Does a single “free” in the subject line send an email to spam? Almost never. Gmail and Yahoo filters add up signals: domain reputation, authentication, recipient complaints, message structure and vocabulary. Email spam trigger words count in that sum, with a modest weight next to sender reputation. Below are about a hundred words and phrases to watch, sorted by category, each with a concrete alternative. Then come the factors that really decide the outcome: complaint rate, authentication, one-click unsubscribe and list condition.
Is a single spam word enough to block an email?
No. A spam filter scores a message on a set of signals, and vocabulary is only one of them. SpamAssassin, the open source filter from the Apache foundation, adds up the points of dozens of rules and flags a message as spam from 5 in its base configuration (the required_score parameter is 5); its wiki even advises new users to raise that threshold to 6 or 7 to limit false positives (Apache SpamAssassin, BasicConfiguration page). A lone word moves a counter without making it overflow. The filter also looks at the whole message (links, images, headers) and at the sender’s history (domain, IP address). A suspect word in an authenticated send from a domain with no complaints passes without a sound. The same word in a send from a recent, unauthenticated domain worsens a file that is already heavy.
A CRM team whose campaigns land in junk without any change in content usually blames the subject line first. The culprit more often sits in domain reputation (see below). Still, words do matter when they pile up: two money promises combined with capital letters and exclamation marks draw a profile that filters recognize at once.
Spam trigger words by category, with an alternative for each
The tables below list the terms most associated with spam in English, grouped by family of triggers. The right-hand column gives an alternative for each spam word: a wording that keeps the information and removes the trigger. A word on this list is not banned. It needs a clean context and rare use.
Promises and exaggerations
These terms announce a benefit without putting a number on it. Replacing the promise with the real condition almost always solves the problem.
| Word or phrase | Why it weighs | Alternative |
|---|---|---|
| Free, 100% free, free trial | First word on the usual spam lists; capitals and the percent sign make it worse | “No commitment for 14 days” (if the duration is exact) |
| Exclusive offer, special offer, incredible deal | Superlative with no verifiable content | “Price reserved for customers who sign up before November 30” |
| Satisfaction guaranteed, 100% guaranteed | Vague guarantee that recalls mass mailings | “Returns accepted within 30 days” |
| No fees, no obligation | Financial promise with no condition | “Shipping included from $49” |
| Gift, free gift, bonus gift | Evokes contests and prize draws | “A free sample with your first order” |
| Refund, money back | Associated with promises of gain | “Returns possible within 14 days, per our terms” |
| Discount, bonus, prize | Commercial terms common in bulk subjects | “20% off the X range until November 15” |
| Dear friend, don’t delete, confirm now | Imitate phishing messages | Describe the action: “Your customer account access expires on the 10th” |
Urgency and pressure
Manufactured urgency is the most common lever of fraudulent campaigns. A real, dated deadline works better than an adjective.
| Word or phrase | Why it weighs | Alternative |
|---|---|---|
| Urgent, hurry, act now | Pressure with no deadline | “Registration closes Wednesday at 6 pm” |
| Limited time, limited offer, while supplies last | Scarcity nobody can verify | “Last day: November 15” |
| Order now, don’t wait, don’t delay | Command repeated across bulk sends | “Book your slot for next week” |
| Click here, click below | Meaningless anchor, very common in phishing | A link that names its destination: “view the November invoice” |
| Buy direct, call now, apply now | Raw sales order | “Book” or “Schedule a call” inside a button |
| Last chance, final notice, once in a lifetime | Pressure and filler | “Only 12 left in stock” |
Winnings and prize draws
Any lottery vocabulary joins the prize scams. If the email really announces a contest, a published set of rules makes it more credible.
| Word or phrase | Why it weighs | Alternative |
|---|---|---|
| Congratulations, you’re a winner | Announces a gain nobody asked for | “Your order is confirmed” |
| You have won, winner, you won | Typical sentence of fake lotteries | “Results of the October 12 contest” |
| You have been selected, you have been chosen | Generic flattery sent to the whole list | Cite the criterion: “You attended 3 of our webinars” |
| Sweepstakes, lottery, prize | Gambling vocabulary | “Contest: rules at the bottom of the message” |
| Double your, profit, jackpot | Promise of multiplication | Quantify the real benefit: “2 months free on the annual plan” |
| Opportunity, just for you, exclusive deal | Hollow mass-mailing formulas | Name the offer and its audience: “For CRM managers” |
Money and payment
Banking terms attract filters because invoice and wire fraud use them. A legitimate invoice is recognized by its verifiable references.
| Word or phrase | Why it weighs | Alternative |
|---|---|---|
| Best price, lowest price | Superlative nobody can check | “2026 pricing: $29 per month” |
| Make money, extra income, earn cash | Core business of scams | Describe the actual role: “Partner referral program” |
| Investment, interest rate, bankruptcy | Consumer finance, heavily monitored | A precise product term: “loan simulation” |
| Save big, cash, cash bonus | Easy money | “Cut your shipping costs by 20%” |
| Credit card, Visa, Mastercard, PayPal | Trips anti-phishing rules in the subject | Keep payment methods for the body or the checkout page |
| Wire transfer, check, debit | Vocabulary of fake invoices | “Payment details are in your customer account” |
| Invoice, billing | Frequent suspicious attachment | “Document no. 2026-0412 from Dupont Ltd” |
| Insurance, refund, earnings | Sector with a high volume of fraud | “Certificate for your home insurance policy” |
Necessity, desperation and privacy
These words play on emotion or intimacy. A B2B sender rarely has a use for them; a dating site has to compensate with an impeccable reputation.
| Word or phrase | Why it weighs | Alternative |
|---|---|---|
| Help me, I need help | Imitates personal-emergency scams | “A question about your file” |
| Desperate, desperation | Emotional blackmail | Describe the need factually |
| Single, hookup, love | Dating sites and fake profiles | “New profiles near you” for a genuine dating service |
| Sexy, adult | Adult content | Describe the product: “evening collection” |
Marketing and sales
Pure sales jargon lowers the credibility of a subject line. A concrete benefit replaces it.
| Word or phrase | Why it weighs | Alternative |
|---|---|---|
| Sale, flash sale, clearance | Massive seasonality of sends | “Last day of the members’ sale: Thursday” |
| Buy, order | Direct sales verb | “Your cart is waiting” |
| Sign up now, subscribe | Repeated command | “The November 18 webinar: agenda and times” |
| Marketing solution, increase sales, revenue | Cold outreach jargon | The expected result: “reduce your bounces by 30%” if you can prove it |
| Easy, easily, no effort | Promise of simplicity without proof | Describe the duration: “set up in 10 minutes” |
Health and pharmacy
Health, weight loss and food supplements face the strongest filter pressure. Words that promise a bodily result are the most penalized.
| Word or phrase | Why it weighs | Alternative |
|---|---|---|
| Lose weight, weight loss, slim down | Bodily promise | “8-week meal program supervised by a dietitian” |
| Wrinkles, anti-aging, look younger | Cosmetics that promise results | Name the active ingredient: “hyaluronic acid cream” |
| Hair loss, erection, performance | Historical categories of pharma spam | Point to a consultation: “Book an appointment” |
| Prescription, doctor, pills | Imitates illegal online pharmacies | Cite the professional: “Dr. Martin, cardiologist” |
| Anxiety, stress relief | Hijacked by miracle products | “Stress management workshop, Tuesday 6 pm” |
| Effective, miracle, guaranteed results | Promise of a guaranteed outcome | A study figure cited with its source |
Subject lines before and after
Rewriting rule: replace the adjective with a verifiable fact. Our guide to eye-catching email subject lines goes further on the craft.
| Before | After | What changes |
|---|---|---|
| URGENT!!! Exclusive offer 100% FREE | Your 14-day trial starts today | One dated fact replaces three triggers |
| Congratulations, you won a free gift | Results of the October 12 contest | Real context replaces flattery |
| Best price guaranteed, buy now | Our 2026 pricing: from $29 per month | A precise price replaces the superlative |
| Click here for your invoice | Document no. 2026-0412 is available in your customer account | The reference proves authenticity |
| Make money easily | Referral program: how it works | The program is named |
| Last chance!!! Limited time | Registration closes Wednesday at 6 pm | A dated deadline replaces scarcity |
| You have been selected for a special offer | 3 webinars attended: here is your certificate | A real criterion replaces the generic formula |
The factors that decide where the email lands
Gmail and Yahoo judge the sender first, then the message. In October 2023 Google and Yahoo announced numeric thresholds, applied since February 2024 to sends to Gmail and Yahoo Mail.
Complaint rate and sender reputation
Google asks senders to keep the spam rate shown in Postmaster Tools under 0.3%; it recommends staying under 0.10% and avoiding reaching 0.30% (Google, email sender guidelines, rules in force since February 1, 2024). Yahoo sets the same 0.3% limit for all senders, calculated on messages delivered to the inbox, with enforcement announced from February 2024 (Yahoo Sender Hub). On a campaign of 100,000 messages, 0.3% means 300 complaints against 100 for 0.10%; the 0.10% threshold equals one complaint per 1,000 delivered messages. A list that exceeds the limit suffers across all of its sends, whatever the vocabulary of the subject line.
Authentication and Gmail and Yahoo requirements
Since February 1, 2024, Google requires senders of more than 5,000 messages per day to Gmail accounts to authenticate with SPF (RFC 7208), DKIM (RFC 6376) and DMARC (RFC 7489), a p=none policy being enough. Yahoo asks bulk senders for SPF and DKIM and a DMARC policy of at least p=none with From-domain alignment (Yahoo Sender Hub, 2026). The mechanisms are explained in our guide to SPF, DKIM, DMARC and BIMI.
One-click unsubscribe
Google requires one-click unsubscribe from bulk senders. The standard that describes it, IETF RFC 8058 (January 2017), relies on the List-Unsubscribe-Post: List-Unsubscribe=One-Click header. Yahoo asks for a working List-Unsubscribe header and requests processed within 2 days. The reason lies in recipient behavior: without a simple way out, people click “report spam”, which feeds the complaint rate directly.
Engagement, message structure and text-to-image ratio
Opens and replies point to an interested audience. Complaints and quick unsubscribes point to the opposite. Shortened links (bit.ly) hide the destination and come up often in fraudulent campaigns: a link on the sender’s own domain reassures more. Writing to contacts who never consented (double opt-in is recommended) also multiplies complaints. On structure, a message made of one large image leaves filters no text to analyze. Our article on emails that contain only images covers this point.
Subject, body and language
The subject is read before the email is opened and shapes the recipient’s first judgment; the body also weighs in the score, along with its links. Language does not protect you: the reference lists are mostly in English, and the levers described above (urgency, winnings, easy money, health) show up in every language under different words. Words from this list therefore apply in their translated versions too: switching language does not get around the filter.
Hard bounces and list hygiene: the lever behind reputation
A hard bounce (permanent failure, SMTP 5xx code, defined by RFC 5321) happens notably when the address does not exist or no longer exists, for example with code 550 5.1.1 (“destination address does not exist”, RFC 3463). Mailbox providers read the hard bounce rate as an indicator of list quality: a domain that regularly writes to nonexistent addresses looks like a sender that bought or scraped its database. Sender reputation then drops. Perfectly written messages go to junk along with the rest.
A soft bounce (SMTP 4xx code: full mailbox, server temporarily unavailable) is monitored before any removal. List hygiene means removing hard bounces, addresses inactive for a long time and contacts who complained. An abandoned address can also be recycled into a spam trap: writing to a trap signals a poorly maintained list.
Catch-all domains complicate the check: they accept any address at the time of SMTP reception, so an invented address can look valid and bounce later. Only a conversation with the mail server reduces that risk; the syntax of the address reveals nothing. A simple internal rule works well: a hard bounce leaves the list after the first send, and so does a soft bounce repeated 3 times in a row.
To see where your database stands, the simplest method is to run a sample of the list through CaptainVerify before the next campaign: the tool flags invalid addresses before sending, which gives an idea of the hard bounce rate to expect.
Testing an email before sending
3 checks combine at low cost. A spam score test (Mail Tester, for example) gives a grade and the rules triggered; SpamAssassin, cited above, produces the same kind of score. Seed testing (sending to test mailboxes at Gmail, Outlook and Yahoo, for example with GlockApps) shows the real destination folder. Finally, Google Postmaster Tools tracks sender reputation and the spam rate over time.
These tests have a limit: a message that passes in testing can fail in production if the list is bad.
Frequently asked questions about spam words
Is one spam word enough to end up in spam?
Generally not. Filters add up signals: sender reputation, authentication, recipient complaints and vocabulary. A single sensitive word, in an authenticated message sent to a list that asked for the content, does not usually trigger any block. Several weak signals piling up, however, tip a message over.
Do forbidden words apply to the subject line or the body?
Both. The subject drives the open and weighs in the analysis; the body also weighs in the score. A suspect phrase in the subject adds to the signals from the body. Keep sensitive words for cases where they are indispensable, in a precise context.
Do capital letters and exclamation marks count as much as words?
They weigh in the same sum. An all-caps subject with several exclamation marks adds spam points to every sensitive word it contains. A normal sentence with a single word from the list adds far fewer.
Do I also need to avoid words from other languages?
Yes. Filters analyze every language, and the same triggers circulate in translation (gratuit, gagnant, gratis, kostenlos). Dropping a foreign “free” into an English email stacks two signals.
What can replace “free”?
The exact condition: “no commitment for 14 days”, “shipping included from $49”, “a sample with your first order”. A duration or an amount replaces the adjective and reassures the recipient as much as the filter.
