Catching the Fraud a Clean Registry Check Misses

A company can pass a clean credit-bureau check (Active, no registry alerts, no court records) and still be a shell built to borrow and default. Veridion scored 1,007 Italian companies on the observable evidence a registry never sees, surfacing the accounts run by directors with criminal records, under active regulatory enforcement, or operating as letter-box shells that a clean check would wave straight through.

1. 1
   Looked past the registry record

   shells built to borrow and default pass a clean bureau check

2. 2
   Scored 1,007 companies

   on 14 evidence flags across 6 dimensions

3. 3
   Weighed the observable footprint

   physical, digital, reputational and operational evidence, from the graph and live searches

4. 4
   Sourced every flag

   to a court, regulator, review or registry record

5. 5
   Surfaced the bad actors

   9 companies run by directors on a criminal or regulatory record, 24 under active enforcement, plus letter-box shells

6. 6
   Caught what the bureau called clean

   fraud-risk accounts a registry check would onboard, one scoring 100/100 on the evidence

A clean registry record the shells are built to pass

The bureau supplies business-risk data at the moment of onboarding and across SME lending, and its fraud screening sat on internal records and the company registry. For most decisions that is enough. For the fraud that costs the most, it is the wrong place to look.

A registry check confirms that a company is registered. It cannot confirm that the company is real, and the shells and synthetic entities built to borrow and default are designed to pass it: they register cleanly and carry ordinary-looking details.

In the sample, a northern-Italian energy retailer held a spotless bureau record (Active, no registry alerts, a named director, €1.4M revenue, no insolvency or court filings) and turned out to be the worst counterparty in the batch once its footprint was read. The brief was to bring in the external footprint the registry never sees, and use it to separate genuine SMEs from fraudulent entities before credit is extended.

One question: does the evidence support the profile?

Veridion scored every company on one question, drawn from the footprint its company knowledge graph already maps: does the observable evidence support or contradict the declared profile? Each company is read through 14 flags across six dimensions: physical-address integrity, legal-entity status, digital corroboration, the director record, market reputation and real trading activity, with a technology read where a site exists.

The flags sum into a 0–100 score, each dimension capped so a single decisive finding stands on its own, and one of four tiers, each mapping to an action: pass and monitor, enhance and verify, hold and investigate, suspend and escalate.

The evidence is specific and sourced. Director risk comes from Italian courts and regulators (AGCM, Consob, Banca d’Italia); reputation from press, enforcement records and consumer reviews; trading from procurement registries (ANAC, EU TED), job postings and LinkedIn; address integrity from ghost-pin detection and co-registration density; cyber from the detected tech stack checked against CVE and EOL databases.

That clean-on-paper energy retailer scored Critical, 100/100, on an active regulatory fine, fraud-complaint reviews at 2.2 stars, a director on the record, a dead domain and high-density co-registration, with four independent sources converging on a company the bureau called clean.

The distribution shows the discipline. Of 1,007 companies, 778 scored Low and only 36 reached High or Critical, so the alarm fires where manual review is warranted rather than everywhere. Calibration keeps the signals honest: thin digital presence is normal for B2B and micro-firms, and a missing named director is normal for an Italian SME, so neither is penalised.

Screen on whether a business is real

The bureau can screen on whether a business is real, not merely whether it is registered.

Because the score decomposes to the dimension and the source driving it, a reviewer sees not just that a company scored high but which evidence put it there and what to act on.

The hardest cases, the ones engineered to look ordinary on the registry, are precisely the ones the evidence catches: a few dozen accounts in a thousand, but in lending and onboarding those are the ones that draw down credit and disappear, each carrying a loss far larger than its share of the book.

The same evidence-versus-claim approach extends beyond Italy, with France and the Balkans next.