Validate Data

Summary

The Validate Data block performs configurable checks on an input dataset to ensure data integrity before it flows downstream. Validation can be applied at different levels - schema, cell values, and record counts - and each check can be associated with a specific failure action (Error, Halt, Warning, Ignore).

This block is useful for preventing workflows from running with malformed, incomplete, or invalid data, while providing flexible error handling and reporting.

How does it work?

• data (mandatory) — connect the dataset to be validated.
• Validation checks are configured in the block options, across three main categories:
• Schema validation — check expected fields exist and match data types.
• Cell value validation — apply rules on individual field values across records.
• Record count validation — enforce minimum and maximum record limits.
• If a validation check fails, the block applies the configured failure action:
• Error: the block fails and workflow execution stops.
• Halt: the branch is halted; downstream blocks on that branch will not execute.
• Warning: execution continues but a warning is raised.
• Ignore: execution continues without interruption.

Optional email reporting can be configured to notify when errors or warnings occur.

Advanced streaming settings determine whether records are passed downstream immediately or only after validation of the full dataset.

Options

Schema validation

Define the expected fields (name + type). All listed fields must be present in the dataset; otherwise, validation fails.

Use this to prevent workflows from running with incomplete or incorrectly typed inputs.

Cell value validation

Define rules on individual fields. For each rule, specify a condition that records must satisfy.

Example:
[Minimum price] > 0

Rules can be evaluated in two ways:

• All rules must pass — if any rule fails, validation fails.
• Any rule can fail — validation fails as soon as one rule fails.

Record count validation

Specify the minimum and/or maximum number of records allowed in the dataset.

Example: Require at least 1 record to avoid running a workflow on an empty dataset.

Failure action

Each type of validation (Schema, Cell values, Record count) has its own failure action setting:

• Error — block fails, workflow stops.
• Halt — block halts, this branch is not executed further, but other workflow branches continue.
• Warning — workflow continues, but a warning is issued.
• Ignore — workflow continues silently.

Email reporting

Choose when the block should send an email:

• Never — no emails sent.
• Always — email every time this block executes.
• On error — only on validation errors.
• On error or warning — on either error or warning.

Note: If the block is halted, halt information is always included whenever an email is sent.

Data streaming behaviour (advanced)

Control how records flow through the block:

• Stream immediately — valid records flow downstream as soon as they are checked. If a later record fails validation, the stream halts; already-passed records may have affected downstream blocks. The input is retrieved once (unless the Problems output is connected).
• Stream after checking all data — the block checks all records first, then streams data downstream only if permitted. Input is retrieved twice if the first output is consumed.

Outputs

• Valid data — the dataset after passing validation (subject to streaming settings).
• Problems — (optional) details of validation failures, if consumed downstream.

Examples summary

• Enforce schema: require Customer ID (text) and Sales (numeric) fields to be present.
• Enforce values: Sales > 0 for all records.
• Enforce record counts: at least 100 records, but no more than 10,000.
• Choose Error if invalid data should stop the workflow entirely, or Warning/Ignore if workflow should continue.
• Configure email reporting to alert data owners automatically when issues occur.