What Is a Trust Score? How TrustVerify Calculates Risk

Identity verification produces a lot of raw signal: a face-match percentage, a liveness verdict, a document-authenticity confidence, a sanctions outcome, an IP reputation. Asking an onboarding team — or a checkout flow — to read all of that on every check does not scale. A Trust Score solves this by condensing the signals into one number from 0 to 100, where higher means lower risk.

The signals behind the score

TrustVerify computes the Trust Score from independent, weighted signals so that no single check can carry a decision on its own:

• Biometric face match: how closely the live selfie matches the photo in the identity document.
• Liveness / presentation-attack detection: confidence that a real, present person was captured — not a photo, replay or deepfake.
• Document authenticity: the quality and consistency of the extracted document data (MRZ/barcode checksums where available).
• AML outcome: whether sanctions, PEP or adverse-media screening returned clear, review or hit.
• Device and IP risk: VPN/proxy/Tor detection, geolocation consistency and device-fingerprint signals.

Why one number, and why independent signals

A single score is operationally powerful: it lets you set one threshold and route automatically. But the score is only trustworthy if its inputs are independent. If a fraudster defeats the face match but the liveness, document checksum and AML signals still fail, the composite score stays low. Combining independent signals is what makes the number robust against a single point of failure — and it is why a verified Trust Score is more reliable than any individual check.

How to read the score

TrustVerify maps the score to a risk band and a recommended action:

• Approve (high score): all checks corroborate a genuine identity — let the user through automatically.
• Review (mid score): one or more signals are inconclusive or in mild conflict — route to a human or request a re-capture.
• Reject (low score): a hard signal failed — a sanctions hit, failed liveness, or a strong identity mismatch.

Setting thresholds for your risk appetite

There is no universal threshold; it depends on the cost of a false accept versus a false reject in your business. A crypto exchange or a lender will set a higher bar than a low-value marketplace. The right approach is to start conservative, watch the review queue, and tune the approve and reject cut-offs against real outcomes. Because the score is deterministic and the underlying signals are logged, every decision is explainable after the fact — essential for compliance and for disputing chargebacks.

Trust Score and honest reporting

A score is only as good as the integrity of its inputs. TrustVerify shows the contributing signals alongside the score on every verification report, distinguishes a value that was cross-checked ("match") from one that was simply read ("verified"), and never inflates a number it cannot evidence. The goal is a score your compliance team — and your auditor — can defend, not a flattering headline figure.