Integrity holds when five things line up: what should be possible, what is possible, what you tell people, what you can show and what is still true. Each gap is a break between two of them.
Between what should be possible and what the system allows
Check whether people, services and agents can do more than their role requires.
Can one customer open another customer’s record?
Keep: the intended access rule and a scoped test with synthetic identities, including what was not tested.
Between what the system does and what you tell people it does
Trace what happens to data and compare it with the promise the business makes.
Does deletion remove the data from every place the policy describes?
Keep: a data-flow map, the exact claim and observed results for the systems in scope.
Between what you say and what you can show
Connect each conclusion to an observed result, a configuration, a document or a clearly attributed statement.
Which test supports this claim, and what did it leave untested?
Keep: the procedure, result, date, environment and the reasoning behind the conclusion.
Between what was shown and what is true now
Keep the build, date, scope and reasons to reassess attached to the evidence.
Does the last review cover the feature you just shipped?
Keep: the assessed build, the relevant configuration, later changes and the trigger for another check.
One finding can touch more than one gap. The Four Gaps organize the work; they sit alongside established security practice and the responsibility of the people who run a system.
Further reading
A free introduction to Permission, Behavior, Proof and Drift, with a worked example and a worksheet.
The method organizes an examination. It does not replace technical testing, establish compliance or certify a system. Record the build, scope, date, source and untested conditions with each conclusion.
Inspect an example assessment report →
The Field Kit pairs the method with an editable workbook and a register of 244 controls. Read a sample evidence record from the kit.