Term · Data, privacy & security
DPIA (data protection impact assessment)
With a DPIA you look in advance at what could go wrong with people's data, before you start a new processing activity.
In short
A DPIA (data protection impact assessment) is an assessment in advance of the privacy risks of processing personal data. It is mandatory if that processing is likely to pose a high risk to people. You describe the processing, the risks and the measures.
What is a DPIA?
A DPIA is an assessment you carry out before you start processing personal data in a new way. You map the privacy risks to people and choose measures to reduce them.
DPIA stands for data protection impact assessment. In the Dutch legal text it is called a gegevensbeschermingseffectbeoordeling. The obligation is set out in Article 35 of the GDPR.
What goes into a DPIA?
You may choose a method yourself. But it must contain at least four things:
- a description of the processing and its purpose;
- why the processing is necessary, and whether it can also be done with less data;
- the risks to the people whose data you use;
- the measures you take, such as pseudonymisation, less access or encryption.
Start early, while you are still designing the processing. Does a high risk remain after your measures? Then you first consult the Dutch Data Protection Authority (Autoriteit Persoonsgegevens, AP) before you begin.
When is a DPIA mandatory?
A DPIA is mandatory if a processing activity is likely to pose a high risk to people. You decide that yourself, with three tools:
- the cases that the GDPR mentions, such as processing health data on a large scale;
- the AP's list, which includes monitoring employees, camera surveillance and profiling;
- nine criteria from the European supervisory authorities. Does your processing meet two or more? Then, as a rule of thumb, a DPIA is needed.
If you do not carry out a DPIA, you must be able to explain why not.
Example in an SME
Example
Suppose you want an AI tool to sort job applicants' CVs. Then you are assessing people with new technology, and it concerns people who depend on your choice. That way you meet several of the nine criteria. According to the rule of thumb, a DPIA is then needed in advance. With AI, the EU AI Act also plays a role alongside the GDPR.
Do you want to use AI with data about customers or employees? Read how I approach AI in your business, with business accounts and a data processing agreement.
Frequently asked questions
Does a small business also have to carry out a DPIA?
That depends on the risk of the processing, not on the size of your business. Does a small business systematically monitor what employees do, for example? Then a DPIA may be mandatory there too.
Do I have to send my DPIA to the AP?
Usually not. Only if a high risk still remains after your measures do you consult the AP in advance. That is called a prior consultation.
May someone else carry out the DPIA for me?
Yes, you may have it carried out by someone inside or outside your business. You do remain responsible. Is a processor involved in the processing? Then they must give you the information you need.
This is an explanation in plain language, not legal advice.
Sources: the Dutch Data Protection Authority (Autoriteit Persoonsgegevens): Data protection impact assessment (DPIA) (in Dutch) and Decision on the list of processing operations for which a DPIA is mandatory (Besluit lijst verwerkingen waarvoor een DPIA verplicht is, in Dutch). Consulted on 7 October 2026.
Written by Bart van Engelenhoven Founder of Uniekaiconsultancy in Ede, the Netherlands. Bart analyses how SMEs work and builds software, integrations and AI applications himself. Last updated: 7 October 2026
Read more
Term
GDPR (AVG)
The GDPR (AVG) is the European privacy law. Who does it apply to, what are the basic rules and what to watch for with software, integrations and AI?
Read moreTerm
EU AI Act (AI-verordening)
The EU AI Act (AI-verordening) is the European law on AI. Which risk categories exist, which rules apply when and what the 2026 amendment means.
Read moreTerm
Pseudonymisation
Pseudonymisation is processing data so you cannot directly link it to a person. How does it differ from anonymisation, and how do you use it with AI?
Read moreNext step
Curious what could work for you?
- The first conversation is free and without obligation
- By phone or at your premises
- You speak directly with me
Bart van Engelenhoven You tell me what takes up your time, I tell you honestly whether I can help.