Do Spam Trigger Words Still Matter? A Practical Email Deliverability Guide

An isolated spam trigger word is not a reliable reason for an email to land in spam. Gmail,
Outlook and Yahoo do not publish a universal list of banned words, and copy alone does not
decide placement. Content still matters when it forms part of a broader pattern, such as false
claims, manufactured pressure, a misleading subject line or hidden text. If a campaign has just
lost inbox placement, check authentication, complaint history, list quality, sending consistency
and the full message before you start deleting words.

The short answer

Spam trigger words are a copy-review aid, not a placement rule. No major mailbox provider
publishes a fixed list, and no provider publishes the score it gives to a word such as "free" or
"urgent". A phrase can contribute to a risk score when it sits inside a suspicious pattern, but
the same phrase can be ordinary in a message from a sender with a good history and passing
authentication.

That leaves one useful conclusion: a word is a hypothesis, not a diagnosis. Rule out the layers
that can break placement on their own, then test the copy with the other variables held steady.

What a spam trigger word actually is

A spam trigger word is shorthand for a phrase that often appears in unwanted mail. It is not an
official filter category with a fixed score in Gmail, Outlook or Yahoo. Words such as "free",
"offer", "urgent" and "winner" became famous because they are common in phishing and low-quality
bulk mail. That history explains the advice. It does not prove that each occurrence causes a
filter to reject a message.

The word is only one part of a message pattern

A filter sees a complete message. It can compare the subject with the body, inspect the links and
formatting, and weigh the sending domain, authentication results, complaint history and recipient
behaviour around it. That means a phrase can contribute to a risk score without being the reason
the message was filtered.

Consider two subject lines:

Your free trial ends tomorrow
FREE TRIAL!!! CLAIM YOUR GIFT NOW!!!

The first is an ordinary account notice. The second combines a pressure cue, all caps, repeated
exclamation marks and a generic offer. No single token explains the difference. The pattern does.

The same applies to body copy. An honest reminder about a free shipping offer is different from a
claim that a recipient won a prize they never entered. A genuine deadline is different from a
deadline invented for urgency. The words may be identical while the context changes the result.

The same phrase can be safe or risky

Plenty of legitimate messages contain "urgent", "free", "limited time" or "final notice". Banks
send security alerts. Stores advertise free delivery. Billing systems send final notices. If a
filter blocked those words on sight, too much ordinary mail would disappear.

Filters can also react to combinations that a static word list cannot capture. Three pressure
phrases in one subject line, a subject that promises a reply that does not exist, or a clean
phrase wrapped in spammy formatting can all raise risk in a way that the phrase alone does not.

That is why "replace free with complimentary" is weak advice. It changes the sentence without
fixing the pattern, and it can make the offer less clear. A cleaner edit usually means replacing
manufactured pressure with a fact: a real date, a real deadline, a specific offer and a subject
line that tells the truth.

What Gmail, Yahoo and the law actually say

The useful evidence comes from the mailbox providers and the law, not from a recycled list of
words.

Google's published filter signals

Google describes Gmail's filtering as machine learning powered by user feedback. In its
overview of Gmail's spam filters,
Google names the signals it uses: characteristics of the sending IP address, domains and
subdomains, whether bulk senders are authenticated, and user input. Spam reports and the way
recipients move messages are part of that feedback.

Google does not publish a trigger-word list in that overview. It does not publish a percentage for
content, a point value for "free", or a formula that explains why one message was filtered. Any
site claiming that it knows the exact weight Gmail gives a word is guessing.

The short list is a public summary, not a full model of the filter. It does not prove that content
is ignored. It proves that the provider does not publish a word map that a sender can optimise
against. No provider publishes the weight it gives to content, so any page that assigns an exact
percentage to a word is presenting an estimate or a vendor test, not a Gmail or Yahoo rule. A
trigger list is a prompt to look, not a filter specification.

Google's sender requirements contain no word list

Google's email sender guidelines do address
message content, but not with a list of forbidden words. They require headers and content to be
accurate and not misleading. They say not to use Re: or Fwd: in a subject line unless the
message is a real reply or forward. They say visible links should be understandable and that
hidden HTML or CSS content can cause a message to be marked as spam.

The same guidelines set the technical and behavioural requirements: SPF or DKIM for all senders,
SPF and DKIM plus DMARC for bulk senders, valid forward and reverse DNS, TLS, RFC 5322
formatting, one-click unsubscribe for marketing and subscribed mail from bulk senders, and a
complaint rate below 0.3 percent. Google's sender FAQ
says bulk senders should keep user-reported spam below 0.1 percent and prevent it reaching 0.3
percent. These sender guidelines apply to mail sent to personal Gmail accounts, not to Google
Workspace recipients.

None of that says "urgent" is banned. It says the message must be accurate, the sender must be
identifiable and authenticated, and recipients must not have grounds to complain.

Accurate content is a requirement, a generic phrase is not

Yahoo's sender best practices start with audience:
send timely and relevant mail to people who requested it, keep complaint rates low and remove
invalid recipients. The guide also says not to use deceptive subject lines.

The US Federal Trade Commission makes the legal version of that rule explicit in its
CAN-SPAM compliance guide:
"Don't use deceptive subject lines. The subject line must accurately reflect the content of the
message." The problem is deception. A real deadline is not deceptive. A fake urgency tactic can
be.

When copy can still affect delivery

Content is not a harmless detail. It becomes relevant when it fits a pattern that a recipient or
a filter has learned to distrust.

Claims, pressure and subject-body mismatch

The riskiest copy problems are usually accuracy problems:

a subject that promises a reply, invoice or account notice when none exists
a reward, discount or outcome claim that is not true
urgency without a real deadline
the same pressure phrase repeated in the subject, preheader, headline and button
an offer that the body does not explain
a sender identity that the reader cannot verify

These patterns can create complaints and deletions. A complaint rate is a direct signal that
recipients did not want the mail. Gmail defines its spam rate as the share of messages delivered
to an engaged recipient's inbox and then marked as spam. It says rates above 0.1 percent already
have a negative effect on inbox delivery for bulk senders.

Formatting, links and structural signals

The shape of a message can matter as much as its vocabulary. Risk can come from all caps,
punctuation walls, image-only messages, a missing plain-text alternative, hidden text, broken
HTML, unclear unsubscribe links, shortened URLs, misleading display text or a new tracking domain
with poor history.

Google's requirement for visible links is practical. If the text says a reader will see an order
confirmation and the link leads somewhere else, the message is misleading before a filter even
looks at it. The same is true of an unsubscribe control that is hard to find.

A word checker may notice a few of these patterns. It cannot see whether the claim is true, whether
the list is appropriate, whether the link domain is trusted or how recipients reacted.

A corporate gateway can apply its own rules

Mailbox providers are not the only filter. A company mail gateway can quarantine a message for
local reasons, such as a banned phrase, an attachment rule, a link policy or an impersonation
check. A message can pass Gmail and still fail in one corporate network.

Treat that as a separate observation. If only one recipient organisation reports the problem,
ask its mail team which rule fired. Do not generalise a local rule into a claim about Gmail,
Outlook or Yahoo.

What to check before you rewrite one word

When placement changes, work through the layers in order. Each step rules out a different cause.
Rewriting copy first can hide the real fault while making the message harder to read.

Step	What to inspect	What it rules out
1	SPF, DKIM and DMARC results on the failing message, plus alignment	Broken records, unsigned mail, misaligned senders
2	Complaint rate and provider reputation trends	Messages that recipients did not want
3	List quality, permission and unsubscribe handling	Invalid, stale or unwanted recipients
4	Volume, sending times, new domains and new IPs	Sudden changes in sending behaviour
5	The complete message: subject, body, links, formatting and HTML	Deceptive or structurally suspicious content
6	A controlled copy test	A copy-related cause, once the other variables are steady

Check the message, not just the DNS zone

The fastest way to misdiagnose this is to check that a domain has an SPF record and then start
rewriting subject lines. A published record says the domain is configured. It does not prove that
the message you sent passed authentication or that its From domain aligned with the
authenticated domain.

Read the Authentication-Results header from a real message. That is stronger evidence than an
abstract DNS check because it shows what the receiving chain saw.

For the authentication part of this order, DMARCFlow's free
DMARC checker,
SPF checker and
Email Header Analyzer cover the record side
and the message side. The header analyzer parses SPF, DKIM and DMARC results and the delivery
route. For an unexpected source, DMARCFlow's paid dashboard reads DMARC aggregate reports and
lists the sending sources behind them. None of these checks scores copy, measures complaints or
predicts inbox placement.

Promotions is not the Spam folder

Gmail's inbox layout documentation describes
Primary, Social, Promotions and Updates as tabs inside the Default inbox. A message in Promotions
has reached an inbox category. It has not landed in the Spam folder.

That distinction changes the diagnosis. If the campaign moved from Promotions to Spam, the
problem is serious and the complaint, authentication and list layers deserve a check. If it moved
from Primary to Promotions, the change may be a classification shift rather than a delivery
failure. Record the folders separately. "It is not where it used to be" is not a measurement.

How to test whether the copy caused the placement change

A copy edit followed by better placement is not proof. It may be a coincidence, a different
audience segment or a provider updating its filters. A controlled test turns the copy hypothesis
into something you can check.

Write the hypothesis before the test. For example: "The generic prize subject is causing the
Hold the other variables steady. Keep the sending domain, IP, DKIM key, audience mix,
Change one thing. Test the subject line before changing the body, links and template together.
Use comparable, consenting segments. Randomly split the audience so both versions reach
Measure the right outcomes. Track placement by mailbox provider, complaint reports, unsubscribes
Repeat. One result can be noise. A repeatable difference across comparable sends is evidence.

A placement panel or reputable placement-testing service can help with the inbox-placement part. A
Postmaster Tools account adds complaint-rate,
reputation, authentication and delivery-error data for mail to personal Gmail accounts. Neither
tool replaces a controlled comparison, and Postmaster Tools does not measure where every message
landed.

What a spam-trigger checker can and cannot prove

A content-only checker can inspect the message you give it. It can flag repeated pressure
phrases, excessive punctuation, obvious formatting problems and some link patterns. That is
useful for copy review.

It cannot see the sending domain's complaint history, the recipients' reactions, the receiving
provider's decision, a hidden sending-source problem or a broken authentication chain. A high
content score is not a placement prediction. A low score is not proof that the copy caused spam.

The most defensible use is as a copy audit, not as a deliverability verdict. For the same reason,
run the header check on both versions of a test. DMARCFlow's Email Header Analyzer can confirm
that authentication stayed healthy while the subject changed. That removes one competing
explanation. It does not prove that the subject caused the result.

Frequently asked questions

Do spam trigger words still matter in 2026?

Yes, but not as a fixed list. Content can contribute to a risk score when it forms part of a
suspicious pattern. An isolated phrase such as "free" is not a reliable predictor of where a
message will land.

Does the word "free" cause an email to go to spam?

No single word has that effect across Gmail, Outlook and Yahoo. "Free" can appear in a normal
trial confirmation or a shipping offer. It becomes riskier when it sits inside pressure, false
claims, poor formatting and a weak sending history.

Does Gmail have an official list of spam words?

No. Google's published sender guidelines and spam-filter overview do not include a list of words
that cause spam. They describe authentication, DNS, TLS, complaint rates, unsubscribe handling,
accurate content and other signals.

Can one word send an email to spam?

One word can be part of a pattern that makes a message more suspicious, but it cannot be treated
as the cause without evidence. Check the authentication, complaint, list, volume and content
layers, then test the copy in isolation.

Should I rewrite my subject line before checking SPF, DKIM and DMARC?

No. A subject edit cannot repair failed authentication, an unexpected sending source or a
misaligned domain. Check those first. They are fast to test and they explain more placement
changes than vocabulary does.

Are spam-trigger word checkers reliable?

Useful for finding copy patterns, unreliable as placement predictors. They cannot observe sender
reputation, complaint history, recipient behaviour or the receiving provider's decision.

Does landing in the Promotions tab mean the email was marked as spam?

No. Google describes Promotions as one of the tabs in the Default inbox. Track Promotions,
Primary, Spam and rejection separately so the measurement matches the problem.

What should I check first when a campaign suddenly lands in spam?

Check the authentication results on the failing message, then complaint-rate trends, list and
permission state, and sending-volume changes. Review the complete message after those layers are
clean.

Does changing "free" to "complimentary" improve deliverability?

Not reliably. It can remove a real offer, make the copy awkward and leave the actual cause in
place. Replace pressure with facts and fix the sending signals before polishing synonyms.

How do I know if content or sender reputation caused the placement change?

You need a controlled comparison. Hold the other variables steady, change one element, use
comparable recipients and measure placement and complaints by provider. A placement change after
an edit is a correlation, not proof.

The practical verdict

Spam trigger words are a starting point for copy review, not a verdict on inbox placement. There
is no official list to clear, and no synonym swap that repairs a broken sending setup. Check the
message authentication first, then complaints, list quality, sending consistency and the full
message. If content still looks like the cause, test one change against a comparable audience.

Fix the broken layer before polishing the copy. A word is a hypothesis. A diagnosis needs
evidence.