You can infer what an organisation measures from the layout it ships, without access to anybody's intentions or documents. Five signals to look at, why asymmetry of effort is the reliable one, and how to state the reading as evidence rather than as an accusation.
What this method is for
Not to catch anybody. To answer a practical question before you depend on something: what is this product built to produce?
The layout will tell you, because a layout is a set of decisions about relative cost and those decisions were made by people optimising something.
Everything below is observable from outside, requires no source access, and takes about ten minutes.
Five signals
Relative prominence. One option is a filled button, the other is grey text. Nothing has been concealed. Something has been weighted, and the weighting is the finding.
Count of actions. Walk each path and count. One tap against four is not an implementation detail; it is the shape of the product's preference expressed in the only unit users actually spend.
What is pre-selected. Whatever is already ticked will be chosen by most people. Which makes the choice of default consequential in a way that deserves its own treatment, and gets it elsewhere.
Depth of the reversal. Where does undo live — beside the action, or four screens in, under a different noun? The distance is a statement about how often the product expects people to change their minds, and how much it would like them to.
Symmetry of wording. Are both options written in the same register, or is one phrased as a loss? No thanks, I don't want to save money against Continue is not a pair of options. It is one option and a discouragement.
Asymmetry is the reliable one
If you only do one of these, do the counting.
If accepting and declining were genuinely equivalent to the business, they would cost the user the same. They almost never do.
Any difference in effort is therefore a measurement of preference — taken by the organisation, recorded in the product, and legible to anyone willing to walk both paths. It is a measurement that exists whether or not anybody sat down to make it, which is what makes it good evidence: it cannot be adjusted for the benefit of an observer who was not expected.
Why this is not cynicism
Teams optimise what they measure, and what gets measured is completion of the flow.
An interface producing more acceptances looks better on the dashboard the team is judged on. A designer meets a brief. A product manager hits a target. An engineer implements a spec that arrived with a ticket number.
Nobody in that chain has to intend anything for the pattern to appear. Which makes this a governance problem rather than a moral one — and governance problems are the better kind, because they have structural fixes, while moral ones only have arguments.
Stating the reading
The discipline is to report what you observed and stop there.
A cancellation requiring a phone call while signup takes thirty seconds: acquisition is instrumented and departure is not.
Export buried three levels down while import is on the front page: the product is built to move data in one direction.
A notification setting that returns to its default after each update: the setting is treated as a preference the product may revisit rather than an instruction it received.
None of those requires a claim about motive, and none of them is weakened by not having one. Motive is unavailable and unnecessary; the behaviour is right there, reproducible by anybody who repeats the walk.
What the reading looks like when it comes back clean
Worth knowing, so the method can return a positive result rather than only ever indicting.
Parity of effort between the two paths. The reversal in the same place as the action. Symmetric wording. State that survives an update rather than quietly expiring.
When you find those together, they are unlikely to be accidental. Each costs something to build and none of them improves a completion metric, which means somebody argued for them against a number.
For the people building it
One question, and it does not need an ethics review.
Would you be comfortable showing this flow to the user you are hoping will not read it carefully?
If the flow depends on inattention, the design has already reached a conclusion about its users. You can see it on the screen, which means so can they, eventually, and probably at the worst moment for you.
Close
Walk both paths and count. The number is the product's opinion of you, stated in the only language that ships.