← All writing

Assurance · 2 min read

Audit evidence that keeps up with delivery.

Turning recurring assurance work into repeatable engineering.

Audit evidence is easier to maintain when it comes from the work itself. A deployment record, an access review, or a completed change can already contain much of what an assurance process needs. The architectural task is to connect that record to a clear control objective.

Start with the claim you need to support. “Changes are reviewed before production” is more useful than “we have a delivery pipeline.” The first statement suggests what evidence would demonstrate the control and what would count as an exception.

Define what the record proves

A useful record has a scope, a source and a time. It should make clear which service or change it concerns, where the information came from, and when the relevant action occurred. A reviewer should be able to understand those boundaries without reconstructing an entire engineering process.

Be precise about the difference between configuration and operation. A screenshot of a required approval setting shows how a system was configured at one moment. A change record with an approval shows a particular instance of the process. The assurance question determines which evidence is useful.

Capture evidence at the source

Where possible, capture a record as part of the workflow that produces it. This reduces the need to recreate events later and keeps the evidence closer to the decision. It also creates a practical ownership question: who maintains the workflow when the service or control changes?

Automating collection still requires care. Access to the evidence, retention periods and sensitive content need explicit decisions. A log that exposes credentials is a new problem, even if it makes a review easier.

Make exceptions visible

An exception should have an owner, a reason and a review point. Keeping it alongside the relevant control record helps distinguish a deliberate risk decision from an unexplained gap. It gives both engineering and assurance teams a common account of what happened.

The goal is a chain a reader can follow: control objective, implementation, evidence, exception and follow-up. When that chain is part of routine delivery, preparing for a review becomes a matter of explaining the work already recorded.

Next: The security automation worth owning. RSS feed

Get in touch

Good security starts
with a conversation.

Have a security challenge, an interesting role, or a different perspective?
I’d like to hear about it.