Pricing fairness review

What it does

It tests whether a pricing model charges people fairly. The prices and offers the model actually produces are analysed across customer groups, including by tenure, age, vulnerability and protected characteristics, and each material difference is traced to its cause. Differences explained by cost or risk stand; those explained by a customer's inertia or weakness do not. The pricing owner is accountable; product governance and the second line carry out or challenge the review. It operates before a model is used and at least annually after.

How to implement it

• List the model's inputs and classify each as cost, risk or behaviour.
• Exclude or constrain inputs that measure willingness to pay.
• Compare prices for customers of similar cost and risk across groups.
• Compare new and existing customers for the same product.
• Set limits on permitted differences and build them into the model.
• Report findings to product governance and act on them.

Evidence it operates

Design: a pricing fairness standard and review procedure. Operation: review reports with analysis by group; decisions on each unexplained difference; model constraints in place; fair value assessments that refer to the review.

Where it fits

At validation and in operation, as part of product governance. It is the primary control for unfair personalised pricing and supports poor customer outcomes from AI-driven journeys on price and value. It draws on bias testing and fairness monitoring for the discrimination analysis.

Limits

Whether a difference is justified is a judgement, and cost and risk arguments can be constructed for most things. Complex models make it hard to say why one customer was charged more. The review sees the organisation's own prices and cannot see whether competitors' algorithms are moving in step, which is a separate risk.