A consumer finance product must balance fraud controls against onboarding friction. How should that trade-off be made rather than argued?

How to make it rather than argue it

Quantify both sides in the same currency. Fraud losses and false-decline losses are both money, and both are measurable. A control that prevents a certain amount of fraud while rejecting a larger amount of legitimate business is a net loss, and that is an arithmetic question rather than a values one.

The number most often missing is the cost of the false positive — the legitimate customer rejected, who does not complain and does not return. It is invisible in fraud metrics and frequently larger than the losses being prevented.

The design that avoids the binary

Graduate the friction by risk and by value, rather than applying one policy to everyone:

• Low risk, low value: no additional friction.
• Higher risk or higher value: step-up verification, applied only where the expected loss justifies the abandonment it will cause.
• Highest risk: refuse, or route to manual review.

The crucial property is that the friction is proportional to what is at stake, which means a single policy for all values is either too permissive at the top or too restrictive at the bottom.

What must be measurable for this to work

• Approval rate, fraud rate and abandonment rate, segmented by risk band, channel and customer cohort.
• The outcome of stepped-up cases, since a step-up that almost always passes is friction with no protective value.
• Cohort behaviour over time, because the effect of onboarding friction shows up in retention rather than in the funnel.

The organisational requirement

The decision belongs to risk and product together, and it must be recorded. An engineer choosing a threshold is making a business decision with insufficient information, and a threshold with no recorded rationale gets adjusted by whoever is currently unhappy.

The framing that ends the argument

"This control prevents X in expected losses and costs Y in expected abandonment." Once both numbers exist, the disagreement moves to the estimates — which is a productive place for it, and one where evidence can be gathered.