Spam trigger words and filters: What actually matters

The spam-word checklist is one of the most durable myths in cold email. It survives because it is easy to act on, and it persists despite the fact that filters stopped working that way years ago.

The word-list myth

The belief is that certain words, free, guarantee, offer, act now, trip the filter, so you avoid them. There was a grain of truth to this a long time ago, and almost none now.

Modern filters do not scan for naughty words in isolation. They judge the sender and the pattern. A reputable sender can use free in a sentence and land in the inbox; a burned domain gets filtered writing nothing but plain prose.

What filters actually weigh

The signals that decide inbox placement are mostly about you, not your wording.

• Sender reputation, the domain and IP history, covered in reputation and blacklists.
• Engagement, whether recipients open, reply, or mark as spam.
• Authentication, SPF, DKIM and DMARC set up correctly.
• Sending behaviour, steady volume, low bounces, no sudden spikes.

Where wording still matters, a little

Content is not entirely irrelevant. A message that reads like textbook spam, all caps, lots of links, hard-sell hype, aligns with what filters expect from bad senders and adds to a picture reputation already paints. But it is a minor input, not the main event, and cleaning up genuinely spammy writing is really about sounding human.

The habits that actually keep you inboxed

None of them are about vocabulary.

1. Verify your list so bounces stay low.
2. Warm new domains and keep volume steady.
3. Send relevant mail people want, so complaints stay near zero.
4. Set up authentication properly and remove opt-outs at once.

Why the myth refuses to die

The word list survives because it is comfortable. It turns a hard systems problem, reputation, engagement, infrastructure, into an easy editing task anyone can do in ten minutes. Deleting the word "free" feels like progress; warming a domain for three weeks does not. Tool vendors keep the myth alive too, because a spam-word checker is an easy feature to build and demo, while genuine deliverability work resists being packaged into a traffic-light score.

There is also a survivorship story: someone removes a few words, the campaign coincidentally recovers because the mailbox provider's trust window shifted, and the word edit gets the credit. Correlation does the rest.

How filters got here

Early spam filters were rule-based: static scores for suspicious words, capital letters and exclamation marks, added up against a threshold. Spammers adapted within months, misspelling words and padding messages, and the arms race made static rules useless. Providers moved to statistical filtering, then to reputation systems that track how recipients treat mail from each sender, and today to machine-learning models weighing hundreds of signals at once. Word choice is one faint input among those hundreds, and it is evaluated in context, not against a blacklist.

The practical consequence: advice built for 2005-era filters is being applied to 2026-era systems. The filter judging your email today knows your domain's complaint rate, bounce history, volume pattern and engagement, and no synonym swap changes any of that.

Formatting signals that still carry weight

While vocabulary barely matters, a few content-level signals do feed the picture:

• Link count and link quality. Multiple links, link shorteners or mismatched anchor URLs pattern-match to phishing. One plain link, or none, is the safe shape for cold email.
• Image-heavy layouts. Emails that are mostly images with little text resemble promotional blasts. A plain-text style fits both filters and reader expectations for a one-to-one message.
• Attachments to strangers. Unsolicited attachments trigger security gateways. Send a link later in the conversation instead.
• Hidden tracking clutter. Heavy open-tracking pixels and rewritten links add machinery that corporate filters inspect. Less is safer.

How to find out where you actually land

1. Run a seed test: send the campaign to your own mailboxes across Google Workspace and Microsoft 365 and note where each lands.
2. Check authentication results in the message headers, all of SPF, DKIM and DMARC should pass.
3. Review bounce and complaint rates per sending inbox, not just per campaign.
4. Watch reply rate as a placement proxy: a sudden drop across all copy variants usually means a delivery problem, not a persuasion problem.
5. Make it routine, placement drifts over weeks, which is why ongoing deliverability monitoring beats one-off panic checks.

A note on European B2B inboxes

Most European business mail sits behind Google Workspace, Microsoft 365 or a corporate security gateway, and the gateways common in DACH enterprises are stricter than consumer filters: they weigh sender reputation, authentication and link behaviour heavily, and word lists not at all. If your mail clears Microsoft and Google on reputation, wording will not be what stops you in a German or Nordic corporate inbox. What does stop senders there is volume from unwarmed domains, so warm-up discipline matters more, not less, when targeting European enterprises.

Stop editing words, fix the system

Time spent hunting for trigger words is time not spent on the things that decide delivery: the list, the reputation, the relevance. Write like a human, keep the technical setup clean, and the occasional ordinary word will not hurt you. If you are in spam, the cause is almost always upstream of your wording, in deliverability fundamentals.