Cold email deliverability, measured

Deliverability is placement, not delivery

The sending tool reports “delivered” when the receiving server accepted the message. It cannot see what the provider did next: the inbox, the Promotions tab at Gmail, or spam. Deliverability is that second decision, per provider, and it is the number a cold sender is actually paid on. A campaign can be 100% delivered and mostly in spam, and nothing in the sequencer will say so; only a placement test does.

The receivers' rules, in their words

Since 2024 the three largest receivers publish sender requirements and enforce them with rejections. For a cold sender the rules that bite are authentication and the spam rate; the rest are in the auth checker.

• Google. SPF and DKIM must both pass for the sending domain; DMARC record required for bulk senders (p=none minimum), aligned with SPF or DKIM. Spam rate: Keep below 0.10% in Postmaster Tools; never reach 0.30%. Senders above 0.3% (since June 2024) are ineligible for Google mitigation; mitigation requires 7 consecutive days below 0.3%.
• Yahoo. SPF required; DKIM required, key ≥1024-bit (2048 recommended). Spam rate: No numeric threshold published; high-complaint domains are deferred.
• Microsoft. SPF must pass; DKIM must pass. Threshold: 5,000+ messages per day to consumer Outlook.com/Hotmail/Live addresses = high-volume sender.

Check a domain against all three with the bulk sender auth checker; the full requirements with their dates are on the bulk sender requirements page. The rejections each provider returns for a failure are decoded in the bounce code decoder.

Infrastructure: the domain and the mailboxes

1. 1

   A separate sending domain

   Cold email goes out from a domain that is not the one your company runs on, so a filtering problem never touches the mail that pays the bills. Whether that is a subdomain or a lookalike is its own decision; the guide on it is linked below.

2. 2

   A few mailboxes per domain

   Volume comes from more warmed mailboxes, not from one mailbox sending more. The cold email volume page has the arithmetic for a target daily total.

3. 3

   Authentication that passes and aligns

   SPF, DKIM and DMARC on the sending domain, with the From domain aligned to at least one of them. The generators produce the exact records; the auth checker verifies them after propagation.

4. 4

   Reverse DNS where you control the IP

   Google requires valid forward and reverse DNS for sending IPs. On Google Workspace and Microsoft 365 the provider handles it; on your own server or a relay it is yours to publish.

Subdomain or separate domain covers the first step; the auth checker and the record generators cover the third.

Reputation before volume

A new domain has no history, and the receivers treat no history as a reason for caution. Warming gives the domain a record before the first cold email goes out: on MailStrike an AI persona on each mailbox holds real two-way threads across the network on a 21-day ramp, so by the time outreach starts the domain has been read, replied to and kept by real mailboxes. Then outreach starts at 20 to 30 a day per mailbox and scales as placement holds, while warming continues at a maintenance level so the record does not decay under the campaigns.

• How long warmup takes: the 21 days, and the measured days to 90% inbox by domain age and provider.
• How many cold emails per day: the providers' caps against the safe volume, and the mailboxes needed for a total.

Content the filters read

Content is judged together with reputation, and a filter that has seen a phrase from spammers weighs it against a young domain more than an established one. The habits that hold up: one message written for one person rather than a template with merge fields, plain text or close to it, no tracking domain the receiver has already flagged, a working reply address, and links only to domains you would send from. The template analyzer scores the subject line and the body separately and names the evidence for each finding, which is a better answer to “how long should a cold email be” than any word count.

The list

The receivers read the list through its results: bounces from addresses that do not exist and complaints from people who did not want the mail. Both are rates, and both are what Google's spam-rate line and the bounce thresholds measure. Verify the list before the first send, remove every hard bounce, and watch the rate rather than the count; the bounce rate guide has the thresholds.

Monitoring: what to watch, and where

• Google Postmaster Tools for domain and IP reputation and the spam rate at Gmail; Microsoft SNDS for the sending IPs at Outlook. Both report standing, not where a message landed.
• A placement test before the first campaign and after any change: inbox, promotions or spam per provider. The Free Deliverability Test is that, with the content scored first.
• The public record: blacklists, authentication, reverse DNS and domain age in one report from the reputation checker.
• Bounces, read by code rather than by count: a 5.7.26 is authentication, a 4.7.28 is volume, a 5.1.1 is the list.

The order that works

1. 1

   Domain and authentication

   Separate sending domain, SPF, DKIM, DMARC aligned, checked.

2. 2

   Warm for 21 days

   Every mailbox, before it sends a single cold email.

3. 3

   Test placement

   Per provider, before the campaign. Fix the provider that fails.

4. 4

   Send at 20 to 30 a day per mailbox

   Scale with placement, add mailboxes for volume, keep warming underneath.

5. 5

   Watch the rates

   Spam rate in Postmaster, bounces by code, and a placement test after any change.