Security · Beginner

A threat model in one hour

In one line: Sketch what you have, who might want it, and where it crosses a boundary — and you get a real priority list instead of a gut feeling.

Four questions

What are we protecting? Customer data, money, availability, reputation. Not everything is worth the same.

From whom? An automated scanner, a random attacker, a competitor, an insider, a breached partner. Each has different capabilities.

Where are the boundaries? Every place data crosses trust zones: browser to server, server to database, system to vendor.

What happens if? Describe three bad scenarios in a line each, and rank by damage and likelihood.

Draw before talking

A simple sketch of the components and the arrows between them reveals more than any discussion. Every arrow is a question: who authenticates whom, what is encrypted, what is logged.

The deliverable

One page with five ranked risks, and one suggested action for each. That's enough to start, and it's already more than most systems have.

Going deeper

Repeat the exercise on every significant architectural change — integrating a new vendor, opening an interface, moving to the cloud. A threat model isn't a one-off document but a habit, and done at design time it's dozens of times cheaper than fixing later.

Security Planning & deciding

Next: The ten failures that recur in every audit · Previous: Logs: store correctly and find fast · Back to library