Case Study: Using Behaviorally-Informed Defaults to Improve Conversion at Loanly

AI · 7 min read

Loanly, a fintech startup offering small personal loans, faced high drop-off on multi-field applications where users hesitated on optional financial details. Product hypothesized that behaviorally-informed defaults—defaults derived from anonymized cohort patterns rather than generic placeholders—could reduce choice paralysis without increasing fraud risk.

The team trained a lightweight model on historical, opt-in user data to suggest sensible defaults for fields like repayment cadence and typical loan amounts for similar profiles. Designers made defaults transparent: suggested values appeared with a short rationale tooltip and a clear ‘change’ affordance. For higher-sensitivity fields, the UI required explicit confirmation and displayed the privacy rationale. Compliance and fraud teams defined thresholds for when to require additional verification.

The experiment increased application completion by 27% and shortened mean application time by 2.3 minutes. Importantly, fraud rates did not rise because verification gates activated conditionally when defaults deviated from baseline risk signals. Loanly now includes behaviorally-informed defaults as a configurable layer in their application builder, with guidance on transparency, consent, and escalation rules to balance conversion with safety.