KoltrixSpam filter myths: trigger words, images and link counts
Modern filters care far more about reputation and engagement than single words. Which content folklore is outdated and which content issues still matter.
On this page(12 sections)
- How modern filtering weighs things
- Myth 1: certain trigger words send you to spam
- Myth 2: images get you filtered
- Myth 3: too many links trigger filters
- Myth 4: spam testing tools tell you if you will be inboxed
- Myth 5: HTML email is riskier than plain text
- Myth 6: a spam folder problem is a content problem
- Myth 7: you can fix placement by tweaking the From name
- Content issues that still matter
- Where to spend effort instead
- Checklist for content
- Bottom line
Somewhere, right now, a marketer is rewriting "FREE" as "complimentary" because a blog post from 2009 said it would keep the email out of spam. Content still matters to filters, but not in the way the folklore suggests.
Most of what decides placement today happens before a filter reads a single word.
How modern filtering weighs things
Large mailbox providers combine many signals, and they do not publish their weights. Broadly, practitioners agree the order of importance looks something like this:
- Authentication: can the receiver verify who sent the message (SPF, DKIM, DMARC alignment)?
- Reputation: what is the history of the sending domain, IP and link domains?
- Engagement: how do recipients, collectively and individually, react to this sender's mail?
- Content and structure: what does the message itself look like?
Content is real, but it is usually the tiebreaker, not the verdict. A well-authenticated sender with a strong reputation and engaged recipients can send a message with "free" in the subject line without trouble. A new domain with no authentication will struggle no matter how carefully it avoids certain words.
With that framing, here is how the common myths hold up.
Myth 1: certain trigger words send you to spam
The idea comes from early rule-based filters that literally added points for phrases. Some open source filters still contain word and phrase rules, but they carry modest weights and are combined with many other tests. The large mailbox providers rely on machine learning trained on enormous volumes of mail and user feedback, where a single word means little on its own.
What is true: content that looks like spam as a whole still scores badly. Aggressive urgency, misleading subject lines, all-caps shouting and deceptive claims correlate with spam. The problem is not a word list; it is a pattern.
Myth 2: images get you filtered
Image-heavy marketing mail lands in inboxes every day. Images are not a problem by themselves.
What is true: messages that consist of a single large image with little or no text are a classic spam technique for hiding content from text analysis. Keep real text in the message, include alt text, and provide a plain-text alternative. A reasonable balance of text and imagery is the safe zone; a specific "60/40 ratio" rule is folklore.
Myth 3: too many links trigger filters
There is no magic link count.
What is true: link reputation matters a great deal. One link to a blocklisted domain or a widely abused URL shortener does more harm than twenty links to your own reputable domain. Links whose visible text shows one domain while the actual URL points to another are a phishing signal. Use links on your own domains, ideally a branded tracking domain, and make the visible text honest.
Myth 4: spam testing tools tell you if you will be inboxed
Tools that run a message through a rule-based filter and produce a score are useful for catching obvious problems: missing headers, broken HTML, authentication failures, blocklisted links.
What is true: they cannot simulate the reputation and engagement data a large provider holds about you and about each recipient. A perfect score means you avoided obvious mistakes, not that you will reach the inbox.
Myth 5: HTML email is riskier than plain text
Both formats reach inboxes. Most commercial mail is HTML.
What is true: sloppy HTML can hurt. Broken markup, hidden text, huge embedded styles copied from a web page and invisible elements stuffed with keywords all look suspicious. Clean, simple HTML with a plain-text alternative is standard practice and does not count against you.
Myth 6: a spam folder problem is a content problem
When mail suddenly goes to spam, teams often start by rewriting copy.
What is true: sudden changes usually come from somewhere else: a DNS change that broke DKIM, a new list source with old addresses, a volume spike, a link domain that got listed, or rising complaints. Check authentication and reputation signals first. Content changes rarely cause overnight shifts unless you added something specific, such as a new link domain.
Myth 7: you can fix placement by tweaking the From name
A recognizable From name helps recipients decide to open your mail, and consistency helps them trust it. It does not override reputation.
Content issues that still matter
Folklore aside, some content and structure problems genuinely hurt:
| Issue | Why it hurts |
|---|---|
| Deceptive subject lines ("Re:" on a first message, fake urgency) | Pattern strongly associated with spam and with complaints |
| Mismatched link text and URLs | Phishing signal |
| Links to blocklisted or shortener domains | Link reputation is scored directly |
| Image-only messages | Hides content from analysis |
| Missing plain-text part or malformed MIME | Looks machine-generated by low-quality tools |
| Broken or bloated HTML | Signals low-quality senders |
| Attachments of risky types | Executables and macro documents are routinely blocked |
| No visible unsubscribe on bulk mail | Drives complaints; one-click unsubscribe is required for bulk senders at Gmail and Yahoo |
Where to spend effort instead
If you want better placement, these matter far more than word choice:
- Authenticate everything with SPF, DKIM and DMARC, aligned with your From domain.
- Mail people who want it. Remove addresses that have not engaged in months.
- Make unsubscribing easy. A complaint costs far more reputation than an unsubscribe.
- Keep volume steady. Avoid sudden spikes.
- Use your own link domains with HTTPS.
- Separate streams so marketing behavior does not affect transactional mail.
- Watch complaint and bounce rates and act quickly when they rise.
Checklist for content
- No deceptive subjects or misleading "Re:" and "Fwd:" prefixes.
- Honest link text, links on your own domains.
- Real text in the message, alt text on images, plain-text alternative included.
- Clean HTML without hidden text.
- Visible unsubscribe on bulk mail, plus one-click unsubscribe headers.
- No risky attachment types.
Bottom line
Trigger-word lists and magic ratios are relics of old rule-based filters. Modern filtering decides mostly on authentication, reputation and engagement, and treats content as one signal among many. Avoid deceptive patterns, keep links and HTML clean, and put the rest of your effort into who you mail and how they respond.
Start with Koltrix
Your domain, one inbox, and an API that sends.
A team inbox where AI sorts and drafts (nothing is sent without your click), plus the transactional API and SMTP relay your product sends with. 7 days free, no card.
