- Fintech
- Custom App Development
- Web Design
- ML & Data Science
The compliance screen people give up on
Identity and anti money laundering checks are not optional, and the screen that runs them is reliably the biggest drop off in a financial product. Reordering what is asked, and when, recovers a meaningful share of applicants without weakening a single control.
Representative engagement. This describes a pattern we build rather than one named client: the situation that produces it, how we approach it, and the range of outcomes that kind of work lands in. Figures are stated as ranges or targets, never as a measured result for a specific customer. Our named client work is on the work index.
- Industry
- Fintech
- typical onboarding drop off recovered by resequencing (target range)
- 20 to 40%
- no check weakened or removed to get there
- Same controls
The problem
A typical onboarding flow asks for everything up front, because that is the order the compliance requirements were written in. The applicant meets a wall of fields before seeing anything of value, and a good proportion of them close the tab. The other half of the problem is what happens on a partial match. A rules engine tuned to be safe refers a large number of legitimate applicants to manual review, and manual review means a queue, and a queue means an applicant waiting days for a decision on a product they wanted immediately. Nobody can loosen the controls, so the only room available is in sequencing, in how a partial match is resolved, and in what the applicant sees while they wait.
What we built
The flow is rebuilt around when each piece of information is actually needed rather than around the order the policy lists it. The applicant reaches value before the heaviest requirements arrive, and each step explains why it is being asked, which measurably matters on identity screens. Checks run progressively. Data available from the provider is not asked for again, document capture is guided rather than left to a file picker, and a soft failure prompts a retry in context instead of a rejection. Review is treated as a product surface rather than a spreadsheet. Cases arrive prioritised with the evidence assembled, so a reviewer decides rather than gathers. Every decision, automated or human, is logged with its reasoning, which is what turns an audit into a query.
What changed
Completion improves because the flow stops asking for everything before giving anything, not because anything was waived. Every control that ran before still runs. The second effect is on the review queue, where the gain comes from assembling the evidence rather than from approving more cases. Both figures below are the ranges this pattern is scoped against, and both depend heavily on the product and the risk appetite behind it.
Built with
- Next.js
- TypeScript
- PostgreSQL
- Persona
- GA4