Back to notes

July 22, 2026 · 5 min read

Data Minimization Is a Feature, Not Just a Compliance Constraint

Teams treat data minimization as a regulatory tax to be minimized. I've started pitching it as a product decision that makes systems faster, cheaper, and easier to reason about.

PrivacyHot Take

Every time data minimization comes up in a project kickoff, it gets framed as a constraint legal is imposing on what the product team actually wants to build. I think that framing is backwards, and I've started actively arguing against it in scoping calls.

A client's event schema had grown to collect full name, email, phone, address, and a handful of behavioral fields on every single tracked event — not because any downstream system used all of it, but because someone years ago decided to 'capture everything in case we need it later.' Nobody ever went back to prune it. The event payloads were enormous, the warehouse table was bloated, and every new integration had to reckon with a schema five times bigger than it needed to be.

What minimizing actually bought them

We cut the event schema down to what each specific use case actually consumed — session and behavioral data in the event stream, PII joined in separately at query time from a single system of record. Payload size dropped by more than half. Warehouse costs dropped with it. And every engineer who touched the event pipeline afterward had a much smaller mental model to hold, because the schema's shape now reflected what it was actually for.

Collecting less data isn't just lower legal exposure. It's a smaller surface area for every engineer who has to reason about the system after you.

Where this pays off beyond compliance

  • Smaller breach blast radius if something does go wrong — you can't leak a field you never collected
  • Fewer downstream consumers accidentally depending on PII that should have stayed out of the event layer
  • Cheaper storage and faster queries, since minimization and performance optimization overlap almost completely
  • Simpler consent logic, because a thinner data footprint has fewer categories to gate behind opt-in rules

The pitch that actually lands with product teams isn't 'this reduces our GDPR risk.' It's 'this makes the system smaller and faster, and the compliance benefit comes along for free.' Data minimization sold as an engineering decision gets adopted. Data minimization sold as a legal mandate gets resented and worked around.