# Acceptance criteria are part of the interface

> Scope fails like an ambiguous UI when completion evidence is unnamed — treat done-when as a control so fixed-scope delivery stays finishable and verifiable.

- Category: Practice
- Reading time: 5 min
- Published: 2026-09-05
- Updated: 2026-09-05

Scope fails the same way an ambiguous UI fails: different people form different models of the system, then argue about whether the work is finished. Writing "build the thing" is not a contract. Naming what evidence will show completion is. Design engineering without the handover already treats scope as an interface: available actions clear, invalid states constrained, completion meaning shared. This note sharpens the acceptance face of that interface.

## Done-when is a control, not a footnote

A useful acceptance criterion is observable. It names the artefact, the state, and how someone who was not in the room can verify it. "Looks better" is not acceptance. "Organisation node resolves to one `@id` across page schema, feed and site index" is.

Brittek's engagement model makes that separation explicit. The System Diagnostic produces validated findings, prioritised remediation and an implementation-ready brief; its scope boundary stops at analysis and planning. Fixed Scope Delivery is separately bounded: explicit scope, price, acceptance criteria and principal delivery. Crossing that boundary without a new acceptance set is how drift returns.

## Evidence before proposal, proof after change

Diagnosis captures evidence before calculation. Implementation should close with verification that the changed evidence produces the expected state. Acceptance criteria are the hinge between those stages: they turn a remediation directive into a re-runnable check rather than a persuasive paragraph.

## Why this compounds



- Clients can dispute or confirm completion without re-litigating intent.
- Specialists can contribute without owning the outcome definition.
- Journal, diagnostic and engagement language stay aligned: observation, recommendation and verification remain distinct.

## What this is not

This is not a plea for heavier process theatre. It is the minimum interface so that fixed-scope work remains finishable — and so that "done" cannot be redefined after the fact to absorb unscoped desire.

## Index

- [Journal](https://brittek.net/journal)

Canonical URL: https://brittek.net/journal/acceptance-criteria-are-part-of-the-interface