How to Write Cold Email Subject Lines That Don't Get Flagged

To write cold email subject lines that don't get flagged, keep them short (2 to 5 words), lowercase, specific, and free of spam-trigger words like "free," "guarantee," and "act now." Avoid punctuation stacks, emojis, and ALL CAPS. But subject lines rarely cause spam placement on their own. Deliverability is decided by your domain and infrastructure first, then copy.

Key Takeaways

• Subject lines alone rarely trigger spam filters. Domain authentication (SPF, DKIM, DMARC), sending reputation, and volume matter far more than word choice.
• Short subject lines of 2 to 5 words consistently outperform long ones in cold outreach, and they read better on mobile where most first opens happen.
• A warmed mailbox tops out at roughly 30 to 50 sends per day at steady state. That ceiling is why Alchemail runs 200+ sending accounts across 100+ domains instead of blasting from a few mailboxes.
• Open rate is unreliable for testing subject lines because tracking pixels get blocked or pre-fetched. Alchemail runs with open tracking OFF and reads total reply rate instead.
• In 2025, Alchemail's infrastructure-first approach generated $55M+ in client pipeline and booked 927 meetings.
• Total reply rate, not open rate, is the clearest signal that your email reached a human inbox and got read.

Why do cold email subject lines get flagged as spam?

Cold email subject lines get flagged when they combine spam-trigger language (like "free," "guarantee," "urgent") with a sending domain that already has weak reputation. A spam filter is a scoring system that weighs sender reputation, authentication, engagement history, and content signals together. A clean subject line on a poorly warmed domain still lands in spam. The subject line is a minor input, not the deciding factor.

Most people blame the subject line because it is the visible part they can edit. The real culprit is usually infrastructure: missing or broken SPF, DKIM, and DMARC records, a cold domain sent at high volume, or a poor sender reputation from earlier bounces and complaints. Fix those before you agonize over word choice. See our guide on [cold email deliverability fundamentals].

Filters do read content, but they read the whole message: the subject, the body, links, images, and the ratio between them. A subject line with one flagged word inside an otherwise clean, well-authenticated email from a warmed domain almost never triggers spam by itself.

What words trigger spam filters in subject lines?

Spam-trigger words are terms filters statistically associate with promotional or fraudulent email, such as "free," "guarantee," "act now," "limited time," "cash," "winner," "risk-free," and "100%." These words raise a message's spam score, but they only tip an email into the spam folder when combined with other risk factors. On a reputable, authenticated domain, one such word rarely causes flagging on its own.

Here is a practical reference for what to avoid and what to use instead:

Fake "Re:" and "Fwd:" prefixes deserve special mention. They can bump open rates short term, but they damage trust the moment the prospect realizes there was no prior thread, and they raise complaint risk. Complaints hurt your domain reputation, which hurts every future send. For more on protecting sender reputation, see [how sender reputation works].

How long should a cold email subject line be?

A cold email subject line should be 2 to 5 words, or roughly 20 to 40 characters. Short subject lines render fully on mobile, where a large share of first opens happen, and they read as personal rather than promotional. Long, benefit-heavy subject lines look like marketing email and invite both filter scrutiny and prospect skepticism. Brevity signals a real person wrote to a real person.

The goal of a cold subject line is not to sell. It is to look like an email a colleague or peer would send, so the prospect opens it out of curiosity rather than dismissing it as a pitch. Lowercase or sentence case tends to feel more human than Title Case, which reads like a newsletter.

Examples of subject lines that stay short and specific:

• quick question
• {{company}} + {{yourcompany}}
• idea for {{team}}
• worth a look?
• {{firstname}}, timing

Specificity beats cleverness. A subject line tied to the prospect's company or role signals relevance, which raises replies. But specificity only pays off once deliverability and targeting are right. A perfectly personalized subject line to the wrong person on a spam-flagged domain still fails.

Should you A/B test subject lines by open rate?

No, you should not test subject lines by open rate, because open tracking is unreliable. Open rate is the percentage of recipients recorded as opening an email, measured by a tracking pixel. Those pixels are increasingly blocked or pre-fetched by privacy tools and mail providers, which inflates or erases the numbers. Alchemail runs with open tracking OFF and measures total reply rate instead.

Total reply rate is the percentage of recipients who reply at all, including "not interested." It is the clearest test of deliverability, not a vanity metric. If people reply at all, the email reached the inbox and a human read it. Positive reply rate then forecasts pipeline. Both matter, for different reasons.

So test subject lines the way that actually reflects results: split your list, send variant A and variant B, and compare total reply rate and positive reply rate across the two. This ties your subject line decisions to inbox placement and human engagement rather than to a pixel that may never fire. Learn more in [why we turned off open tracking].

Do AI-generated subject lines get flagged more often?

AI-generated subject lines are not flagged more often for being AI-written, because filters do not detect authorship. They flag patterns: promotional language, spam-trigger words, punctuation stacks, and repetition across many sends. The real risk with AI is sameness. If a tool generates near-identical subject lines at high volume from the same domains, filters can learn that pattern and downgrade it.

Used well, AI helps you write shorter, more specific, more varied subject lines at scale. Used badly, it produces generic, benefit-stuffed lines that read like marketing and repeat across your whole campaign. The fix is variation tied to real prospect data, not templated filler. Personalization only pays off once deliverability, targeting, and offer are right, so build those first, then let AI vary the copy. See [using AI to personalize cold email at scale].

The infrastructure point still governs. AI cannot rescue a subject line sent from an unwarmed domain at high volume. Because a warmed mailbox tops out near 30 to 50 sends per day, scaling requires more accounts and domains, not more sends per box. That is why Alchemail runs 200+ accounts across 100+ domains.

FAQ

Does the subject line affect deliverability more than the body?

No, the subject line does not affect deliverability more than the body. Filters score the entire message plus sender reputation and authentication together. The body's link count, image-to-text ratio, and overall content weigh as much or more than the subject. And all of it sits below domain reputation and infrastructure. Fix authentication and warming first, then treat both subject and body as secondary content signals.

Can one spam word in the subject line send me to spam?

Rarely, one spam word in the subject line will not send you to spam on its own. Spam filters use cumulative scoring, so a single flagged word on a well-authenticated domain with good reputation usually passes. The danger is stacking: a trigger word plus a cold domain, plus high volume, plus a link-heavy body. Remove obvious triggers, but do not assume word choice alone decides placement.

Should I personalize the subject line with the prospect's name or company?

Yes, personalizing with the company name usually outperforms generic subject lines, because it signals relevance and looks like a real one-to-one email. First name works too but feels more common. Personalization only pays off once deliverability, targeting, and offer are correct. A tailored subject line to a well-matched prospect on a healthy domain raises replies. The same line on a spam-flagged domain still fails.

Are emojis in subject lines bad for cold email?

Yes, emojis are generally bad for cold email subject lines. They raise filter scrutiny because they correlate with promotional and spam email, and they undercut the credibility of a one-to-one business message. A cold email should read like a real person writing to another person. Plain, lowercase, specific text does that. Save emojis for warm, opted-in audiences where they match expectations.

How do I test whether my subject lines are hurting deliverability?

Test subject line impact by splitting your list and comparing total reply rate between variants, not open rate. Open tracking is unreliable because pixels get blocked or pre-fetched. If total reply rate drops on one variant while the domain, list, and offer stay constant, the subject line is likely the cause. Reply rate confirms the email reached a human inbox, which is what actually matters.

What is the safest subject line format for cold email?

The safest cold email subject line format is 2 to 5 words, lowercase or sentence case, specific to the prospect or their company, with no spam-trigger words, no emojis, no ALL CAPS, and no punctuation stacks. Examples include "quick question," "idea for {{team}}," or "{{company}} + {{yourcompany}}." Keep it honest, avoid fake "Re:" prefixes, and remember infrastructure decides placement before copy ever does.

How many emails per day can I send from one mailbox safely?

A warmed mailbox tops out at roughly 30 to 50 sends per day at steady state. Pushing past that ceiling risks reputation damage no matter how clean your subject lines are. To scale volume safely, add more sending accounts and domains rather than raising per-mailbox volume. Alchemail runs 200+ accounts across 100+ domains for this reason, which supported $55M+ in client pipeline in 2025.