Skip to main content
Follow these steps to create a new validation on a dataset.

Steps

1. Open the creation form

Navigate to Validations in the left sidebar and click New Validation.

2. Select a validation type

Choose the type of check you want to run. Available types include:
  • Completeness — Check for missing or null values
  • Uniqueness — Detect duplicate entries
  • Validity — Validate values against a format or list
  • Reliability — Check row counts and freshness
  • Numeric Distribution — Track statistical metrics (mean, median, percentiles)
  • Custom SQL — Write a custom SQL expression to measure any metric
  • Delta Validation — Compare metric values between runs
See Validation Types for a full description of each.

3. Select your data source and dataset

Choose the Data Source and Dataset the validation will run against. Enter a name for the validation, or accept the auto-generated suggestion.

4. Configure filters (optional)

Apply a filter condition to restrict which rows are included in the metric calculation. For example, filter the tr_volume column to Less Than 5000 to validate only low-volume transactions.

5. Define the threshold

Set a threshold to define what constitutes an incident. When the metric crosses this boundary, Datachecks creates an alert. See Thresholds for the available threshold types.

6. Set the schedule

Choose how often the validation should run — hourly, daily, weekly, or a custom cron expression.

7. Create

Click Create to save the validation. It will appear in the validations list and run according to the schedule you set.

What’s next