Why cold emails land in spam

Why do cold emails land in spam when SPF, DKIM and DMARC all pass?

Authentication and placement are different questions. SPF, DKIM and DMARC prove a message was not forged. Whether a mailbox provider puts that message in the inbox is decided separately, mostly by the reputation of the sending domain and IP address, by how recipients have engaged with previous mail from the sender, and by the rate at which people mark messages as spam.

A green DNS checker is therefore a necessary condition and not a sufficient one. Our three domains passed all three checks and still landed in spam folders, which is the single most common surprise for senders who have just finished DNS setup and assume the hard part is over.

Does shared SMTP hosting cause spam placement?

Shared SMTP is the most likely cause of a total placement failure, and was ours. On shared hosting the outbound IP address is used by many customers at once, so that address carries the combined filtering history of everyone using it. A sender with careful copy, verified addresses and clean authentication still inherits whatever the worst sender on that IP has been doing.

Dedicated sending infrastructure removes the shared component. One Google Workspace mailbox per sending domain means the reputation being built belongs to that domain alone, and a mistake on one domain does not contaminate the others.

How do I know if my cold email is landing in spam?

Send a probe from each sending mailbox to a seed mailbox you control, then read over IMAP which folder the probe landed in. Nothing else answers the question directly.

Open rates cannot answer it. Apple Mail Privacy Protection pre-fetches images and registers opens no human performed, and a message sitting in a spam folder can still record an open when the folder is viewed. Reply rate cannot answer it either, because a zero reply rate is equally consistent with poor targeting, weak copy, or nobody seeing the message at all.

A seed probe separates those explanations. Use one seed per provider your prospects actually use, because placement at Gmail predicts nothing about Microsoft 365, where enterprise tenants add filtering layers of their own.

Why does Google Postmaster Tools show no data for my domain?

Google Postmaster Tools needs roughly 250 messages per day to Gmail addresses from a single domain before spam-rate charts populate. Cold outbound from a warmed Google Workspace domain runs about 20 to 25 messages per day, an order of magnitude below the threshold.

Low-volume senders are therefore structurally invisible to the tool most often recommended for diagnosing deliverability. Seed probing exists to fill exactly that gap.

How long does domain warmup take, and can it be shortened?

Warmup takes two to four weeks of calendar time and cannot be compressed by sending more mail. Mailbox providers evaluate sending history across days, so the constraint is elapsed time rather than message count.

One consequence is frequently missed: warming ten domains in parallel still takes two to four weeks, exactly as one domain does. Additional domains buy daily capacity, never a faster start. A campaign that must launch in week one needs a channel that requires no warmup at all.

How many spam complaints can a cold sender afford?

Google asks senders to keep spam complaint rates below 0.3 percent, and low volume makes that threshold far harsher than it appears. At 23 messages per day from a single domain, one person clicking “report spam” represents roughly 4 percent for that day — more than ten times the limit.

Two practical consequences follow. Make opting out easier than complaining: an RFC 8058 List-Unsubscribe header gives Gmail and Outlook their native unsubscribe button, which is the control a mildly irritated recipient reaches for instead of the spam button. And treat a single complaint as a signal worth investigating rather than noise, because at cold-outbound volume there is no statistical cushion.

Are warmup replies real replies?

Warmup replies are generated by other warmup mailboxes and must never be counted as engagement. Warmup services work by having participating mailboxes email each other and reply automatically, which builds sending history — but the traffic is machines talking to machines.

Warmup mail is easy to identify once you know the pattern. Subjects are generic office filler such as “RE: Travel Plans” or “RE: Q3 Budget”, almost always with an RE: prefix that fakes an existing thread. Sender domains are unrelated to each other and appear exactly once. Many networks embed a fixed nonsense token in the body so their bots can recognise each other.

In one of our mailboxes, 226 of 279 messages carried such a token. A sender counting those as replies would report an outstanding reply rate built entirely from automation.

What does a realistic cold email result look like in 2026?

Published benchmarks for 2026 put the meeting-booked rate from cold email between 0.1 and 0.4 percent, with positive reply rates of roughly 0.3 to 0.8 percent — down from 1 to 2 percent a few years earlier. Total reply rates commonly cited sit near 3 to 5 percent.

Small numbers have an uncomfortable implication for anyone judging a new campaign. At a 0.2 percent booking rate, 387 messages predict fewer than one meeting, so zero meetings from a few hundred sends carries almost no information about the offer. A campaign needs thousands of delivered messages before its results mean anything, and messages filtered to spam do not count toward that total.

What actually fixes spam placement?

Placement improves through sender reputation, and reputation is built by evidence rather than configuration. In rough order of impact:

• Move off shared IPs. Dedicated mailboxes per sending domain remove the largest uncontrolled variable.
• Warm properly, then stay inside the cap. Two to four weeks, then the provider’s cold-outbound limit rather than a number chosen for convenience.
• Measure placement daily during the warm. A seed probe costs one message per mailbox per day and is the only way to catch a problem while it is still cheap to fix.
• Make unsubscribing easier than complaining. TheList-Unsubscribe header, honoured immediately and permanently.
• Verify addresses before sending. Bounces damage reputation quickly. Our own bounce rate across 387 sends was 0.52 percent, which is healthy; anything approaching 2 percent needs attention before volume increases.
• Write to one recipient at a time. Near-identical bodies across many recipients are a bulk signal. Deliberate typos are not a fix — filters key on reputation, not on tidiness, and a typo sent to someone who sells writing reads as carelessness.

The mistake underneath all of the above

Our real error was not shared hosting. Shared hosting was the cause; the error was having no way to detect the consequence. Sending ran for weeks with no measurement of where mail was landing, so a fixable configuration problem became several hundred wasted sends and three burned domains.

Deliverability is not a setup task that completes. Reputation moves daily, and the only defensible position is a measurement you look at every morning during a warm.

Related reading

• What is an AI SDR? — how automated outbound works, and where it fails
• How agencies get clients without referrals — building a pipeline that does not depend on word of mouth
• Cold email automation — what the engine does end to end
• Pay-per-meeting outbound — the commercial model