Email Scoring That Tells You
What “Valid” Never Will.

Email scoring rates each address on a 0 to 100 scale that estimates how likely a real, active person reads mail there, instead of the binary valid/invalid verdict a standard checker returns. Verifox builds the score from 40+ behavioral signals layered on top of the deterministic checks that email verification runs, then maps it to one of four tiers with a recommended action.

Four stages run in sequence: a behavioral fingerprint built from engagement signals like open recency, reply cadence, and device diversity; catch-all intelligence that probes the mail server beyond a simple RCPT TO check; pattern analysis against millions of known role, disposable, and generated-address templates; and a final synthesis that compresses everything into one 0 to 100 number with a tier assignment. Cached results return in under 50ms.

85 and above is safe to send: the mailbox is active and accepting mail. 50 to 84 is worth a look, so segment it into a slower cadence or re-engagement flow. 20 to 49 is risky and should sit out your primary campaigns. Anything under 20 is dead, fake, or a spam trap; remove it immediately. That one sort keeps bounce rates under the 2% threshold most mailbox providers punish.

Catch-all servers accept mail for any address, which is why 20 to 40% of B2B lists fail there and most validators return “unknown.” Verifox analyzes domain-level bounce history, known catch-all configurations, and behavioral engagement signals behind the SMTP response to resolve those addresses to a concrete score with 98.2% accuracy, including on Google Workspace and Microsoft 365 catch-all setups.

Verification answers “does this mailbox exist?” with deterministic checks: syntax, MX records, the SMTP handshake, disposable and role detection. Scoring answers “should I send here?” by adding behavioral evidence on top. You can try the binary checks free on the free email checker; scoring matters most where those checks stall, on the catch-all domains the 9-point verification engine flags.

The underlying verification engine measures 99.99% accuracy on a 1,000-address benchmark list, and the AI scoring layer resolves catch-all addresses at 98.2% accuracy, the segment where binary tools simply give up. Scores are probabilistic by design: a 92 is not a guarantee, it is a measured confidence level. The tier thresholds are calibrated so that acting on them keeps bounce rates below the 2% line.

Both. CSV upload handles bulk jobs at roughly 10,000 emails per minute on the standard tier, and the REST endpoints documented in the API reference return single scores at a median of ~380ms, sub-50ms when cached. AI agents can call the same engine through the native MCP server with no glue code.

Sort once, then act by tier: full sends to 85+, a slower nurture cadence for 50 to 84, suppression for 20 to 49, and immediate removal under 20. Teams running email warmup point early sends exclusively at the top tier to build reputation fast, then confirm the effect with an inbox placement test before opening up the middle of the list.

That is its main job. Bounces and spam-trap hits are what land senders on blocklists, and scoring removes both before you press send. One unscored campaign can push you past the 2% bounce threshold, and recovery takes 72+ hours once providers throttle you. Pair scoring with blacklist monitoring so that if something does slip through, you hear about the listing within minutes instead of weeks.

Sign up free and you get 1,000 credits, or 2,500 with a work email, no card required. Credits are pay-as-you-go, never expire, and the pricing page shows volume rates for your region. Addresses are processed in memory and discarded once scoring completes: nothing stored, logged, or sold, backed by SOC 2 Type II.