TL;DR
- Privacy by design means you build data protection into a product before you collect any data.
- Privacy by default means the most private setting applies until the user changes it.
- GDPR Article 25 requires both. Swiss FADP Article 7 requires both as well.
- The EDPB calls the two concepts complementary. Each one is weaker without the other.
- Regulators have fined companies for public-by-default accounts and for archives that cannot delete data.
- Practical measures include encryption, data minimisation, access controls, private account settings, and automatic data deletion.
What Is Privacy by Design and Privacy by Default?
Privacy by design (opens in a new tab) means you build data protection into a system's architecture before you collect data. Privacy by default means the system starts with the most protective settings, so users do not need to change anything. GDPR calls the pair "data protection by design and by default" in Article 25.
Here is what you need to know first:
- Privacy by design and data protection by design describe the same idea. Ann Cavoukian popularised "privacy by design." The GDPR uses "data protection by design and by default."
- The duty falls on the data controller. The controller decides why and how personal data is processed.
- The data subject is the person the data relates to. Article 25 exists to protect their rights.
- The EDPB says the duty applies to every controller, including small businesses.
- The duty also covers systems that existed before the GDPR. Old platforms need review.
Why Privacy by Design and Privacy by Default Matter
They reduce legal exposure, breach costs, and the need to rebuild systems later. Regulators also treat poor defaults as violations in their own right. These figures show the scale:
- Fines: Breaching Article 25 can cost up to €10 million or 2% of worldwide annual turnover, whichever is higher. This limit sits in Article 83(4)(a) of the GDPR (opens in a new tab).
- Enforcement: European regulators issued about €1.2 billion in GDPR fines in 2025, according to the DLA Piper GDPR Fines and Data Breach Survey (opens in a new tab). Total fines reached about €7.1 billion between 25 May 2018 and 10 January 2026.
- Breach volume: Regulators received an average of 443 breach notifications per day in the year to January 2026. That is 22% more than the year before.
- Breach cost: The IBM Cost of a Data Breach Report 2026 (opens in a new tab) puts the global average at $4.99 million. That is a record and 12% above last year. The US average is $11.5 million.
Article 25 also builds in proportionality. You weigh the state of the art, the cost, and the risk to people. Also, you do not need the most expensive control. You need an effective one.
Origins of Privacy by Design
Ann Cavoukian developed privacy by design while she served as Information and Privacy Commissioner of Ontario. In October 2010, regulators met in Jerusalem for the 32nd International Conference of Data Protection and Privacy Commissioners. They adopted a resolution on privacy by design (opens in a new tab).
The resolution recognised it as an essential component of fundamental privacy protection. It encouraged organisations to make privacy their default mode of operation. The GDPR turned the idea into law. Regulation 2016/679 has been in force since 25 May 2018, and Article 25 carries the duty.
What Is Privacy by Design?
Privacy by design means choosing technical and organisational measures (TOMs) at the planning stage and keeping them effective while the system runs. Article 25(1) names pseudonymisation as one example. The measures must put principles such as data minimization into practice.
Four measures appear in most designs.
Encryption Protocols
Encryption makes data unreadable to anyone without the key. Encrypt data in transit and at rest. Article 32 of the GDPR lists encryption as a security measure.
End-to-end encryption goes further. Only the sender and the recipient can read the content. The service operator cannot. You must choose this at the design stage because adding it later changes how the whole system works. Our zero-knowledge encryption guide (opens in a new tab) covers the technical details.
Secure System Design
Secure design limits what each part of a system can see. Store identifiers apart from content. Use pseudonymization (opens in a new tab) so records do not point to a named person. Use tokenization for payment data. A token replaces a card number and has no value outside your system.
Keep production data out of test environments. The ICO tells designers to avoid real user data when they build prototypes.
Access Control Mechanisms
Access control decides who can see which data. Role-based access control (RBAC) grants access by job role. Give each role the minimum access it needs. Log every access to sensitive records. Review permissions when people change roles. Remove access on the day someone leaves.
Data Minimization Rules
Article 5(1)(c) requires personal data to be adequate, relevant, and limited to what is necessary. Article 5(1)(b) adds purpose limitation. You collect data for a stated purpose and do not reuse it for an incompatible purpose.
Ask one question about every field: which purpose needs it? Remove the field if you have no answer.
How Does Privacy by Design Prevent Data Privacy Risks?
It removes risk at the source. Data you never collect cannot leak, be misused, or trigger a deletion request. A privacy risk assessment at the start shows where risk sits before you build.
Jaap-Henk Hoepman of Radboud University proposed eight privacy design strategies (opens in a new tab). Engineers use them as a privacy engineering checklist. Four strategies focus on the data itself:
- Minimise: limit the personal data you process.
- Separate: process data in separate places to prevent linking.
- Abstract: limit the detail of the data you process. The original paper called this "aggregate."
- Hide: protect data from view and make it unlinkable.
Four strategies focus on process:
- Inform: tell people about the processing in good time.
- Control: give people control over their data.
- Enforce: commit to privacy-friendly processing and enforce it.
- Demonstrate: show that you process data in a privacy-friendly way.
Which Lifecycle Controls Does Privacy by Design Include?
Privacy by design covers four stages of the data lifecycle: collection, processing, retention and deletion, and third-party sharing.
Controlled Data Collection
Collect only the fields you need. Set the lawful basis for processing before you collect anything. Where consent is the basis, use opt-in consent that people give by a clear action. Avoid dark patterns that push people toward sharing more.
Secure Processing Mechanisms
Encrypt data, pseudonymise it where possible, and restrict access. Keep audit logs. Test security before launch and after each major change. The EDPB asks controllers to review the effectiveness of their measures regularly, not once.
Retention and Deletion Governance
Article 5(1)(e) requires storage limitation. Set a retention period for each type of data. Automate deletion when the period ends. Test that deletion reaches backups and archives. The Deutsche Wohnen case below shows what happens when it does not.
Third-Party Data Protection
Every processor and sub-processor adds risk. Article 28 requires a written contract with each processor. The EDPB states that the controller stays responsible for its processors and sub-processors.
Run vendor due diligence before you sign. Ask where the vendor stores data, which sub-processors it uses, and how it deletes data. Our glossary explains data processors (opens in a new tab), sub-processors (opens in a new tab), and the data processing agreement (opens in a new tab).
The 7 Foundational Principles of Privacy by Design
Cavoukian set out seven foundational principles. Regulators and standards bodies still use them as a reference.
Principle 1: Proactive, Not Reactive. Preventative, Not Remedial
Act before harm occurs. Do not wait for a breach or a complaint. For example, run a data protection impact assessment before you launch a new feature, not after users raise concerns.
Principle 2: Privacy as the Default Setting
Users get full protection without any action. New accounts start private. Analytics stores nothing on the device unless the user agrees. This principle matches GDPR Article 25(2).
Principle 3: Privacy Embedded into Design
Protection forms part of the architecture. It is not an add-on. For example, build a deletion function into the data model on day one instead of writing a script after launch.
Principle 4: Full Functionality: Positive-Sum, Not Zero-Sum
You do not trade privacy for features. Both can work. For example, you can count visits and referrers without storing a persistent visitor identifier. You still get useful traffic data.
Principle 5: End-to-End Security: Full Lifecycle Protection
Protect data from collection until secure deletion. This principle is wider than end-to-end encryption. It covers every stage listed in the lifecycle section above.
Principle 6: Visibility and Transparency: Keep It Open
Tell people what you collect and where you store it. Let others verify your claims. Publish your sub-processors. Open-source code allows independent checks.
Principle 7: Respect for User Privacy: Keep It User-Centric
Put the person's interests first. Give clear notices and simple controls. Support data subject rights, including access, correction, and erasure.
What Is Privacy by Default?
Privacy by default means that, without any user action, only the personal data necessary for each specific purpose gets processed. Article 25(2) applies this to four things: the amount of data collected, the extent of processing, the storage period, and who can access the data. Personal data must not become available to an indefinite number of people without the person's own action.
Privacy-friendly default settings follow from this rule. The user can open things up. The system never does it for them.
Data Visibility
New accounts, profiles, and posts start private. Sharing requires a clear user action. Regulators tested exactly this setting in the Instagram and TikTok decisions covered below.
Behavioural Tracking
Behavioural tracking follows what people do across pages or sites to build a profile. It must stay off until the person opts in. The ePrivacy Directive requires prior consent before a site stores or reads non-essential data on a device.
A cookie banner with pre-ticked boxes does not meet that standard. A cookie wall that blocks access unless people accept tracking raises a second problem. Article 7(4) of the GDPR tells regulators to check whether a service depends on consent that it does not need. Read more in our guide on whether you need a cookie banner for analytics (opens in a new tab) and in the consent banner glossary entry (opens in a new tab).
Location Sharing
Location sharing starts off. Ask at the moment a feature needs it. Use approximate location when it is enough. Precise location trails can reveal special categories of data, such as health or religion.
Third-Party Integrations
Integrations start disabled. Each one adds a recipient of personal data. Show the user what an integration receives before they enable it. Pass the minimum data.
Data Minimization: Collecting Only What Is Necessary
Optional fields stay optional and empty. Forms do not ask for a date of birth or phone number without a purpose. Article 25(2) ties the default to the amount of data collected, so every extra field needs a reason.
Automatic Privacy Protection Through Default Settings
Protection should work without user effort. Examples include automatic retention expiry, default encryption, and short session timeouts. The ICO notes that privacy by default does not always mean switching everything off. The default must match what the purpose needs and no more.
What Is Not an Implementation Strategy for Privacy by Default?
Relying on users to find and change settings is not an implementation strategy for privacy by default. The system must protect people before they act. These approaches fail the test:
- Pre-ticked consent boxes or sharing toggles.
- Public-by-default profiles, especially for children.
- Collecting data "just in case" with no named purpose.
- Controls hidden deep in menus or worded to discourage a choice.
- Keeping data with no deletion date.
- A privacy policy as the only safeguard, with no change to system behaviour.
The CNIL's innovation lab (LINC) flagged two ways to push people into sharing more than they need. One is pre-checked boxes. The other is a promise of a better experience. The EDPB notes that default settings cover choices made by controllers and by data subjects. Your starting values come first.
Privacy by Design vs Privacy by Default
Privacy by design is how you build the system. Privacy by default is how you configure it when it ships. Design covers architecture and controls under Article 25(1). Default covers starting settings and data scope under Article 25(2).
| Aspect | Privacy by design | Privacy by default |
|---|---|---|
| Core question | How do we build protection into the system? | What does the system do before the user changes anything? |
| GDPR clause | Article 25(1) | Article 25(2) |
| When it applies | At planning and while processing runs | At first use and with every new feature |
| Who acts | Architects, engineers, the DPO | Product, UX, and engineering teams |
| Typical measures | Encryption, pseudonymisation, access control, deletion logic | Private profiles, tracking off, optional fields empty, short retention |
| Failure example | An archive that cannot delete data | Child accounts that are public by default |
| How to test it | Can you show the control works? | Does a new user get the minimum with zero clicks? |
Is Encrypted Data Processing Privacy by Design or Privacy by Default?
Encryption is a design measure. You choose it when you plan the system. Encryption that applies to every user without any action also qualifies as a default.
For example, an end-to-end encrypted messaging app that never allows encryption to be switched off meets both. The design makes the operator unable to read messages. The default means no user has to enable it. Encryption alone is not enough. You still need minimisation, access control, and deletion.
Why Privacy by Design and Default Must Work Together
The EDPB describes the two concepts as complementary. They reinforce each other. People gain more from strong defaults when the design is also sound, and the reverse holds too.
Secure Systems Can Still Fail Without Strong Defaults
A well-built system can still expose people. The Instagram decision below shows this. Children's accounts started public, and the exposure came from a setting, not from an attack.
Strong Privacy Defaults Cannot Fix Insecure Systems
A private account on a system with open admin access still leaks. The Deutsche Wohnen case below shows a design failure. The archive could not delete data, so no default setting could fix it.
Real Protection Requires Both Design and Default
Check three things before launch:
- The architecture limits access and supports deletion.
- A new user starts with the minimum data scope.
- You can show evidence for both.
Privacy by Design and by Default Under the GDPR
GDPR Article 25 requires controllers to apply data protection by design under paragraph 1 and by default under paragraph 2. Recital 78 supports the rule. You can read the full text on EUR-Lex (opens in a new tab).
What Does GDPR Article 25 Require?
These are the GDPR Article 25 requirements in plain terms:
- Article 25(1): Apply appropriate technical and organisational measures. Apply them when you decide how processing works and while it runs. Consider the state of the art, the cost, the context, and the risks to people.
- Article 25(2): Process only the data necessary for each purpose by default. This covers the amount collected, the extent of processing, the storage period, and accessibility.
- Article 25(3): An approved certification mechanism under Article 42 can help you show compliance.
- Recital 78: Producers of products and services should consider data protection when they design them. Controllers should pick providers that do.
The measures must put the Article 5 principles into practice. These include lawfulness, purpose limitation, data minimization, and storage limitation. Article 25 does not replace the need for a lawful basis for processing under Article 6.
What Do the EDPB Guidelines 4/2019 Say?
The European Data Protection Board (EDPB) adopted its Guidelines 4/2019 on Article 25 (opens in a new tab) on 13 November 2019 for consultation. Version 2.0 followed on 20 October 2020. The key points are:
- The duty applies to all controllers, regardless of their size.
- It applies to existing systems, which controllers must keep updated.
- Controllers must apply the measures before processing and review them continually.
- Controllers remain responsible for their processors and sub-processors.
- The guidelines list key design and default elements for each Article 5 principle.
- The EDPB encourages certifications and codes of conduct, and it gives extra recommendations for small and medium enterprises.
What Is the Fine for Breaching Article 25?
The maximum is €10 million or 2% of worldwide annual turnover, whichever is higher, under Article 83(4)(a). Regulators often add findings under Article 5. Those fall in the higher tier of €20 million or 4% of turnover under Article 83(5)(a). These administrative fines can apply to any controller.
Privacy by Design and Default Under the Swiss FADP and UK GDPR
The revised Swiss Federal Act on Data Protection (FADP) (opens in a new tab) requires both. Article 7 has applied since 1 September 2023. The Swiss FDPIC explains the rule (opens in a new tab) in two parts. Privacy by design means applications must process data in line with the FADP from the planning stage. Privacy by default means settings must limit processing to the minimum needed, unless the person decides otherwise.
| Item | GDPR | Swiss FADP |
|---|---|---|
| Provision | Article 25 | Article 7 |
| In force | 25 May 2018 | 1 September 2023 |
| Design duty | Measures at planning and during processing | Measures from the planning stage |
| Default duty | Only necessary data by default | Minimum processing by default |
| Regulator | National supervisory authorities and the EDPB | The FDPIC |
The UK GDPR contains a matching duty. The ICO updated its guidance on data protection by design and by default in February 2026. The update followed the commencement of the Data (Use and Access) Act. Read our guide on why Swiss infrastructure matters for data privacy (opens in a new tab).
Real-World Examples of Privacy by Design and Privacy by Default
These privacy by design examples show what regulators check. One is an illustrative scenario. Three are real decisions.
Social Media Example: Implementing Privacy by Default
Case study 1: Instagram (2022). The Irish Data Protection Commission (DPC) adopted its decision on 2 September 2022 and published it on 15 September 2022. It fined Meta Platforms Ireland €405 million across several infringements. The findings included Articles 25(1) and 25(2).
The DPC examined two issues. Instagram's business account feature publicly showed the email addresses or phone numbers of some children. Personal accounts of children were public by default. The EDPB issued a binding decision on 28 July 2022 during the process. Meta stated that it had changed the default settings and would appeal. See the DPC press release (opens in a new tab).
Case study 2: TikTok (2023). The DPC adopted its final decision on 1 September 2023 and announced it on 15 September 2023. It fined TikTok Technology Limited €345 million in total. The inquiry covered 31 July to 31 December 2020. The DPC found that public-by-default settings for child accounts breached Articles 25(1), 25(2), and 5(1)(c). The largest part of the fine, €180 million, covered transparency failures, according to the IAPP. See the DPC decision page (opens in a new tab).
What both cases teach you:
- Audit the defaults for every account type, especially accounts held by minors.
- Make sure your DPIA covers the real risks, including users below your minimum age.
- Keep a record that shows why each default exists.
Data Retention Example: Deutsche Wohnen (2019 to 2026)
What happened. The Berlin data protection authority inspected the German property company Deutsche Wohnen SE in June 2017 and March 2019. It found an archive that stored tenant data with no way to delete it. On 30 October 2019, it fined the company about €14.5 million for breaching Articles 25(1) and 5. The authority had warned the company about the archive in 2017. See the DLA Piper analysis (opens in a new tab).
What happened next. The Court of Justice of the EU ruled in case C-807/21 that fines can apply directly to companies. They apply only where the company acted with fault. On 9 June 2026, the Berlin Regional Court I confirmed the breach and cut the fine to €900,000. The judgment is not yet final, according to Legal Tribune Online (opens in a new tab). The court noted that the company had worked on a technical fix, but also said it could have moved faster.
The lesson. The court reduced the amount but kept the finding. Deletion must exist in the design. Fix known audit findings before they reach a regulator.
Web Analytics Example: Measuring Traffic Without Identifiers
Traditional analytics sets a cookie to recognise returning visitors. A privacy-first design counts visits from request-level signals and stores no persistent identifier. It sets no cookie, so no cookie banner is needed for that measurement. The data stays in one region, and the team collects only what each report needs. Learn how this works in our guide to cookieless analytics strategies (opens in a new tab) and our checklist on GDPR-compliant analytics (opens in a new tab).
How Do You Implement Privacy by Design and Default? 8-Step Checklist
Follow these steps to implement privacy by design in a product or a website. They also meet the default rule.
- Map your data. List what you collect, where it lives, who can access it, and why you hold it.
- Define purpose and lawful basis. Write one purpose and one lawful basis for each data item.
- Minimise. Remove every field that has no purpose. Apply Article 5(1)(c).
- Run a privacy risk assessment. Complete a DPIA before building when Article 35 or your regulator's list applies.
- Set the defaults. Make profiles private, tracking off, optional fields empty, and retention short. Test as a new user with zero clicks.
- Add technical and organisational measures. Use encryption, pseudonymisation, RBAC, and logging. Involve privacy engineering in the design review.
- Control retention and deletion. Automate deletion and test it on backups.
- Govern vendors and keep evidence. Sign Article 28 contracts, track sub-processors, and run vendor due diligence. Record each decision. Consider certification mechanisms under Article 42.
The ICO has a six-phase guide that follows a product from kick-off to post-launch. Use it to schedule these steps in your delivery plan.
How Pulse Analytics Applies Privacy by Design and by Default
Privacy-focused analytics is a practical example of how organisations can apply data protection principles to everyday technology.
Pulse Analytics, Ciphera's web analytics product, follows both principles. The table shows how.
| Privacy principle | How Pulse Analytics applies it |
| Data minimisation | Sets no cookies and keeps no visitor identifier past the end of the day. |
| Privacy by default | Stores nothing on the visitor's device, so in many countries a site using it needs no cookie banner. |
| Data protection | Stores its data on servers in Switzerland. |
| Transparency | The dashboard's code is public on GitHub under AGPL-3.0, and the command-line tool under Apache-2.0. |
| User control | Provides data export options and the ability to delete analytics data. |
| Purpose limitation | Focuses on website performance measurement rather than advertising-based behavioural profiling. |
Try Pulse Analytics
Pulse measures your traffic without cookies, and the free plan covers one website with 10,000 events a month.