Every lookup shows a trust score from 0 to 100 — higher means lower risk. Here’s exactly what goes into it, in plain language, and why it’s built to be hard to game.
The number you see is a trust score from 0 to 100 — the higher it is, the lower the risk it carries. It’s the mirror image of the spam-and-scam risk we compute behind the scenes. A brand-new, unknown number doesn’t start at the top: it starts below a verified, accountable owner, because “we’ve never seen a problem” is not the same as “this is safe.”
The starting point depends on verification. An unknown, unclaimed number begins at a guarded baseline. When someone verifies ownership — a personal number or a business — that baseline rises, because a real, accountable owner now stands behind it. But the badge is not a shield: a verified number that gets reported for abuse falls faster than an anonymous one. You can’t buy a clean reputation.
Two things push it down: spam and scam reports from the community, and consistently poor reviews. Two things hold it up: a clean history and positive reviews. Recency matters most — a report from last week weighs far more than one from last year, and a number’s influence roughly halves every three weeks. So a number that cleans up its act, or gets reassigned to a new owner, can recover over time rather than being branded forever.
Reports aren’t votes you can stuff. First, you can only report a number after verifying your own — anonymous mass-reporting isn’t possible. Second, we weigh each reporter by whether their past reports agreed with the wider community: a reporter with a solid track record counts more, while a cluster of brand-new accounts all reporting the same number counts for very little. Faking broad, long-term consensus is the hard part — by design. Reports from accounts with a verified business carry weight and build credibility; reports from free or personal accounts are a bounded, deferring signal that can nudge a cold unknown number but cannot sink a verified one on its own.
We also fold in public complaint data — for example, government and regulator spam-complaint feeds. These can corroborate a bad number, but they’re deliberately capped: a single outside mention nudges an unknown number only a little, never into the danger zone on its own, and it always defers to first-hand reports from verified callers. Outside data never touches reporter credibility, so it can’t be laundered into fake consensus.
Some calls arrive with caller-ID authentication (STIR/SHAKEN) — a cryptographic sign-off from the originating carrier. We rarely have this for any single lookup, and a consumer search never carries it; it reaches us mainly from operators and voice platforms on the inbound API. When it is present, an authenticated call gently raises our confidence in that call, while a call that fails verification is a supplementary negative signal for that call alone — never stamped onto the number itself. Separately, a pattern of unauthenticated or failed calls seen by several independent sources is bounded corroboration that the number may be spoofed: it can nudge risk upward, but only with that corroboration, always capped, and always deferring to first-hand reports from verified callers.
The score is an informational signal to help good calls get through and warn about likely fraud. It’s never the sole reason anyone is denied anything, we keep a human in the loop, and every report can be challenged. If you think a score is wrong, you can report a number or dispute a report — more on our approach on the Trust & safety page.