A claim-level scoring system that shows its working.
Claim-level precision
Every statement scored independently
Shows its working
Each score lists the claims it rests on
How it works
Most due diligence platforms assign a single score to an entire response. Applied Diligence works at the level of individual claims — the discrete, verifiable statements within a subject's response.
Each claim is scored independently against available evidence before being aggregated into an overall score. You know exactly what is wrong, why it is wrong, and which source revealed it.
Scoring logic
Every claim receives one of five verdicts before contributing to the overall score.
A reliable source directly supports the claim.
No source contradicts the claim, though active confirmation was not possible.
Evidence on the case both supports and contradicts the claim.
A reliable source directly contradicts the claim. Triggers automatic score penalty and priority flag.
Nothing on the case supports it except the subject's own answers. Treated as a gap, not a neutral outcome.
Source taxonomy
Every claim is scored against three evidence sources, weighted by how reliably each one can confirm or contradict it: the subject's own other answers, documents in the case corpus (uploaded files, registry and sanctions check results, the questionnaire transcript), and news coverage.
Confidence interval
Every score is shown with a confidence interval. It starts at ±25 and narrows by 3 for each claim independently verified against the evidence, to a floor of ±5 once seven or more are. A wide band means few claims could be checked on this case — not that the subject is hiding something, and not a judgement about their country.
High confidence
Seven or more claims independently verified against the evidence on the case. This is the narrowest band the engine produces.
Limited evidence
One claim independently verified. The band narrows with each further one.
Scoring profiles
The three source weights above — questionnaire, corpus, news — can be adjusted per engagement to reflect your organisation's risk posture. This is set up for you when you onboard; there is no self-service configuration today.
One additional control exists: whether a confirmed contradiction caps the investigation's headline score. It's on by default, and can be turned off per engagement if your review process is built to handle contradictions differently.
Contradiction caps the headline score
On by default. A confirmed misstatement dominates the score rather than being diluted by volume of corroborated claims — see The Five Verdicts above.
One investigation — score by round
Round 2: material omission flagged in ownership disclosure.
Score history
Within a single investigation, the score is snapshotted at every round of questions and answers — not just recalculated in place. As follow-up questions go out and new evidence lands in the corpus, you can see exactly how the score moved between rounds, not just where it ended up.
This is scoped to one investigation — it tracks how the evidence for this engagement developed over its own rounds, not a subject's history across separate cases.
Audit trail
Every investigation keeps a chronological log — questionnaires sent, responses received, sanctions and registry checks run, with their outcomes. Screening hits are called out explicitly, and the case's risk rating is derived from that log, not hardcoded.
A generated report's headline score is computed the same way the case's credibility score is — the plain average of assessed claims, capped when a confirmed contradiction is found. When that cap applies, the report's findings name the specific contradiction and why it was flagged. The full claim-by-claim breakdown — every claim, not just the ones that capped the score — still lives only in the investigation workspace, not the audit log or the report itself.
Screening: OFAC SDN — US Treasury
Result: 0 possible matches
Engagement: Response received — Veracity score: 74
Risk rating derived from latest score and any confirmed contradictions, not from screening status alone.
Get started
We offer structured demonstration sessions for qualified enterprise clients.