Score your own system on our eight steps.
Built for compliance, risk and product teams. Answer for the affected person, not for your staff, then give each step the score our codebook would give it from your documents. This page computes the average, the weakest step and whether standing is blocked. Nothing you enter leaves your browser. This is the same scale our index uses, applied to one role without the operator, without a second coder and without a check of your documents. It is a first reading, not an assessment. A result from this page may not be published or described as a Human Option score; only profiles published by the index carry that name.
4: exists in practice and works, shown by a binding document. 3: exists with a material limit. 2: on paper only, or through an outside route. 1: a trace. 0: shown absent. ND: you have no document either way. Company statements alone cap a score at 2.
Can the affected person see the data used about them, by right, before or at the decision?
Is the person told, at the decision, that an automated system took part and what it did?
Can the person be heard before the decision is final, and does a named role hold documented authority to override the output?
Is the person told that a decision was made, at the time it is made?
Are the reasons specific to the person and sufficient to act on?
Can the person correct the inputs, and is the decision re-run when they do?
How deep does recourse go: none, automated reconsideration, human review without authority, human review with authority, or independent contestability?
Can the person refuse the automated path without losing the service, the history or their standing?
Colorado's automated decision law (from 1 January 2027) and California's automated decision-making rules map onto steps 2, 4, 5, 6 and 7. A 0 or ND on any of those is where to start. Codebook · Methodology · Ask about an independent assessment