Privacy

Privacy by design: why we build products that see less

It's easy to build software that collects everything just in case. It's harder, and better, to decide what not to collect. Here's how privacy by design works in practice.

Software has a habit of collecting everything it can, just in case it turns out to be useful. Every click, every field, every location ping. Storage is cheap, the thinking goes, so why not keep it all?

Because every piece of personal data is also a responsibility and a risk. It can be lost, leaked or misused, or simply make people feel watched. The most trustworthy products we know make a different choice: they decide what not to collect.

UK GDPR has a name for this. Article 25 requires data protection by design and by default. In plain English, privacy should be built into how a product works from the first sketch, and the most private option should be the starting point rather than a setting people have to go looking for.

What it means in practice

Privacy by design isn't one feature. It's a set of habits:

  • Collect the minimum. Only gather data you need for a clear purpose.
  • Use it only for that purpose. Don't quietly repurpose it later.
  • Make private the default. Sharing should be a choice people make, not one made for them.
  • Show your working. Explain what you collect and why, in words people understand.
  • Give people control. Let them see, export and delete their data.
  • Log access. When someone looks at personal data, record who, what and when.
  • Delete on schedule. Data you no longer hold can't leak.

Here's how those habits look in three projects we're involved in.

Contribio: summaries, not surveillance

Contribio is a participation co-operative for universities, rewarding students for contribution such as team projects, short lessons and helping at events. A platform like that could easily become a monitoring tool. It's designed not to be.

University staff see a summary of engagement rather than a minute-by-minute record. Opening a student's full record requires a stated reason, and every view is logged on the student's own profile. Sponsors see outcomes, never individual students. Nothing outside the app is tracked. And when a student wants to show an employer what they've achieved, they choose who receives a signed credential and what it shows.

The result is a system students can trust, which is exactly what a participation platform needs in order to work.

Negotium Pro: your position stays yours

Negotium Pro is negotiation training and live coaching for procurement teams, provided by Klarintel Limited and Liberty Web Ltd. Its users work with commercially sensitive information: targets, walk-away points and supplier details.

So, on a live call, the other party's speech is transcribed locally in the user's browser by default, and only the text is saved. Research on a counterpart comes from public sources, is reviewed by the user field by field, never infers sensitive personal traits and expires after 30 days. Users can read exactly what each AI coach knows about them and make a coach forget them with one click. When an administrator opens account content for support, the access is recorded. And users can export or delete everything from their settings.

Fashion Without Cruelty: no tricks in the inbox

Privacy isn't only about databases. It's also about how you treat people's attention. Fashion Without Cruelty's newsletter, The Thursday Note, uses subscribers' addresses only for the note, offers one-click unsubscribe and avoids "last chance" subject lines and sponsored placements dressed up as editorial. Where the site earns commission from a link, it says so.

And this website

When we rebuilt libertyweb.uk, we applied the same thinking. The site serves its fonts from our own server rather than a third party, runs no advertising or tracking scripts, and the contact form asks for only what we need to reply.

A privacy by design checklist

Use this at the start of your next website, app or AI project:

  • Map the personal data you plan to collect, and write down why you need each item.
  • Remove every field and tracker you can't justify.
  • Choose private defaults, and make sharing an explicit choice.
  • Decide how long you'll keep each type of data, and automate deletion.
  • Limit who can access personal data, and log when they do.
  • Write a privacy notice in plain English, and test it on someone outside the project.
  • Check your suppliers: where data is processed, and what their agreements say.
  • Plan how you'll handle requests to access, export or delete data.
  • Carry out a data protection impact assessment if the processing is likely to be high risk.

The Information Commissioner's Office publishes detailed guidance on data protection by design and by default, and on impact assessments, which is well worth reading before you build.

Trust is the product

Privacy by design can feel like extra work at the start. In our experience it pays for itself. Products that respect people are easier to explain, easier to defend and easier to love. And when most software wants more of your data, choosing to see less is a genuine advantage.

If you're planning something that handles personal data, get in touch. We're always happy to talk it through.

Written by Liberty Web

Liberty Web Ltd is a British web agency building technology for good: websites, AI agents and automation, ecommerce and website support.

Building something that handles personal data?

We help organisations design websites, apps and AI agents that respect the people who use them.