How Email Spam Filters Work: A Technical 101 for Cold Email Operators

By Puzzle Inbox Team · May 16, 2026 · 8 min read

Email spam filters use multiple signals to decide whether your message lands in inbox or spam. Here is how they actually work — and how to stay out of spam.

What Spam Filters Actually Do

Email spam filters analyze every incoming message and assign it a spam score. Above a threshold, the message goes to the spam folder. Below it, the message lands in the primary inbox. The score is calculated from dozens of signals across three main categories: sender reputation, content analysis, and recipient engagement.

Understanding how these signals work lets you optimize cold email to avoid spam classification.

Category 1: Sender Reputation Signals

The biggest factor in spam filtering. Email providers (Gmail, Outlook, Yahoo) maintain reputation scores for every sending domain and IP address.

• Domain age and history: Brand-new domains have low trust. Aged domains with clean sending history score better.
• IP reputation: Established IPs (like Google's) score high. New custom SMTP IPs score zero until built up.
• Authentication pass rates: SPF, DKIM, DMARC alignment. Failed authentication signals possible spoofing.
• Sending volume patterns: Sudden volume spikes look spam-like. Consistent gradual ramping looks legitimate.
• Bounce rate: Bouncing emails signals low list quality (likely list-buyer behavior).
• Spam complaint rate: Recipients marking your emails as spam destroys reputation. Above 0.3% is severe damage.

Category 2: Content Analysis Signals

Spam filters analyze message content for spam-like patterns:

• Spam trigger words: "Free," "guaranteed," "limited time," excessive use of exclamation marks. Filters use ML to detect spam phrasing patterns.
• Link density: Emails with multiple links (especially shortened URLs or unfamiliar domains) score higher for spam.
• HTML vs plain text: Heavy HTML emails (especially with marketing templates) often go to Promotions. Plain text feels personal.
• Attachments: .exe, .zip, suspicious attachments trigger spam.
• Image-to-text ratio: Image-heavy emails (especially with little text) signal image-spam.
• Tracking pixels: Some spam filters detect tracking pixels and treat them as marketing.

Category 3: Recipient Engagement Signals

How recipients interact with your emails informs future placement:

• Reply rate: Recipients who reply to your emails signal "this is a real conversation." Boosts reputation.
• Mark as important: Recipients flagging messages as important signal "I want this."
• Move to inbox from spam: Recipients rescuing emails from spam strongly signals "not spam."
• Spam complaints: Recipients marking emails as spam directly damages reputation.
• Delete without reading: High delete rates signal "not interesting" — moderate negative signal.
• Unsubscribe clicks: Mild negative signal but legitimate behavior.

Gmail-Specific Spam Filtering

Gmail uses multiple filtering layers:

• Pre-delivery filters: Authentication checks, IP reputation, sender reputation
• Categorization: Inbox primary, Promotions, Social, Updates, Forums
• User-level personalization: Each Gmail user's filter learns from their personal behavior

Cold emails ideally land in Primary Inbox. Promotions tab is functionally similar to spam for cold email purposes — recipients rarely check Promotions for cold pitches.

Outlook-Specific Spam Filtering

Microsoft uses Exchange Online Protection (EOP) and Microsoft Defender:

• Connection filtering: IP reputation, sender authentication
• Anti-malware scanning: Attachment and link analysis
• Anti-spam analysis: Content patterns, sender history
• Anti-phishing: Sender impersonation detection
• SafeLinks: Wraps URLs for click-time scanning

How Cold Email Operators Stay Out of Spam

1. Use real Google Workspace or Outlook 365: Inherit established IP reputation
2. Configure SPF, DKIM, DMARC correctly: Pass authentication checks
3. Pre-warm inboxes: Build reputation gradually before cold sending
4. Send 15-20/day per inbox: Stay under volume thresholds
5. Plain text emails: Avoid Promotions tab classification
6. No links in first email: Reduce spam content score
7. Tight ICP targeting: Lower spam complaint rates
8. Verify lists: Keep bounce rate under 2%
9. Active follow-ups: Get replies that build reputation
10. Monitor Postmaster Tools: Catch reputation drops early

Spam Filter Testing Tools

• GlockApps: Sends test emails to seed addresses across major providers, reports placement
• Mail-Tester: Free spam score analyzer
• Google Postmaster Tools: Real Gmail data on your domain
• SenderScore: Reputation score for IPs and domains