Microsoft Fabric

Convertmax Fabric sync publishes the same reporting datasets used for BigQuery into a Microsoft Fabric Open Mirrored Database that you own. Convertmax writes Parquet files to the OneLake landing zone; Fabric mirrors them into Delta tables for SQL analytics and Power BI.

This is a warehouse export. It is separate from any optional Microsoft Fabric OAuth connection used to browse workspaces. Sync uses a dedicated Microsoft Entra writer app and client secret—not that OAuth token.

Use this guide when Data Warehouse Sync is enabled for your organization.

What gets synced

Six tables, matching the BigQuery warehouse reporting contract:

Each row also includes:

• _source_id — stable identity of the source reporting row
• __rowMarker__ — Open Mirroring upsert marker

Primary key for mirroring: (tenant_id, time, _source_id)

Re-delivered rows with the same key update in place. They do not create duplicates.

When a site can export

A site is exported only when all of the following are true:

1. The Data Warehouse Sync add-on is active for that site’s organization
2. A Fabric destination is configured (Entra directory ID, writer app ID, workspace ID, mirrored database ID, and writer client secret)
3. Export is On in Integrations → Data Warehouse

One Open Mirrored Database can be assigned to only one Convertmax site.

Authentication

Convertmax writes to your OneLake landing zone with a Microsoft Entra application (service principal) and client secret.

Create or designate a writer app in your Entra directory, grant it permission to write to the Fabric workspace that owns the mirrored database, and provide:

• Microsoft Entra directory (tenant) ID
• Application (client) ID
• Client secret
• Fabric workspace ID
• Open mirrored database ID

Store and share the secret only through channels your security team approves. After you save it in Convertmax, the secret is not shown again in the UI.

Prepare the Fabric destination

In Microsoft Fabric:

1. Confirm a workspace with Fabric capacity.
2. Create an Open mirrored database (the OneLake landing zone Convertmax will publish into).
3. Grant the writer app access to that workspace so it can land files in the mirrored database landing zone.

Convertmax does not create the mirrored database for you. The database must exist before you turn export on.

Set up in the Convertmax app

When Data Warehouse Sync is enabled:

1. Open Integrations → Data Warehouse → Microsoft Fabric.
2. Enter the directory ID, writer app ID, workspace ID, and mirrored database ID.
3. Paste the writer client secret.
4. Turn export On.

If you do not see Microsoft Fabric, the add-on may not be enabled for your organization—contact Convertmax support.

Changing the workspace or mirrored database starts a new backfill. Convertmax does not remove files from the previous destination.

How sync works

On a schedule (and when Convertmax runs a sync for your workspace), Convertmax:

1. Reads reporting rows for the entitled site.
2. Writes Snappy-compressed Parquet batches into that site’s Fabric landing zone, including upsert keys.
3. Lets Fabric replicate those files into the six mirrored tables.

First run backfills available history. Later runs are incremental, with a 48-hour lookback so late updates in that window are sent again as upserts.

Empty tables still receive a typed zero-row Parquet file so the schema can appear in Fabric before data arrives.

In Convertmax, Files published (and the timestamp) means files landed in OneLake. It does not mean Fabric has finished replication. Confirm replication in Fabric.

Data behavior

Do not treat this as exact change-data-capture.

Validate the export

After setup and the first successful publish:

1. In Fabric, confirm files appear in the mirrored database landing zone and that replication is running.
2. In Fabric SQL on the mirrored database, confirm the six tables exist and row counts look right:

SELECT 'visits' AS table_name, COUNT_BIG(*) AS rows FROM dbo.visits
UNION ALL SELECT 'acquisition_events', COUNT_BIG(*) FROM dbo.acquisition_events
UNION ALL SELECT 'conversion_events', COUNT_BIG(*) FROM dbo.conversion_events
UNION ALL SELECT 'custom_events', COUNT_BIG(*) FROM dbo.custom_events
UNION ALL SELECT 'click_events', COUNT_BIG(*) FROM dbo.click_events
UNION ALL SELECT 'daily_searches', COUNT_BIG(*) FROM dbo.daily_searches;

3. Compare counts and revenue to Convertmax reporting for the same site and time range.
4. After a later sync, confirm unchanged keys do not inflate counts.

Allow time for Fabric replication after Convertmax shows Files published.

Troubleshooting

Microsoft Fabric is missing in Integrations

Confirm Data Warehouse Sync is enabled for your organization, then contact Convertmax support with your workspace name.

Saved in the app but no files in OneLake

Confirm the Entra directory ID, app ID, workspace ID, and mirrored database ID are correct, that export is On, and that the writer app can write to the workspace. Then contact Convertmax support with your workspace name and the approximate time you completed setup.

Files published, but tables are empty or stale in Fabric

Convertmax has landed files; Fabric may still be replicating. Check the mirrored database replication status in Fabric. If replication is healthy and tables stay empty, contact Convertmax support.

Permission or secret errors

Confirm:

• the writer app is in the correct Entra directory
• the client secret is current (secrets are not shown again after save—rotate and re-enter if needed)
• the app has access to the Fabric workspace and Open mirrored database

Counts keep growing for the same period

Row keys should upsert. If counts inflate after a re-sync, confirm you are querying the mirrored tables (not raw landing-zone files) and contact Convertmax support with the table name and time range.