# Data Protection Impact Assessment (DPIA)

> A risk check you write first. Then a risky data use may start.

**Field:** Data Engineering  
**Category:** Data Privacy  
**Also known as:** DPIA, Privacy Impact Assessment, PIA

## Think of it like

It is like testing the ice before you skate. You check first, then you go.

## Definition

A data protection impact assessment is a written risk check before a new use of personal data. Article 35 of the GDPR asks for one when a use is likely to be high risk for people. Tracking people on a large scale, scoring them or using health data are examples. The check describes the use and asks if it is really needed. Then it lists the risks and the steps that cut them. If a high risk is left, the firm must ask the watchdog before it starts.

## Example

A gym wants members to get in with a face scan. Its DPIA flags face data as sensitive. So the gym keeps a card as another way in, and deletes the scans of anyone who leaves.

## Why it matters

It finds the harm on paper, while the plan can still change. Skipping a required one can bring a fine even if nothing goes wrong.

## Common mistake

Writing it after launch to fill a file. It has to come before the data is used.

## Three principles

- **Before you start.** Check while the plan can still change.
- **For high risk.** Needed when people could be harmed.
- **Ask if risk stays.** Talk to the regulator before going ahead.

## Related terms

- [Privacy by Design](https://learn.benibix.com/term/data/privacy-by-design)
- [Data Protection Officer (DPO)](https://learn.benibix.com/term/data/data-protection-officer)
- [Data Protection Authority](https://learn.benibix.com/term/data/data-protection-authority)
- [Records of Processing Activities (RoPA)](https://learn.benibix.com/term/data/records-of-processing-activities)
- [AI Impact Assessment](https://learn.benibix.com/term/ai/ai-impact-assessment)

## Told apart from

- [DPIA vs AI Impact Assessment](https://learn.benibix.com/compare/dpia-vs-ai-impact-assessment): People think a DPIA and an AI impact assessment are the same check, and they are not. A DPIA looks only at the risks of using personal data. The GDPR asks for it. An AI impact assessment looks at every way an AI system could harm people, such as bias or safety. It counts even when no personal data is used. One AI project can need both.
- [Transfer Impact Assessment vs DPIA](https://learn.benibix.com/compare/transfer-impact-assessment-vs-dpia): People think a transfer check is the same as a DPIA, and it is not. A DPIA asks what a new use of data could do to the people in it. A transfer impact assessment asks one narrower thing. Will the laws of another country undo the safeguards on data sent there? A project that sends data abroad may need both.

![Four steps in a row: describe the use, ask if it is needed, list the risks to people, and pick fixes that cut them. From the last step two paths lead down. If the risk is now low, start and keep the check up to date. If a high risk is left, ask the regulator first.](https://learn.benibix.com/diagrams/mlaw-dpia-steps.svg)

---

Source: https://learn.benibix.com/term/data/data-protection-impact-assessment
Author: Behnia Zarei (a developer in Vienna, Austria, who builds and teaches data, AI and business intelligence at BENIBIX)
Publisher: BENIBIX Learn (https://learn.benibix.com/)
Last updated: 2026-11-18
Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Quote freely with attribution.
