Mailient Research
Original measurement rather than commentary. Every edition publishes its sample, its date, what was measured, what deliberately was not, and the raw per-domain data as a CSV so anyone can check the numbers instead of taking them.
Q3 2026 · measured 5 September 2026 · top 50,000 domains
Of the 36,548 sampled domains that actually receive email, 17,494 publish no DMARC policy that would stop a forged message being delivered. Only 10,436 are set to p=reject.
We make an email tool, so we are not a neutral party and there is no point pretending otherwise. What we can do is show the working: the sample is a public research list, the records are public DNS, the method says what was left out, and the per-domain data is downloadable. Anything in these pages can be checked without asking us.
The same records, read live for a single domain, with the exact DNS entries that fix whatever is missing. Free, no account.
Check a domain