Data Protection Impact Assessments: A Practical Guide

Harper James advises start-ups, scale-ups and established businesses on UK GDPR compliance, including when a Data Protection Impact Assessment (DPIA) is required and how to use one to support a sound commercial decision. A DPIA is a structured review of how proposed processing affects individuals, the risks involved and the controls needed before the processing begins.

When a DPIA is required

A DPIA is required where processing is likely to result in a high risk to individuals’ rights and freedoms. Common examples include large-scale processing of special category data, systematic monitoring of public areas, profiling that produces significant effects, biometric identification, employee monitoring and the use of new technologies involving personal data. The assessment should take place during project design, not after a supplier has been appointed or a product has launched.

The first step is to describe the processing clearly: what data is collected, whose data it is, the purpose, the lawful basis, retention period, systems used, recipients and international transfers. Businesses should also record whether a processor is involved and check the relevant contract, security measures and sub-processing arrangements. Practical DPIA resources provide useful material for building a repeatable internal process and keeping assessments consistent across projects.

How to complete the assessment

Next, identify risks such as excessive collection, inaccurate data, unauthorised access, unfair decisions, loss of confidentiality or a failure to honour data subject rights. Assess the likelihood and severity of each risk, then specify controls with an owner and implementation date. Controls include data minimisation, access restrictions, encryption, testing, human review of automated decisions, clear privacy information, deletion schedules and supplier audits. The assessment should also explain why the processing remains proportionate to the business purpose.

A DPIA is a working document, not a one-off form. Review it when the purpose, technology, supplier, data set or risk profile changes, and set a review date for higher-risk projects. If high risk remains after controls are applied, the organisation must consult the Information Commissioner’s Office before starting the processing. For AI projects, the assessment should sit alongside procurement checks, model governance, security testing and a clear process for investigating inaccurate or unfair outputs.