Skip to main content
General Validation reconciles source and target data so you can verify that a pipeline or migration produced the expected outcome. You use the hosted application to discover datasets, define checks, and review runs. Validation executes in your Azure subscription. Configuration, schemas, run status, verdicts, counts, and artifact references are stored by General Validation. Source and target rows, credentials, staging, exact result values, and captured failed rows stay in your Azure environment. Your browser reads exact results directly from your customer-owned storage. See Security and the data boundary.

​
Before you start

Have these ready:
  • A Microsoft work account in the Entra directory your organization will use.
  • An Azure subscription in that directory and an administrator with Owner rights who can deploy the execution environment.
  • A supported source and target dataset that should reconcile. Start with one well-understood pair and a row-count check.
  • Access to a directory administrator if your tenant requires admin consent.
Supported inputs include qualified flat Parquet and Delta datasets, exact-file CSV, Azure SQL Database tables, and supported Fabric Lakehouse and Warehouse tables. Review Supported scope for the format, schema, and access requirements.

​
1. Start your trial

Open General Validation and choose Sign in with Microsoft. Use your work account. For a new organization, enter the name your team should see and choose Start 7-day trial. This connects your Entra directory, makes you the first organization Owner, and starts the trial immediately. No credit card is required. The trial includes five active Dataset Pairs and ten Pairs created in total, including archived or deleted Pairs. See Trial and billing. If your directory already has an organization, ask an Owner to invite your work account. Sign in and choose Accept invitation. Invitations do not send an email automatically.

​
2. Connect your execution environment

Open Environment and enter an environment name, subscription ID, and Azure region. Optionally enter an Entra group ID for colleagues who need to read exact results and failed-row artifacts in the browser. Choose Continue to setup. If the page shows Admin consent needed, a directory administrator completes the consent step, then an organization Owner chooses Retry preparation. Review the resources and permissions, choose Deploy to Azure, and complete the deployment in Azure. Return to General Validation and choose I’ve deployed it, verify and finish. Continue once the environment is ready. Set up your Azure environment walks through the complete process. Azure resource consumption is billed to your subscription, including during the free trial.

​
3. Discover and prepare your datasets

In Discovery, register an Azure Data Factory or Microsoft Fabric workspace. The discovery source identifies the metadata to browse; the separate execution factory runs your checks. Select the source and target datasets in the catalog and choose Import selected, or use Import & prepare for an individual item. Import starts schema and source-access preparation automatically. Wait for both datasets to become ready. If preparation needs attention, open the Dataset and follow its specific remediation. Catalog visibility alone does not mean a dataset can run.

​
4. Create a Pair and a Test

Create a Dataset Pair from your prepared source and target. Confirm the selection before saving: a Pair’s source and target cannot be changed later. Add a row-count Test first. Set the comparison and tolerance to match what you expect from this pipeline. Add aggregates or keyed value checks after that first control works. Follow Create your first Test for the authoring steps and supported check types.

​
5. Run and review

Run the Test and open its result. The run first executes in your customer Data Factory, then delivers the result metadata. Review the final verdict and any reported error after delivery completes. When you open a delivered result, exact comparison values and captured failed rows load directly from customer storage using your Microsoft identity. Your account needs the corresponding Azure storage permission as well as General Validation membership. Continue with Runs, results, and evidence and When to validate to turn your first check into a useful control.