Verified email

Vendors compress a spectrum into one badge. It is worth decompressing, because the difference decides your bounce rate:

• Published — the address appears on a page the company controls. The strongest ordinary evidence.
• Checked — a mail server accepted the address in a probe. Good, and increasingly unavailable as providers stop answering probes honestly.
• Patterned — first.last@ derived from a known format at that domain. A guess with decent odds, which is not the same as verified.
• Syntactic — the string is a valid address and the domain resolves. Almost no information.

A headline accuracy figure is measured on the vendor's own sample, in a way you cannot reproduce, and it is dominated by large well-known companies whose addresses are easy. The number that matters is the bounce rate on your list, in your sector, at your target company size — which you learn by sending a small batch and measuring, not by reading a marketing page.

Treat any published accuracy claim as a comparison the vendor designed. The useful question is what evidence stands behind each individual record.

Roughly a quarter to a third of business contact records go stale each year through job changes alone, and verification is a statement about the moment it was performed. An address checked eighteen months ago and never re-checked is not verified; it is historical.

Which is why the useful metadata on a record is not a badge but a date — when the evidence was gathered, and from where.

Pool addresses in SalesShift carry a grade rather than a single verified flag, and anything the enrichment crawler discovers on a company's own site is stored as unverified — because finding a string on a web page is not verification. Provenance is kept per record: which URL, which fields, when.