What is a cold email? The honest definition
One researched message to one stranger is a different animal from a million scraped addresses. Here is where the line sits and how to stay on the right side of it.
A cold email is a commercial message sent to one specific person who has never heard from you, for a reason you can state in a single sentence. That stated reason carries most of the difference between outreach and spam, and it is what separates cold emailing that works from the kind that gets your domain blocked.
This article defines the term and draws the lines around it. The sending mechanics (separate domains, warm-up ramps, daily caps) live in our cold email infrastructure guide, and the statutes get a full treatment in is cold email legal.
The definition, drawn precisely
"Cold" describes the relationship. The recipient never subscribed, never bought anything, and would not recognize your name in the From line. The message is commercial: you want a reply, a call, a trial, an introduction. And it is individual. You picked this person after looking at what they do, and the email should prove it.
Sales teams call the practice outbound prospecting. Recruiters sourcing candidates, founders recruiting design partners, and freelancers pitching editors run the same play under different names.
Three neighbors it gets confused with. A newsletter goes to people who asked for it. A transactional email, like a receipt or a password reset, goes to people who triggered it. Spam goes to whoever a scraper found. Cold email sits in the fourth quadrant alone: unasked for, but aimed.
Cold email vs spam: the three lines
Consent separates neither category, because both arrive uninvited. The separation happens on relevance, identification, and the out.
| Question | Cold email | Spam |
|---|---|---|
| Who gets it | A named person you researched | Every address a scraper found |
| Why them | A specific, stated reason | They had an inbox |
| Who sent it | Real name, real company, working reply address | Forged or rotating identities |
| The way out | A plain "tell me to stop" that is honored | Hidden, fake, or a click trap |
| Daily volume | Dozens per sender | Millions |
The legal line runs close to that table. CAN-SPAM requires truthful headers and subject lines, a physical postal address, and an opt-out honored within ten business days, and it prices violations per email. The practical line is stricter, because mailbox providers count complaints instead of reading statutes. Gmail's bulk-sender rules, published February 2024, set a spam-complaint target under 0.1% and a hard ceiling at 0.3%; cross the ceiling and Gmail withholds delivery help until you stay under it for seven consecutive days. Our own send path pauses automatically at those exact thresholds, because stopping a campaign is cheaper than a week of earning back trust.
Cold email vs cold calls vs ads
The three ways to reach a stranger have different cost structures, and the differences decide when each one makes sense.
| Cold email | Cold calling | Paid ads | |
|---|---|---|---|
| Cost per contact | Research minutes plus a fraction of a cent to send | A person's time, one call at a time | The auction price of every click |
| What failure costs | Your domain's sending reputation | Little beyond the wasted hours | Budget |
| Scales with | Domain reputation, slowly, through warm-up | Headcount | Money, instantly |
| Feedback quality | Written replies you can reread and count | Live objections, the richest signal there is | Click rates with no words attached |
| Channel ownership | Your domain and your list | Your phone | A rented auction that reprices daily |
The row that matters is failure cost. Ads waste money and calls waste time, but both losses stop when you stop. Cold email failure compounds: spam complaints degrade the domain that carries the rest of your mail. That asymmetry is why this cluster spends an entire article on setting up a separate outreach domain.
When it works, and when it cannot
Cold email works when three conditions hold at once. The target is a business, so the address is professional and the message concerns the recipient's job. The deal size pays for research, because ten minutes per prospect only makes sense when a closed deal is worth thousands. And the sender is a human who can hold the conversation a reply starts. Send fifty researched emails, not five thousand scraped ones.
It cannot work as consumer marketing. Law is one reason: Canada's CASL requires consent before commercial email, and Germany reads its ePrivacy rules the same way. Arithmetic is the other. A 0.3% complaint ceiling means three complaints per thousand delivered messages, and consumers who have never heard of you tap "report spam" without a second thought. Three annoyed strangers per thousand. No purchased consumer list survives that ceiling for long. Cold email to consumer inboxes at scale is spam, whatever the tool selling it to you calls it.
One myth worth retiring: Gmail defines a "bulk sender" as 5,000+ messages a day, and small senders read the number as an exemption. It is the threshold for extra requirements like one-click unsubscribe. The authentication and complaint rules apply at any volume, and we follow them at dozens of messages a day.
The anatomy of a legitimate cold email
Every line of a defensible cold email does one of three jobs: identify you, deliver the reason, or offer the out. The example below is ours, written for this article; both companies and the postal address are invented.
From: Maya Chen <maya@brightloop-hq.com>
Subject: Question about Ridgeline's onboarding queue
Hi Sam,
Your careers page lists four support hires this quarter, which
usually means the onboarding queue is growing faster than the team.
Brightloop automates the account-provisioning half of that queue.
Worth a 15-minute look, or is onboarding not your side of the org?
If you would rather not hear from me again, say so and I will stop.
Maya Chen, Brightloop
2201 Fourth Ave, Suite 900, Seattle, WA 98121
Identification is the From line, the signature, and the postal address; the address is a CAN-SPAM requirement, and a registered PO box satisfies it. Value is the middle: one observed fact about the recipient, one plain sentence about what you do, one small question. The out is the sentence near the end, and it must be honored the moment someone uses it. On newsletter sends our workspace stamps RFC 8058 one-click unsubscribe headers automatically; a one-to-one cold email carries its out as a sentence instead, but the discipline behind both is identical. Whoever says stop goes onto a suppression list that is checked before every future send and never expires.
Notice what the example omits. No "hope this finds you well." No attachment. No fourth paragraph. It runs four sentences because the reader decides in two.
Where to start
Work these in order:
- Rules first. Confirm what your target countries require. The legal article linked above covers CAN-SPAM, GDPR, and CASL in operator terms.
- Domain second. Buy a separate outreach domain and publish SPF, DKIM, and DMARC on day one. The domain article linked above walks the checklist in order.
- Volume third. Set a daily cap before the first send. How many cold emails per day publishes the caps math from our own throttle policy.
- Copy fourth. Start from working cold email templates and spend most of your effort on the subject line, which decides whether anything else gets read.
- Persistence fifth. Most replies arrive after the follow-up, so plan the sequence before message one goes out.
Copy sits at step four on purpose. A perfect email from an unauthenticated domain never reaches the inbox it was written for.