How spam filters actually score your newsletter
TL;DR
Spam filters score newsletters across five factor groups - subject-line risk, body composition, link hygiene, authentication signals, niche calibration. Sender reputation explains 60-70 percent of the placement decision; content explains 20-30 percent; authentication the rest. Score above 80/100 lands reliably; 60-80 throttles to Promotions; below 60 spam-folders.
"Will this land in spam?" is the wrong question. The right question is: "What is the cumulative score, and which factor groups are dragging it down?" Spam filters do not flip a coin or run a single regex. They run a weighted sum across five factor families, score the message in milliseconds, and route accordingly. Understanding the factor families lets you predict the outcome before send.
Factor group 1: subject-line risk
The first factor filter looks at, because it is the first signal available. The same factors that drive subject-line open rate also drive spam classification - the two systems are not independent. A subject with three exclamation marks reads as urgent to a human and as spam-like to a filter, because the historical training data linked one to the other.
Specific subject-line triggers: ALL CAPS over 30 percent of non-space characters, three or more exclamation marks, stacked urgency words ("FREE ACT NOW GUARANTEED"), use of restricted financial terms ("guaranteed return," "no risk") in regulated jurisdictions, and unusually long subjects (over 90 characters) which correlate historically with spam-template reuse.
Factor group 2: body composition
The body of the email contributes about 25 percent of the content score. The factors filters track:
- Image-to-text ratio. Over 70 percent reads as a promotional template. Under 30 percent is the safest band. The historical reason: the early 2000s spam-image attack. Filters now penalise heavy image use to compensate.
- Hidden text. CSS like
font-size: 0,color: #fffon a white background, ordisplay: nonewrapping keyword-stuffed text. Old trick, well-detected, never use. - Broken HTML. Malformed tables, unclosed tags, stray script tags. Templates that pass through too many ESPs accumulate cruft.
- Suspicious URLs in the body. Naked IP addresses, unicode-confusable characters in domain names, mismatched anchor text vs. href.
Factor group 3: link hygiene
Every URL in your email contributes to the score. Three patterns drive most penalties:
- URL shortener use (bit.ly, t.co, tinyurl). Filters cannot inspect the destination cheaply, so they treat shorteners as risk.
- Redirect chains. URL A redirects to URL B redirects to URL C. Each hop adds risk weight.
- Mismatched anchor text vs. href.
<a href="https://attacker.com">https://yourbrand.com</a>is the canonical phishing pattern.
The fix for link wrapping: use your ESP's branded link wrapper rather than a public shortener. Mailchimp's list-manage.com, Klaviyo's tracking domain, Iterable's links.iterable.com all carry pre-warmed reputation and don't trip shortener heuristics.
Factor group 4: authentication signals
Authentication is the single most underweighted factor by content writers and the single most weighted factor by Gmail and Outlook. SPF, DKIM, and DMARC together establish whether the message is from who it claims to be. Without all three properly aligned in 2026, high-volume senders are spam-foldered or rejected outright by major mailbox providers.
- SPF: a DNS TXT record listing servers allowed to send for your domain. Required.
- DKIM: a cryptographic signature that proves the email body was not tampered with. Required for any volume.
- DMARC: a policy on what to do when SPF/DKIM fail.
p=quarantineorp=rejectwith reporting on.p=nonegives reporting but no protection. - List-Unsubscribe header: required by Gmail and Outlook for bulk senders since 2024. Must be present.
Factor group 5: sender reputation (the hidden weight)
The factor that explains 60-70 percent of placement is invisible to content writers: sender reputation. It is built from your domain's send history, your IP's send history, the complaint rate against your sends, the engagement rate from recipients, and the historical pattern of bounce-vs-deliver ratios.
You cannot fix reputation in a single send. You can degrade it in a single send by mailing a stale list and getting a 0.5 percent complaint rate. Once degraded it takes weeks of clean sends to recover. The repair process: shrink your audience to your most engaged 50 percent, mail them only, and let reputation rebuild before you re-add the rest.
The 80/100 threshold
If a single number is what you want, here it is: score 80+ on a calibrated content + authentication + reputation model and you land in inbox at major providers. Score 60-80 and you typically deliver but get throttled or routed to Promotions in Gmail. Below 60, spam-folder probability climbs sharply.
85-100: Inbox-safe at Gmail, Outlook, Apple Mail. Free of obvious flaws.
70-85: Mostly inbox; some Promotions routing in Gmail.
60-70: Inbox-Promotions split. Watch unsubscribe rate carefully.
40-60: Spam folder probability ~30%+ at major providers.
Under 40: Effectively undeliverable to most addresses.
Score your draft before you send
The spam score guide covers the methodology and benchmarks. The footer compliance checker validates List-Unsubscribe, physical address, and consent language.
Check your spam score →Frequently asked questions
What is a good spam score?
Above 80. Between 60-80 typically delivers but throttled. Below 60 has high spam-fold probability.
Top triggers in 2026?
3+ exclamation marks, ALL CAPS >30%, image-to-text >70%, missing List-Unsubscribe, URL shorteners, urgency stacking.
Sender reputation vs. content?
Reputation = 60-70% of placement. Content = 20-30%. Auth = the rest. Great content from a poor sender still spam-folders.
Image-to-text ratio?
Over 70% looks promotional. Under 30% safest. Modern filters compensate for the historical image-attack pattern.
Why does List-Unsubscribe matter?
Gmail and Outlook require it for bulk senders since 2024. Absence is a strong negative signal.