The preparatory step is to identify what matters to you as a practitioner, your company as a product owner, and your prospective customers as purchasers and users of the product. As this is an ethically focussed method, these are the ethical properties that the design should have, that distinguish a "good" design that provides a trusted transparent collaboration with the user, from a "bad" design that does not.

The first analysis step is to identify which of the desirable properties relevant to your product or service might be vulnerable in a way that creates a precarious situation for the user, namely a jeopardy. The scope of the property should be within the influence of the design, but need not be within the system boundaries and need not be vulnerable in all circumstances or for all users. The vulnerability should be foreseable given the known or suspected limitations of the user research and design activity so far. The aim is a qualitative analysis, not a quantitative assessment of the numbers affected or the tolerability of the potential harms caused if the vulnerable property is actually compromised or lost.

The second step is to identify evocative words, phrases, images, or objects from the domain of the design that provoke a creative response to the vulnerable property. Each provocation starts a conversation about the problem and helps communicate it to the team.

Step three is to iterate over all the vulnerable properties, supported by their provocations, to identfy if there is a potential issue. If there is, the team discuss it, and decide if it should be confirmed as a threat to a desirable property that needs to be captured as a new jeopardy to monitor and mitigate, or added to the description of an existing jeopardy, or discarded as a broadly tolerable nuisance that requires no further action beyond your normal user support process. The focus of the discussion should alternate between the property, and the causes of its vulnerability, and the people affected by it, swapping between people and property as many times as needed to achieve a shared understanding.

Step four of the method is to structure all of the discussion from the previous step into a visualisation of the problems identified. It is quite likely this will prompt further discussion and possibly useful insights into the reasons and mechanisms, so some iteration may be needed. One suitable visualisation would be a bowtie diagram.

The final step is to communicate the findings to everyone who needs them. This should be in a form they can use through the product life-cycle, and extend and update as new information comes to light.