Skip to content

sendx/email-skills / skills/email-deliverability-expert-skill

email-deliverability

Verified

Diagnose, audit, and fix email deliverability problems using a proven, systematic methodology. Use this skill whenever someone asks for a deliverability audit, says "audit my email", "check my deliverability", "why are we hitting spam", or "why are my emails going to spam" — or mentions inbox placement problems, emails going to spam, low open rates, IP/domain reputation damage, blocklisting, warming up IPs or domains, shared pool contamination, bounce rate spikes, mailbox provider blocks, sender reputation, or email authentication failures. Also trigger when someone is setting up email sending for the first time and wants to get it right, wants a review of their email setup, or asks about ESP infrastructure health, sender compliance, list hygiene, traffic shaping, or reputation recovery planning. Even if they just say "our emails aren't working" or "open rates dropped" — this is the skill to use.

references/provider-specific-tactics.md

View on GitHub
# Provider-Specific Tactics

How to diagnose and fix deliverability issues at each major mailbox provider and spam filter, based on SendPost's experience managing large-scale sending operations.

---

## Gmail

### How Gmail Thinks
Gmail uses correlation-based filtering — it connects multiple data points to identify coordinated operations. It's not just looking at one signal; it's looking for patterns.

**Gmail's strategy:** "Make it unprofitable, not impossible." They don't block everything outright — they make spamming so costly in terms of blocked mail and wasted resources that legitimate senders optimize while bad actors give up.

### Correlation Points Gmail Watches
- Physical business addresses shared across domains
- Web hosting IPs shared across "different" brands
- Similar domain naming patterns
- Sender identity mismatches (display name vs. email address)
- Similar or identical content sent from different domains
- Lack of consistent branding across sends

### Gmail-Specific Warmup
- Start at ~200 emails/day for campaigns
- Never exceed 4x previous day's opens
- Domain reputation in Google Postmaster Tools is your primary signal
- Reputation levels: High → Medium → Low → Bad
- Moving from Low to High takes weeks of consistent positive signals

### Gmail-Specific Fixes
- **Content filtering:** Address all correlation points. If multiple domains share infrastructure markers, Gmail treats them as one sender.
- **Domain reputation "Bad":** Full warmup restart needed. Consider whether the domain is recoverable or if starting fresh is more practical.
- **Low engagement:** Gmail heavily weights engagement. Sending to inactive subscribers actively damages your reputation there.

---

## Microsoft (Outlook / Hotmail)

### How Microsoft Differs
Microsoft tends to be more aggressive with blocking than Gmail. Where Gmail might spam-folder your emails, Microsoft might outright reject them.

### Microsoft-Specific Warmup
- Start very low: 25 emails/day when recovering from damaged reputation
- Never exceed 4x previous day's opens
- Microsoft separates Outlook.com and Hotmail — track both independently
- Recovery is often slower at Microsoft than Gmail

### Microsoft-Specific Issues
- **Outages affect senders:** Microsoft's own outages can cause mass bounces. The ESP should detect this and pause sending: "If everybody starts getting blocked at Microsoft because it went down again, we should pause sending and not just bounce a gazillion emails off of Microsoft's infrastructure."
- **Aggressive gray-listing:** Microsoft uses gray-listing (code 400 soft bounces) extensively. Legitimate senders retry and get through; spammers don't. Understanding this behavior is critical — it inflates apparent soft bounce rates for legitimate senders.

### Microsoft-Specific Fixes
- When Microsoft blocks: 48-hour pause from that IP, then restart at minimal volume
- New IP may be preferable to rehabilitating a blocked one
- Track Hotmail and Outlook.com as separate segments during warmup

---

## Yahoo

### How Yahoo Differs
Yahoo is generally more forgiving than Gmail or Microsoft. Good news: easier to recover. Bad news: people sometimes use Yahoo success as false confidence about overall deliverability.

### Yahoo-Specific Warmup
- Start at ~50 emails/day when recovering
- Never exceed 4x previous day's opens
- Yahoo performance recovering doesn't mean Gmail performance is fine

### Yahoo-Specific Tactics
- Yahoo's engagement weighting is lighter than Gmail's
- Still important to segment by engagement, but Yahoo gives more room for recovery
- Watch for Yahoo-specific spam folder placement vs. blocks — they behave differently

---

## iCloud / Apple Mail

### The iCloud Signal
iCloud/Apple Mail open rates below 10% indicate severe inbox placement problems. Because Apple Mail Privacy Protection inflates open rates, genuinely low iCloud opens mean almost nothing is being delivered to inbox.

### iCloud-Specific Considerations
- Apple MPP (Mail Privacy Protection) pre-fetches images, which inflates open rate numbers
- If you're seeing sub-10% opens at iCloud despite MPP inflation, the actual engagement is dramatically worse
- iCloud placement is heavily reputation-dependent
- Recovery approach follows the standard warmup playbook but with extra attention to engagement quality

---

## ProofPoint

### How ProofPoint Works
ProofPoint is a corporate spam filter (not a mailbox provider). It protects enterprise email systems — so when you're sending to business addresses at companies using ProofPoint, different rules apply.

### ProofPoint-Specific Issues
- **Sending rate sensitivity:** "The rate at which they were sending for one customer into the ProofPoint spam filter was not working. It was generating huge numbers of soft bounces."
- **IP-level blocking:** ProofPoint blocks at the IP level. Getting delisted requires contacting ProofPoint directly.
- **Throttling requirement:** The ESP's MTA must throttle sending rate to ProofPoint-filtered domains. Sending too fast = automatic soft bounces.

### ProofPoint-Specific Fixes
- Get the IP delisted from ProofPoint's blocklist directly
- Implement MTA-level throttling for ProofPoint-filtered domains
- Reduce sending volume to ProofPoint destinations during recovery
- ProofPoint soft bounces are often rate-related, not reputation-related — distinguish between the two

---

## SpamHaus

### What SpamHaus Does
SpamHaus maintains blocklists that many mailbox providers and spam filters reference. A SpamHaus listing can cascade across multiple providers simultaneously.

### Automatic SpamHaus Listing Triggers
- **HELO/rDNS mismatch:** "If the HELO, the server announcement that comes through the SMTP connection, doesn't match the IP's reverse DNS — that is an automatic SpamHaus listing. If you hit enough networks with it."
- **Spam trap hits:** Sending to SpamHaus-maintained spam traps
- **High complaint rates:** Excessive spam reports from multiple sources

### SpamHaus Diagnosis
Use the free SpamHaus domain reputation checker to check:
- Negative reputation scores
- Infrastructure quality assessment
- Domain creation dates (new domains are higher risk)
- Whether multiple domains share the same IP address

### SpamHaus Delisting
- Fix the root cause first — delisting without fixing the cause results in re-listing
- Submit delisting request through SpamHaus portal
- Provide evidence of remediation steps taken
- Monitor closely after delisting — re-listing happens fast if the problem recurs

---

## Cross-Provider Strategy

### The Segment-by-Provider Approach
During any warmup or recovery, segment sending by mailbox provider. Each provider has different:
- Rate tolerances
- Engagement weighting
- Block vs. spam-folder behavior
- Recovery timelines

Real-world example of segment management during warmup:
"I literally one time had to create eight separate segments for the different mailbox providers and spam filters and then literally update those every day and say okay, well today I sent to the last one to three days of engaged people at ProofPoint, Microsoft, Gmail, Hotmail, whatever."

### Reading Provider Signals Together
- **All providers declining:** Likely a content or data quality issue
- **One provider declining, others stable:** Provider-specific throttling or reputation issue
- **Sudden drops everywhere:** Authentication failure, DNS change, or SpamHaus listing
- **Gradual decline:** Engagement erosion — probably sending to too many inactive contacts

### The 4x Rule
Across all providers: "Never send more than 4x previous day's opens."

This is a universal safety guardrail. If you sent 100 emails yesterday and got 25 opens, today's maximum is 100 emails (4x25). This prevents over-sending to unengaged audiences and naturally scales volume as engagement improves.
rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first.Stars: 194kUpdated Claude CodeCodexVerifiedInstall