Cold emails landing in spam: find the cause

Three outcomes, three starting points

Look at what your sending logs say before you change anything.

• Rejected (5xx) or deferred (4xx) with a code: the receiver told you which rule failed. Read the code in the table below and fix that record or setting.
• Accepted but filed as spam: the receiver judged the sender or the message. Look at the list, the complaint rate and the message itself.
• Bounced as an unknown user: the address does not exist. That is a list problem, covered in the verification post.

Gmail rejection and deferral codes, with the fix

Google sends a rejection code and a reason when it turns a message away for failing the sender guidelines, and shows the same information in Postmaster Tools. The codes below are quoted from Google's own list; the fixes follow the setup steps Google points to.

When there is no error and the mail still goes to spam

Without a code, the signal comes from recipients and from the message. Google says the user-reported spam rate has a graduated effect on delivery, that rates above 0.1% already hurt inbox delivery for bulk senders, and that 0.3% or higher is worse. It also says that if messages from a domain are frequently reported as spam, later messages are more likely to be marked as spam.

The same guidelines name practices that invite those reports and that filtering looks at:

• Buying addresses, or mailing people who never signed up. Google and Yahoo both say not to.
• Subject lines that fake a reply or forward, such as Re: or Fwd: on a first message.
• Hidden content made with HTML and CSS, or links whose destination a reader cannot tell.
• Names or symbols meant to suggest the sender has been verified when it has not.
• No easy way to unsubscribe. Google notes that an easy opt-out can improve open rates, click-through rates and sending efficiency.

Find the mailbox that is hurting you

A domain reputation problem and a one-mailbox problem look the same in a dashboard average. Split the numbers by mailbox before you decide. The table is a fictional week for four mailboxes on two domains. The bounce rate is bounced divided by sent, and the complaint rate is complaints divided by messages delivered to the inbox, both from your own logs.

Compare the rows with each other and with your own previous weeks. Both mailboxes on the second domain bounce far more than the first domain's mailboxes, which points at the list those mailboxes were given, while the last mailbox also shows the highest complaint rate, so it deserves a pause and a look at its recent messages first.

Fix in this order

The order matters because the early steps are cheap and the later ones take weeks to show up in reputation.

1. Pause the affected mailbox, not the whole account, while you read the codes and the recent complaints.
2. Confirm SPF, DKIM, DMARC alignment and reverse DNS for the sending domain. Fix any record a code points at.
3. Verify the list that mailbox was sending to and remove undeliverable and risky addresses. Add everything that bounced or complained to the suppression list.
4. Resume at a lower volume. Google advises reducing volume when bounces or deferrals appear and increasing slowly afterwards, and notes that it can take time for an improved spam rate to show up in how mail is classified.
5. Rewrite the message only after the steps above, and remove anything in the practices list.

Where ColdMeetings fits

ColdMeetings sets up sending domains, mailboxes and the SPF, DKIM and DMARC records, checks addresses before an import, and keeps a do-not-contact list. Bounces and complaints suppress the address, and a preflight check before a campaign starts looks for a mailbox on a domain with verified SPF, DKIM and DMARC records. It is software and an API for running the operation; the reputation of your domain follows who you mail and what you send, and new mailboxes ramp up with real, low-volume sends.