What words or phrases trigger spam filters? (e.g., "free," "Viagra," urgency words)

Still have a question, spotted an error, or have a better explanation or a source we should cite?

Short answer: word lists matter less than you think, but they are not zero. The classic offenders still exist in filter rule sets. The longer answer is that filters score the whole message, not single words, so "free" inside a clean newsletter from a warm sender is not the same signal as "FREE!!! ACT NOW" from a cold IP with no DKIM.

Here is what actually happens. Open-source filters like Apache SpamAssassin ship with hundreds of named rules, and many of them do trigger on specific words and phrases. A few real rule names from the SpamAssassin rule set:

  • FREE_PRICE flags "free" near a price.
  • URG_BIZ flags urgent business language.
  • MILLION_USD flags large dollar amounts in body text.
  • LOTS_OF_MONEY flags repeated money references.
  • VIAGRA_OBFU flags obfuscated drug-name spellings like "v1agra" or "v-i-a-g-r-a".
  • DEAR_SOMETHING flags generic "Dear Friend" or "Dear Winner" openers.

Each rule adds a small score. Hit one and you are fine. Hit five and you sit at the edge. Combine those hits with a missing DKIM signature, a new sending IP, and a complaint rate above 0.3 percent and you are in the spam folder. SpamAssassin's published rule list documents the exact rules and weights if you want to see the full menu.

Major mailbox providers (Gmail, Yahoo, Outlook) do not publish their internal rule sets, but their filters work on the same principle: small signals stacked up. Google's bulk sender guidelines say outright that they look at content patterns together with authentication, complaint rate, and engagement (Google sender guidelines). They do not say "do not use the word free." They say keep your complaint rate under 0.3 percent and authenticate with SPF, DKIM, and DMARC.

The categories that still hurt you in 2026:

  1. Pharma and adult terms. Viagra, Cialis, weight loss miracles, CBD without context. These get flagged hard because spammers still abuse them daily.
  2. Money and urgency stacked together. "Limited time" plus "act now" plus "guaranteed" plus a dollar sign in the subject line. One of these is fine. Four of them in one email reads as a Nigerian-prince template.
  3. Obfuscation. Replacing letters with numbers ("FR33", "V1AGRA"), inserting hidden characters, using ALL CAPS for whole sentences. Filters are trained to spot the workaround, not the word.
  4. Phishing language. "Verify your account", "your account will be suspended", "click here to confirm". These trip rules even for legitimate senders if the rest of the message looks off.

What does not matter much anymore: the word "free" on its own, "newsletter" in the subject, the word "click", the word "promotion". Filters expect commercial mail to contain commercial language. For more on why context beats keywords, see how spam filters decide if an email is spam and heuristic rules in spam filtering.

The practical advice has not changed in ten years. Authenticate properly. Keep complaints low. Write to people who asked to hear from you. Do not stack three urgency words into one subject line. Do not obfuscate. The rest of the keyword-avoidance folklore (no "free", no exclamation marks, no red text) is overstated. If you are a clean sender with a warm domain, "free trial" in your subject line will not put you in spam. If you are a cold sender with no DKIM and 2 percent complaints, changing "free" to "complimentary" will not save you either.

For the bigger picture on what filters actually look at, see what signals do spam filters look at.

Contributors

Who worked on this answer

Every name links to their profile. Every company links to their site. Real people, real accountability.

Ask an AI · tailored to your setup

Learn how modern spam filters really evaluate content.

I'm writing subject lines and body copy for an upcoming campaign and I want to avoid spam filters. I know words like "free" and "act now" can be risky, but I also need to be compelling. How do modern spam filters actually work, are they just scanning for bad words, or is it more nuanced? What should I actually worry about vs. what's outdated advice?

Edit the yellow boxes, then send to the AI of your choice.