Every validation response includes a risk_score between 0 and 100. Zero means no risk; one hundred means the highest risk. The score is designed to be dropped straight into a downstream rule \u2014 no math required.
Use the recommended action as a starting point and tune for your own risk appetite.
Strong positive signals across every factor. Treat as a clean, trusted contact.
A small number of weak signals. Acceptable for most flows; add friction for high-value actions.
Multiple negative signals stacking up. Send to a queue, ask for step-up verification, or hold the order.
Strong, layered fraud signals. We recommend rejecting outright or routing to a fraud analyst.
Four independent factors feed into every phone score.
Four independent factors feed into every email score.
Run a check against the public demo endpoint and see how the score is interpreted.
{
"valid": true,
"type": "landline",
"country": "CH",
"risk_score": 4.2,
"formatted": "+41 44 668 18 00"
}We deliberately do not publish the exact weighting or internal thresholds of each factor. That is what keeps the score useful: if it were transparent, fraudsters would game it. We tune the model continuously based on aggregate signals across all customers — no individual data is ever shared between accounts.
Need a custom scoring profile for your industry (banking, gambling, e-commerce, SaaS)? That ships with the Pro plan.