Fair question, but no. Unknown senders go to quarantine, and I check quarantine at least daily. So some messages get delayed, but all acceptable messages get delivered. I was already doing this, because we had false positives before switching strategy, but it feels a little more urgent now.
It works out like this:
First stage filtering, out of 100% not blocked during helo/ehlo
- 53.6% are discarded for no valid recipients
- 46.4% are allowed to proceed
Second stage filtering, out of 100% with valid recipients
- 14.3% are flagged as confirmed spam,
- 1.5% are flagged as probable spam, and
- 2.7% are flagged for unknown sender.
- 81.5% are allowed to proceed
Third stage filtering, out of 100% allowed to proceed to third stage
- 4.2% are blocked for content or other reasons
- 0.2% are added to the total as quarantine false positives
- 96.0% are released to the users.
Currently, known-sender false positives are 64% of all false positives, so my improperly delayed message volume has jumped from about 2 messages per day to 5 messages per day.
Although unknown senders arrive continuously, errors do not repeat (except on a same-day basis.) Once a message is released as a false positive, that sender becomes a known sender overnight.
Caveats: I said this elsewhere, but I don't think I mentioned it on this topic.
I have split senders into three categories, to limit risk:
- Consumer mailbox messages - The message is from Gmail, Hotmail, Yahoo, etc. These are currently exempted from the control, although we still track known sender status. We have existing problems with these messages being quarantined incorrectly based on spam score, so I don't want to make the problem worse.
- Aligned domain messages - SMTP From address and Message From address are the same organization. The unknown sender control is only applied to these messages at present.
- Unaligned domain messages - The two address are different organizations. This is mostly advertising from Email Service Providers (ESPs) like Sendgrid and ConstantContact. Known and Unknown sender status is currently tracked but not enforced, but I plan to control this group in the future.
A sender can become known for three reasons:
- An account or domain is in a corporate database as a vendor, client, employee, etc. Not all of my potential database sources have been integrated my solution at present.
- An account or domain has been sent a message from us.
- An account or domain has sent a message to us, and it has been allowed through our filtering process and was not blacklisted based on after-delivery complaints.
For consume mailbox domains, risk and known-sender status are tracked by individual accounts. For other domains, trust and known-sender status are tracked by domain.