Dynamics 365 Customer Insights Common Errors, Login, Sync, and Integration Fixes
Why Dynamics 365 Customer Insights Errors Keep Happening
Picture this: your marketing team is waiting on a fresh segment build. Your data sources were all green yesterday. You come in this morning, open Dynamics 365 Customer Insights, Data, and you're greeted with "Completed with errors," a cryptic reference to a dataflow that was skipped, or worse, a blank screen where your unified customer profiles should be. I've seen this exact situation on dozens of enterprise tenants, and the frustration is completely valid, especially when this platform sits at the center of your customer data strategy.
Here's what makes Dynamics 365 Customer Insights particularly tricky to troubleshoot. The product went through a major rebrand on September 1, 2023. What used to be called Dynamics 365 Customer Insights is now officially Dynamics 365 Customer Insights - Data, and Dynamics 365 Marketing became Dynamics 365 Customer Insights - Journeys. That name change still causes confusion, especially when you're searching for documentation, error messages reference old names, or your admin portal hasn't fully propagated the branding update.
Beyond naming, the platform has a genuinely complex architecture. There are two separate refresh cycles, system refresh controlled by admins, and data source refresh owned by individual data source owners, and they don't automatically talk to each other the way you'd expect. Power Query connectors have their own authentication layer. Delta Lake tables have their own versioning constraints. And then there's the unification pipeline, deduplication, matching, and merging, where a single schema mismatch can cascade into no customer profiles being built at all.
The error messages Microsoft surfaces don't help much. "Dataflow refresh was skipped because the System refresh settings are not applicable for this datasource" sounds like the system did something on purpose. It didn't. It's telling you that you haven't configured a scheduled refresh on the Power Query side, but only if you know how to decode that message. Most people don't on first read.
The good news: almost every common Dynamics 365 Customer Insights error has a defined resolution path. Credential errors, Delta Lake ingestion failures, Text/CSV connector authentication breakdowns, segment eligibility problems, all of them are fixable. This guide walks you through each one, in plain language, with exact steps from official Microsoft documentation. Browse all Microsoft fix guides →
Whether you're seeing a Dynamics 365 Customer Insights Data ingestion error, a Power Query authentication failure, a missing segment member issue, or an export problem in Power BI Desktop, you're in the right place. Let's fix it.
The Quick Fix, Try This First for Dynamics 365 Customer Insights Errors
Before diving deep, run through this checklist. In my experience, roughly 60% of Customer Insights errors come down to one of three root causes: stale credentials on a Power Query connection, a missing scheduled refresh on the data source itself, or an unsupported Databricks reader version. Here's how to knock all three out quickly.
Step 1, Refresh your Power Query connections. Go to your data source in Customer Insights - Data, open it in Power Query, and select Manage connections on the transformations page. Hit the refresh icon next to each listed connection. Then refresh your browser tab. Save the data source. This alone resolves the "credentials are missing or invalid" error in most cases because MFA tokens expire silently.
Step 2, Check your data source refresh schedule. Navigate to Data sources, find your Power Query-based source, and look at its refresh settings. If you see "Completed with errors" alongside a message about dataflow refresh being skipped, the problem is that the data source has no standalone refresh schedule configured. The system refresh your admin set up only handles non-Power Query sources. Click Edit refresh settings on the Power Query data source and configure a dedicated schedule.
Step 3, Validate your Databricks reader version. If you're ingesting Delta Lake format tables and seeing "Delta Lake ingestion job failed to get version," open your Databricks workspace and check the minReaderVersion on the affected table. Customer Insights - Data supports Databricks features with a minReaderVersion of 2 or earlier. If the table uses version 3 features, you need to remove those features and revert the reader version before re-attempting ingestion.
If the problem is with the Text/CSV connector specifically, where reauthentication fails and the error "The credentials provided for the source are invalid" loops endlessly, don't try to authenticate through the error dialog. You need to go through the Power Query transformations page, use Manage connections, and refresh the connection from there. The Configure connection button in the error message itself won't work in this scenario.
This error trips up a lot of Customer Insights administrators because the platform makes it seem like something went wrong during a scheduled task, but the actual problem is that no task was scheduled in the first place for that data source.
Here's exactly what's happening: Dynamics 365 Customer Insights - Data runs two kinds of refresh. The system refresh is a tenant-wide schedule set by an administrator. It handles the full pipeline, unification, insights, segments, exports. But Power Query-based data sources have a separate, owner-controlled refresh schedule that runs independently. When system refresh runs and finds a Power Query data source with no schedule of its own, it copies the previous state of that data source rather than pulling fresh data. That's the "skipped" behavior you see in the status message.
To fix this, you need to configure the dedicated refresh schedule on the Power Query data source:
- In Customer Insights - Data, go to Data > Data sources.
- Locate the Power Query data source showing "Completed with errors."
- Click the three-dot menu next to it and select Edit refresh settings.
- Enable the scheduled refresh toggle and set your desired frequency, hourly, daily, or weekly depending on how often the source data changes.
- Save the settings.
Once saved, the next system refresh will trigger the Power Query data source on its own schedule. You'll know it worked when the data source status shows "Refreshed successfully" and a current timestamp, not just "Completed" with a stale time.
One thing to watch: the data source owner setting matters here. Only the person who owns the connection can configure its refresh schedule, or an administrator with access to all connections. If you're an admin and can't see the option, check whether the connection is assigned to a service account or a specific user who may have left the organization.
If you're seeing "Data source credentials are missing or invalid. Please update the connection credentials in settings, and try again" in Dynamics 365 Customer Insights - Data, the underlying cause is almost always an expired or incomplete multi-factor authentication token on the Power Query connection.
This happens more often than you'd think. When a connection is first authenticated, the MFA session is valid for a certain window. Over time, especially if the connection isn't used frequently, or if your organization rotates MFA tokens, the session expires. Customer Insights tries to use the old credentials, fails silently, and surfaces this error on the next refresh.
Here's how to fix it:
- Open the affected data source in Customer Insights - Data.
- The connection owner needs to complete a fresh MFA challenge. This might be a push notification to a phone authenticator app, a one-time passcode, or an RSA token, depending on your organization's configuration.
- If you have access to your organization's user profile management self-service portal, log in and verify that your MFA or two-factor authentication requirements are active and up to date on your account.
- If an administrator manages user accounts in your org, contact them and ask them to verify the credential requirements on your account, specifically that MFA or 2FA is properly configured and the token isn't expired.
- After completing re-authentication, trigger a manual refresh of the data source and confirm it completes without the credentials error.
A common mistake here is trying to re-enter passwords or API keys when the actual missing piece is the MFA step. The credentials themselves are usually still valid, it's the second factor that's lapsed. If you see this error immediately after onboarding a new service account, double-check that MFA is enabled and properly configured for that account before connecting it to Customer Insights.
This one is specific to customers ingesting data stored in Delta format. If your Delta data source in Dynamics 365 Customer Insights - Data is trying to read a table that requires a Databricks minReaderVersion of 3 or higher, you'll see:
Delta Lake ingestion job failed to get version <DataSourceId>.And when you try to view the columns in that table, you'll also see: "Failed to save data source schema." That second error is the downstream consequence, Customer Insights can't even inspect the schema because the ingestion job can't establish a version context.
The root cause is a hard version ceiling in the platform. Dynamics 365 Customer Insights - Data supports Databricks features with a minReaderVersion of 2 or earlier. If your table was upgraded to use features that require version 3, things like deletion vectors, column mapping in certain configurations, or other Databricks-specific reader protocol enhancements, those tables are not compatible.
To resolve this:
- Open your Databricks workspace and navigate to the affected Delta table.
- Identify which version-3 feature is applied to the table. Run
DESCRIBE DETAIL <table_name>in a Databricks notebook to inspect theminReaderVersionandminWriterVersionproperties. - Remove the version-3 feature from the table. The specific command depends on which feature was enabled, consult the Databricks documentation for Supported Databricks features and versions to identify the correct removal step.
- After removing the feature, verify the table's
minReaderVersionhas reverted to 2 or lower. - Return to Customer Insights - Data, open the Delta data source, and attempt to refresh it.
If you can't remove the feature because it's needed by other consumers of the same Delta table, the alternative is to create a separate Delta table copy specifically for Customer Insights ingestion, without the version-3 features applied. This is the cleaner long-term approach in environments where the Delta table is shared across multiple downstream systems.
The Text/CSV connector in Dynamics 365 Customer Insights - Data has a quirk that catches people off guard: when you need to reauthenticate, the normal re-authentication flow through the error dialog doesn't work. You click "Configure connection," the error persists, and you're stuck in a loop. This is a known behavior documented by Microsoft.
The fix requires going around the error dialog entirely and working directly from the Power Query transformations page. Here's the exact path:
- In Customer Insights - Data, go to Data > Data sources and open the Power Query data source that uses the Text/CSV connector.
- This brings you to the Power Query editor. On the transformations page, look for the Manage connections option in the toolbar or the Home ribbon.
- A panel or dialog will appear listing all connections associated with this data source.
- For each listed connection, find the refresh icon next to the connection name. You can also open a connection for editing to access the refresh control. Refresh every connection listed.
- After refreshing all connections, close the Manage connections panel and do a full browser page refresh, press F5 or Ctrl+R.
- Return to the data source and click Save.
You need to be the owner of the connection, or have administrator permissions that give you access to all existing connections, to perform these steps. If you're an admin but don't see all connections listed, check whether the connection is owned by a different user account. In that case, you'll need to either impersonate the connection owner (if your org's policies allow it) or reassign connection ownership before proceeding.
After saving, trigger a manual refresh of the data source and confirm the credentials error is gone from the data source status page.
If you're hitting a sign-in error that reads something like "The service principal for resource is disabled," the problem lives in your Azure Active Directory (now Microsoft Entra ID) configuration, not in Customer Insights itself. This is an environment management issue, and it's common in organizations that have recently changed their Azure tenant settings, rotated app registrations, or had a service principal expire or be disabled by a conditional access policy.
A service principal is essentially the app identity that Customer Insights uses to authenticate against Azure resources. If that identity is disabled, either manually or by an automated policy, Customer Insights can't sign in to access downstream dependencies like Azure Data Lake, Dataverse, or other integrated services.
To investigate:
- Open the Azure portal and navigate to Microsoft Entra ID > Enterprise applications.
- Search for the application name associated with your Customer Insights environment. Filter by "Disabled" status to find candidates.
- If you find the service principal in a disabled state, open it and click Enable under the Properties section, provided your Azure admin confirms this is safe to do.
- If the service principal was disabled by a conditional access policy, work with your Azure administrator to create an exemption or adjust the policy scope.
- After re-enabling, return to Customer Insights - Data and attempt sign-in again.
A second scenario under login errors: if your Customer Insights environment was provisioned before the September 2023 rebrand, you may occasionally encounter references to old environment names or stale tenant configuration. If sign-in fails with a generic authentication error, check whether your Customer Insights environment URL is still resolving correctly and that the environment itself is in an active, non-suspended state in the Power Platform admin center under admin.powerplatform.microsoft.com.
Advanced Dynamics 365 Customer Insights Troubleshooting
If you've gone through all five fix steps above and you're still seeing errors, it's time to move into deeper diagnostic territory. These are the scenarios I see most often in enterprise and domain-joined environments where standard fixes don't stick.
Unification Pipeline Failures, Match, Merge, and Deduplication
The Dynamics 365 Customer Insights - Data unification pipeline runs in three stages: deduplication, matching, and merging. A failure at any stage produces a separate error, "Match failed error," "Merge failed error," or issues with deduplication output. These are distinct from data ingestion errors and require their own diagnostic approach.
Start by navigating to Data > Unify and examining the status of each unification stage. Look for tables in the unification output that are missing or have unexpected row counts. The most common causes are schema mismatches between source entities (a field type changed upstream), a missing business unit value in your data model, or a rule conflict in your match configuration that produces zero matches.
If you see "Detected DataVerse dependencies in msdynci_customerprofile" error, this indicates that something in your Dataverse environment has a hard dependency on the Customer Insights customer profile table, typically a custom workflow, plugin, or Power Automate flow, that's blocking the unification output from being written. You'll need to identify and remove or pause those dependencies before re-running unification.
Segment Membership Issues
Segments returning no members is a surprisingly common issue that can have several root causes. First, check whether the segment definition references attributes that come from a unified customer profile, if the last successful unification run is stale or failed, segment evaluation will correctly return zero members because there are no current profiles to evaluate against. Fix the underlying unification issue first.
If unification is healthy, look at the segment eligibility for export. An export target may have specific requirements, for example, certain destinations only accept segments above a minimum size threshold, or require specific attribute mappings that the segment doesn't satisfy. Navigate to Segments, select the affected segment, and check its export eligibility status.
Power BI Desktop Integration Errors
If your Customer Insights environment doesn't show up in Power BI, or you're getting errors on date fields when loading entities in Power BI Desktop, there are two separate problems at play. For missing environments, verify that the account you're using in Power BI Desktop has the correct permissions in Customer Insights - Data, specifically the Viewer or Contributor role. Date field errors in Power BI Desktop typically stem from a locale mismatch between the Customer Insights data format and Power BI's regional settings; adjust the locale in Power BI Desktop's query settings to match the format coming from Customer Insights.
Export Failures, Name Already Exists
If exports fail with "the following name already exists," it means a prior export destination was configured with the same identifier that your current export is trying to use. This happens after environment copies, tenant migrations, or when someone manually created an export target with the same name in the destination system. Navigate to Exports in Customer Insights - Data, find the failing export, edit it, and either rename the export or point it to a new destination target that doesn't have a name conflict.
Prevention & Best Practices for Dynamics 365 Customer Insights
Most of the errors in this guide are preventable with the right operational habits. I've watched organizations spend hours each month fighting Customer Insights refresh failures that could have been avoided entirely with a few upfront configuration decisions. Here's what actually works.
Use service accounts for data source connections, not personal user accounts. When a human user owns a Power Query connection and then leaves the organization, changes their password, or has their MFA token expire, every data source tied to that account stops refreshing. Service accounts with managed credentials and long-lived MFA tokens (or certificate-based authentication where your policy allows it) eliminate this entire category of credential expiry failures.
Set refresh schedules on every Power Query data source at creation time. Don't leave this to chance. When you add a new Power Query data source, configure its dedicated refresh schedule before you save and close it. This one habit prevents the "dataflow refresh was skipped" error from ever appearing in the first place. Schedule it for slightly before your system refresh runs so that fresh data is ready when unification kicks off.
Pin your Delta tables to supported Databricks reader versions. If your org uses Delta Lake format for ingestion, implement a governance rule that tables connected to Customer Insights must stay at minReaderVersion 2 or lower. Document this constraint in your data platform runbook and build a monitoring check that alerts when a reader version is upgraded on any table flagged for Customer Insights ingestion.
Audit your segment definitions after every unification schema change. Whenever a source entity schema changes, a field is renamed, a type is changed, or an entity is removed, review all segment definitions that reference that entity. Schema drift is the leading cause of unexpected zero-member segments, and catching it proactively saves a lot of downstream confusion for marketing teams who rely on those segments.
- Set up a dedicated refresh schedule on every Power Query data source immediately after creation, before closing it for the first time.
- Assign data source connections to service accounts with managed, long-lived credentials rather than individual user accounts.
- Run a monthly audit of Data sources to catch any connections showing "Completed with errors" before they affect downstream segments or exports.
- Document your Databricks Delta table reader versions in your data platform runbook and flag any table used by Customer Insights for version-change review.
Frequently Asked Questions About Dynamics 365 Customer Insights Errors
Why does my Dynamics 365 Customer Insights data source say "Completed with errors" even though the data looks fine?
"Completed with errors" on a Power Query data source most commonly means the dataflow refresh was skipped, the system refresh ran, found no standalone schedule on the Power Query source, and copied the previous data state instead of pulling fresh data. Your profiles look correct because they're based on the last successful refresh. The fix is to configure a dedicated scheduled refresh on the Power Query data source via Edit refresh settings. Until you do that, the system will keep serving stale data silently on every run.
I keep getting "Data source credentials are missing or invalid", I already re-entered my password and it's still there. What am I missing?
This is almost certainly an MFA issue, not a wrong password. The credentials themselves are fine, it's the multi-factor authentication step that's expired or hasn't been completed for the connection. The connection owner needs to go through a fresh MFA challenge (phone authenticator push, one-time code, or RSA token depending on your organization's setup). If you manage your own account through a self-service portal, verify your MFA configuration is active. If an admin manages accounts in your org, ask them to check the MFA/2FA requirements on your account specifically.
What Databricks reader version does Dynamics 365 Customer Insights - Data actually support?
Customer Insights - Data officially supports Databricks features with a minReaderVersion of 2 or earlier. Any Delta table that requires reader version 3 or higher, including tables using certain deletion vector configurations or newer Databricks reader protocol features, will fail ingestion with the "Delta Lake ingestion job failed to get version" error. You need to remove the version-3 features from the table and revert to reader version 2 before the ingestion job will succeed.
My Text/CSV connector keeps saying credentials are invalid and clicking "Configure connection" does nothing. How do I fix it?
Don't use the Configure connection button in the error message, it won't work for the Text/CSV connector in this scenario. Instead, open the data source in the Power Query editor, go to Manage connections on the transformations page, and use the refresh icon next to each listed connection to renew them individually. After refreshing all connections, do a full browser page refresh (F5), then save the data source. You need to be the connection owner or an administrator with access to all connections to do this.
Why are my Customer Insights segments returning zero members when I know the underlying data is there?
Zero-member segments usually point to one of two things: either the unification pipeline hasn't run successfully recently (meaning there are no current unified profiles to evaluate the segment against), or the segment references attributes from an entity whose schema has changed since the segment was created. Check the unification status first under Data > Unify. If unification is healthy and current, review the segment definition to make sure every referenced attribute still exists in the unified profile schema with the correct data type.
What's the difference between system refresh and data source refresh in Customer Insights - Data?
System refresh is a tenant-wide schedule controlled by a Customer Insights administrator. It drives the full downstream pipeline, unification, segments, insights, and exports, but it does not automatically refresh Power Query-based data sources. Data source refresh on a Power Query connection is a separate, owner-controlled schedule that must be configured independently on each Power Query data source. If you only have a system refresh configured and no data source refresh schedule on your Power Query sources, those sources will serve their last-known state on every system refresh cycle instead of pulling new data.