Retail Lending Pricing Optimisation
The brief was deceptively simple: three pricing scenarios, one new loan product, one question — which tier converts best? What followed was more complicated, and more instructive, than the product team expected.
A major UK retail lender was preparing to launch a new personal loan product with a three-tier pricing structure. The product team had two competing hypotheses: that customers would accept a modest rate premium for same-day approval, and that brand trust would outweigh rate comparisons for their existing customer base. Before committing to a pricing architecture, they ran the decision through Simulatte. Three scenarios were modeled across 4.2 million synthetic agents calibrated from UK census data, Financial Conduct Authority lending records, and panel interviews with active personal loan applicants.
The conventional wisdom — confirmed by an internal survey — said speed mattered. Same-day approval would command a premium. The data seemed clear. But Simulatte modeled something the survey couldn't: actual decision behavior under real choice conditions. Not what customers say they'd pay, but how they behave when rate, brand, and friction are all present simultaneously.
The 0.6% speed premium performed exactly as the survey predicted — among customers under 40 who actively used comparison sites. But that segment was already the lender's most loyal, lowest-margin cohort. The premium wasn't extracting value from new customers. It was monetizing existing ones who would have converted anyway.
The real finding was in the 45+ segment. These customers — the highest-value, longest-tenure accounts — barely registered the speed benefit at all. What moved them was something else entirely: specific trust signals in the product copy. Not the rate. Not the approval time. The particular phrasing around "your existing account data" and "no re-application required" reduced their switching friction to near-zero — but only when the messaging was personalized to their relationship tenure.
The simulation also surfaced a counterintuitive finding about the pricing band itself. The standard-rate product was underperforming because it was priced within the margin of error for comparison site filtering — effectively invisible in aggregate rankings. A modest 0.3% reduction moved it into the top-three visibility position on major comparison sites, dramatically increasing the inbound pipeline from price-sensitive under-35 applicants.
Three pricing scenarios went in. A fourth emerged from the simulation — one the team hadn't considered. The lender adjusted their launch architecture before committing to infrastructure. Estimated revenue impact over the product's first 18 months: £2.1M higher than the original middle scenario.
The simulation also produced verbatim persona responses. Daniel, 38, Senior Accountant, Leeds: "I'd move for a better rate but it has to be worth the admin. Half a percent isn't worth an afternoon on the phone with a new bank." Margaret, 57, Retired Teacher, Birmingham: "I've been with the same bank for 30 years. They know me. That's worth something that no rate discount replaces." These weren't anecdotes. They were representative outputs from behavioral clusters the simulation had identified as strategically distinct.
This is what decision infrastructure produces that survey research cannot: not just the headline number, but the behavioral logic underneath it. The mechanisms. The conditions. The words that actually matter to the people who matter.
