Verifiers split results into buckets, and no two vendors draw the lines in the same place. "Risky" is the loosest of those labels. At one tool it means catch-all. At another it bundles catch-all with role addresses, disposable domains and full mailboxes. Before you act on a risky count, open the vendor's result-code page and read what went into it.
Unknown is narrower. The server didn't give a straight answer, so nothing was learned about the mailbox. Greylisting, throttling, a block on probe traffic, a timeout and a 252 reply all end there. Many unknown addresses are real, which is why deleting them throws away live contacts. Our answer on what unknown means walks through each cause.
Here's the practical split. A risky row is information you already have: the domain accepts everything, or the inbox belongs to a team. You decide whether to mail it. An unknown row is information you don't have yet, and a re-check a day later often turns it into valid or invalid.
ZapBounce swaps the risky bucket for four results and three flags. Each address gets one of four results: valid, invalid, catch_all or unknown. Role, disposable and free-provider are separate flags set on every row. So a role address on a catch-all domain shows both facts, instead of one vague word hiding which one applied. The result codes reference lists every value.
Billing follows the same line. A catch_all result costs a credit, because it took a real handshake plus a second probe and tells you something true about the domain. An unknown result costs nothing, because nothing was learned. At $5 for 1,000 checks on the pricing page, you pay only when a mail server gave an answer.
Related questions
Should I send emails to risky addresses?
Send to them in a small separate batch after your valid rows land cleanly. Amazon SES asks senders to keep bounces under 2%, so stop and drop the rest if that batch goes higher. Role addresses in the risky group suit a company-level message better than a personal pitch.
Is a catch-all email risky or unknown?
Catch-all belongs with risky, because the server did answer: it accepts every address at the domain. About 28% of addresses in a 10,000-email B2B benchmark sat on catch-all domains. ZapBounce reports these rows as catch_all, a result of its own.
Do I pay for unknown email results?
Not at ZapBounce. An unknown result is never billed, and neither is a duplicate address in the same batch. You pay for valid, invalid and catch_all results, where a mail server gave an answer.
Is risky the same as invalid?
No. Invalid means a definite no, such as a 550 for that recipient or a domain with no mail server. A risky address can still accept mail, so deleting risky rows removes contacts who would have received your message.