every request. Do these steps:

An invalid_client error means the credentials are wrong or the credential pair was deleted. All requests must use HTTPS on https://api.docupath.app.

Q: The downstream system cannot parse our exports after a format change.

A destination format change applies to all future exports, tenant-wide. Documents exported earlier keep their original format. Tell the downstream teams before you change the format, and test the new output with them.

Q: How do I rotate API keys without downtime?

Caution: Credentials do not expire automatically. Add key rotation to your security calendar. Use a separate credential pair for each system.

Q: Can I limit a credential to one organization or document type?

No. API credentials are tenant-scoped. They cannot be restricted to specific organizations or document types inside the tenant. Depending on your plan, optional controls such as an IP whitelist or scoped permissions can be available per credential.

Problems with the templates that map extracted fields to the schema of your downstream systems, with mustache variables such as {buyer_name}.

Q: The export output has empty values.

A variable name is missing or misspelled. A misspelled variable renders as an empty string, and validation does not catch it. Compare each variable in the template against the Variables Panel for that document type. Correct the names and save.

For fields that are truly optional, wrap the section in a conditional block so that empty tags are not rendered: {{#buyer_name}}...{{/buyer_name}}. Give a fallback with an inverted block: {{^total_amount}}0.00{{/total_amount}}.

Q: My new template is not used in exports.

Q: The output shows the wrong fields.

The wrong document type is probably selected. Variables are document-type specific. Open the template, check the File type dropdown, and select the correct document type. Add the XML body for each document type that the template must cover.

Q: Can I hardcode a static value in a template?

Avoid it. Hardcoded values make templates hard to maintain. Set computed or static values with Transformations or Instruction Builds. Use the template only to map the fields to the target structure.

Q: I changed a template. Do old exports change?

No. Template changes apply only to future exports. Documents exported earlier keep their original format. Coordinate template changes with the downstream teams before you apply them.

Q: How do I track template versions?

The platform has no built-in version comparison. Use a naming convention with a version number, for example UBL-Invoice-Nordics-v1.0. Archive old versions with a suffix, for example -archived. Keep a change history in your internal documentation. Only Admin users can edit templates, and the limit is 1 MB per template.

Problems with the tamper-evident audit trail that answers "who did what, and when".

Q: How do I find out which rule changed a field on a document?

Q: My search returns too many results.

The date range is too broad. Start with a narrow range, for example the last 7 days. Widen the range only when necessary.

Q: Expected events are missing from my search.

A search on a person alone can miss events. Also search on action keywords, for example "Approve", "Role", or "Retention". Always expand the entries to see the before and after values; the summary line does not show them.

Q: How do I export the logs?

There is no built-in bulk export. Copy the entries manually from the interface; log entries are copy-friendly for support tickets and audit notes. For long-term compliance, save critical entries in an internal audit repository.

Info: Log retention follows the tenant's data retention policy. Logs are tamper-evident: they cannot be modified or deleted. Access is available to Admin and Manager roles.

Q: Do the logs show system performance data?

No. Activity Logs record who did what, and when. They do not contain system-level diagnostic traces or performance metrics. Also, before and after values exist only for configuration changes; document lifecycle events show the action without field-level detail.

How to get help when this article does not solve your problem: the Help Center, the Support Questionnaire, and the Messenger.

Q: How do I contact support?

You have three paths, all in the navigation bar at the top of the platform:

You can also send an email to support@docupath.ai.

Q: Which severity do I select on the Support Questionnaire?

Q: The form rejected my attachment.

Attachments must be JPG, PNG, JPEG, or WEBP files, with a maximum size of 10 MB. Convert the file, or reduce its size, and attach it again.

Q: How do I report a problem with one specific document?

Open the document in the Review Screen. Then open the Support Questionnaire from the Help Center icon on that screen. The Document ID, Document Type, Document Number, and Sub-Organization fields fill in automatically. Describe the problem and click Submit Request.

Q: I cannot escalate a Messenger conversation to a human.

Human escalation depends on your support model:

Q: Support tickets from a new organization go to the wrong place.

New organizations and sub-organizations are not added to Partner-Led routing automatically. An Admin or Manager must open System Settings > Bypass default support requests, select the new organizations under "Select organizations to bypass", and click Save. The routing method section and the organization selection each have their own Save button; save both.

Problems with your personal profile: your display name and your email address.

Q: I changed my name or email, but the old value still shows.

Changes take effect immediately, but the display can lag. Refresh the browser. Your name appears in the Main Menu, in approval records, in Activity Logs, and in reviewer panels.

Q: I cannot log in after I changed my email address.

Email changes are not validated against your Single Sign-On (SSO) provider. If your tenant uses SSO (Google or Microsoft), an email change without coordination can break the login. Contact your IT team to align the addresses.

Caution: Coordinate every email change with IT before you make it, when SSO is in use.

Q: Old log entries still show my previous name.

This is expected. Name changes are not applied to historical Activity Log entries. Past actions keep the name that was active at the time. This protects the audit trail.

Q: Can an administrator update user profiles in bulk?

No. There is no admin interface for bulk profile updates. Each user updates their own profile in Settings > User Settings. Do not use a shared inbox address for an individual account; it weakens the audit traceability.

Problems with tenant-wide configuration: duplicate handling, rejection notifications, data retention, and branding. Only Admin users can change System Settings, and all changes are logged.

Q: Correct documents are flagged as duplicates.

Duplicate detection matches on the document number only. Some industries reuse document numbers across suppliers or periods, which creates false positives. Change the handling mode:

Content-level deduplication is not supported.

Q: Nobody receives rejection notification emails.

Two conditions must both be true: the feature is enabled, and at least one recipient is configured. If either is missing, no email is sent, and there is no fallback to an administrator address.

Info: Failed deliveries are logged but not retried. Check the audit trail if an email did not arrive, and verify the address and the mail server.

Q: The download link in a rejection email does not work.

Large documents are delivered as a presigned download link instead of an attachment. The link is valid for 7 days from the rejection. After 7 days, log in to Docupath and retrieve the document from the Rejected tab.

Q: What happens when I shorten the data retention period?

All documents older than the new retention window are deleted permanently, usually within 24 to 48 hours. The retention period applies to all documents and statuses in the tenant, without exceptions per document type.

Warning: This action is irreversible. Deleted documents cannot be recovered, also not by support. Plan retention changes with your legal, compliance, and operations teams before you save. The range is 30 to 2,555 days.

Q: My new tenant logo does not show.

The browser cache probably holds the old logo. Do a hard refresh of the browser. Use an SVG or PNG file; the recommended size for PNG is 300 x 120 pixels. Branding changes apply to the full tenant and are visible to all users.

Q: Can I apply a System Setting to only one organization?

No. System Settings are tenant-wide. They cannot be scoped to specific organizations or user groups. Use organization and sub-organization settings for per-organization behavior, such as urgent marking or self-service access.