How to choose an ethical tech consultancy without buying theatre

Ethical technology work fails in a predictable way. A set of principles is agreed, a committee is created, a policy is published, and the systems that make the decisions carry on unchanged. Everyone is sincere. Nothing is different.

The test is simple. Ask what will be materially different in the running system when the engagement ends. If nobody can name a control, a log, a threshold or a refusal path that did not exist before, the engagement is theatre.

Who on the team has shipped and operated a production system, not just advised on one? What does your deliverable change in code? How do you handle the case where the ethical answer is to not build the feature? What evidence does the system produce after you leave? Who maintains the controls you install? And what have you refused to build, and why?

The last question is the most revealing. A consultancy with no refusals has no line.

Our position is that ethics and governance advice only holds when the same team can implement it. We build and run our own systems with our own capital, which means the standards we recommend are standards we have already had to live with in production.

That is the whole argument for hiring a builder rather than an advisor: the recommendation and the implementation cannot drift apart if they come from the same hands.