From 30% to 90%, surfacing data worth trusting

The legacy report surfaced only 30% of the data points a compliance officer actually needs to make a decision. In practice, that meant a handful of business fields (status, jurisdiction, a single address, incorporation date, one SIC code), a table of officers (role, appointed date, date of birth, occupation, correspondence address), a flat list of UBOs, and screening with ongoing monitoring that only ever flagged screening updates. No financials, no risk indicators, no adverse or insolvency data, no way to see how ownership actually connected. The other 87% lived elsewhere: in public registers, in supplier data nobody had wired up, in spreadsheets officers kept on the side.

Worse, what it did surface was often wrong. Labels and officer counts regularly disagreed with the public register, and once one figure was wrong, officers stopped trusting the rest of the report, including the screening results and the ownership information.

The trust gap forced officers into a backwards routine. They'd pull a report, then leave the platform to verify the same data against other sources, because they didn't trust it enough to act on alone. If they found a discrepancy, they'd resolve it themselves and record the corrected figure directly in their own case management system. The report's only remaining value was that it was easier to file into that system than the sources they use to validate it.

In practice, officers were doing the due diligence twice: once against Thirdfort's report, and once for real. For a product meant to speed up risk decisions, that was backwards enough on its own. It was also an audit liability, the report on file, the one a decision could later be traced back to, was frequently not the version the officer had actually trusted.