When deliverability drops, everyone panics and “fixes” everything. New domains. New copy. New tool. Same results. Because nobody isolates the bottleneck.
This guide is cold email deliverability troubleshooting for people who need answers in the next 30 minutes, not a lecture on “domain reputation theory.”
Goal: identify the real failure mode fast: Data, Copy, or Domain/Sending. Then touch the minimum number of variables.
TL;DR
- Bounces spiking? It’s almost always data or infrastructure (wrong mailbox, bad verification, broken routing). Stop sending and classify bounces.
- Delivered but in spam? It’s usually domain/sending pattern or tracking/links, not “rewrite your value prop.”
- Replies dropped but placement looks fine? It’s usually copy + targeting. Diagnose with isolation tests, not vibes.
- If you send high volume to Microsoft consumer inboxes, they can reject non-compliant mail (SPF/DKIM/DMARC). Microsoft explicitly calls out enforcement starting May 5, 2025 with a 550 5.7.515 style rejection for missing authentication. Read it from Microsoft, not Twitter.
Source: Microsoft Tech Community announcement
Start with symptoms, not theories
Pick the symptom that matches what you see today. Then run the 30-minute triage.
Symptom A: Bounce rate spikes (sudden 2x+ jump)
Common causes:
- Bad list segment (fresh scrape, old export, wrong geo)
- Verification drift (you stopped verifying or changed provider)
- You switched inboxes, SMTP, or routing
- Provider throttling turns into deferrals that your tool misreports
What it usually is: Data first. Then infrastructure.
Symptom B: Spam placement rises (Gmail spam folder, Outlook Junk)
Common causes:
- Tracking pixel
- Link domains (redirects, shorteners, tracking domains)
- New sending pattern (volume jump, too many new recipients)
- Domain reputation decline
What it usually is: Domain/sending pattern and “message shape” (pixels, links). Copy matters, but later.
Symptom C: Replies drop but bounce/spam doesn’t look worse
Common causes:
- You changed ICP, offer, or personalization logic
- You hammered the same accounts (fatigue)
- Your “personalization” got fake and obvious
What it usually is: Copy + targeting.
Symptom D: Gmail blocks or temp errors show up
Common causes:
- Authentication failures
- Reputation issues
- Sudden scaling
What it usually is: Domain/auth + sending pattern.
The 30-minute triage: isolate Data vs Copy vs Domain
Set a timer. Run this in order. Each step tells you what to do next.
What you need (before you start)
- Last 7 days of campaign metrics by mailbox provider (Gmail vs Microsoft vs Yahoo)
- Bounce logs with SMTP codes/messages
- 10-20 recent prospects per segment
- 2 seed inboxes you control (one Gmail, one Outlook.com)
- Ability to send two micro-tests (tracking on vs off)
Minute 0-5: List quality audit (fast, brutal)
Objective: decide if your list is poisoning everything.
Check 50 random rows from the last 1-2 sends:
- Are you emailing role addresses (info@, sales@, support@)? Those inflate bounces and spam complaints.
- Are you emailing free consumer domains when your ICP is businesses? That changes filtering behavior fast.
- Are you emailing the same domain repeatedly (10+ people at one company)? That triggers complaints and blocks.
- Are titles and company names correct, or obviously wrong? Bad enrichment causes fake personalization, which causes spam clicks.
Quick rule: If your list looks even slightly questionable, treat deliverability as a data problem first. Bad data makes every other fix look like it worked.
If you want a deeper standard for data checks, keep it operational: run deterministic checks (duplication, role, disposable, no-MX, typo domains). “Verified” badges are marketing. The inbox doesn’t care.
Minute 5-10: Bounce taxonomy (the only part that’s not optional)
Stop looking at “bounce rate.” Classify bounces into buckets:
Hard bounces (data problems)
Examples:
5.1.1user does not exist5.1.0address rejected- “No such user”
- “Mailbox unavailable”
Action: remove immediately. If this bucket jumps, your list or verification is failing.
Authentication / policy bounces (domain problems)
Examples:
- “SPF fail”
- “DKIM missing/invalid”
- “DMARC fail”
- “Policy rejection”
Action: do not “fix with copy.” Fix DNS/auth and alignment.
Microsoft explicitly states they will reject messages from high-volume senders that do not meet authentication requirements, with enforcement starting May 5, 2025.
Source: Microsoft Tech Community
Rate limiting / deferrals (sending pattern problems)
Examples:
4.7.0temporary deferral- “Try again later”
- “Too many messages”
- Greylisting
Action: reduce per-inbox volume, spread sends, slow ramp. This is “you scaled like a maniac” in SMTP form.
Decision after 10 minutes
- If hard bounces dominate: Data bottleneck.
- If auth/policy dominates: Domain bottleneck.
- If deferrals dominate: Sending pattern bottleneck.
- If bounces are flat but spam/replies changed: keep going.
Minute 10-15: Seed testing (2 inboxes, 2 variants, 1 purpose)
Objective: determine if “message shape” triggers spam.
Send 2 emails to your seed inboxes:
- Plain-text: no links, no tracking pixel, no attachments
- Production-like: same copy but with your normal tracking and one normal link
Send from the same sending identity you use in production.
Record outcomes:
- Gmail: inbox vs spam
- Outlook.com: inbox vs junk
- Any warnings like “Be careful with this message”
Interpretation
- If plain-text lands in inbox but production-like goes to spam: your bottleneck is tracking and links, not copy.
- If both go to spam: domain reputation or sending pattern. Copy might be weak, but you have a delivery fire first.
- If both land in inbox but replies are down: copy/targeting problem.
Minute 15-20: Spam folder sampling (real recipients, not your seeds)
Seeds lie. Your real list tells the truth.
Pick 10 real prospects across providers:
- 4 Gmail / Google Workspace
- 4 Microsoft (Outlook/M365)
- 2 Yahoo/AOL if you have them
Send a single, plain follow-up that looks like a real human follow-up:
- No link
- No pixel
- One short question
- Same sender name
Then ask 2 friendly contacts at Gmail and Outlook to whitelist-check where it landed if possible.
What you’re looking for
- Provider-specific failure (Gmail hates you, Outlook is fine)
- Universal failure (everyone hates you)
Provider-specific failures scream “domain reputation with that provider” or “message shape triggers that provider.”
Minute 20-23: Tracking pixel on/off A/B (stop guessing)
Open tracking relies on a 1x1 image request. That is not “neutral.” It changes the email.
Validity’s deliverability reporting also calls out how open rate measurement relies on a single-pixel image and how changes like Apple’s Mail Privacy Protection distort opens. That’s measurement chaos, not signal.
Source: Validity, The State of Email in 2024 (PDF)
Micro-test
- Send 20 emails: 10 with tracking ON, 10 OFF
- Same list quality and provider mix
- Compare:
- Bounce rate
- Spam placement (use replies like “went to spam” if you can, plus seed outcomes)
- Reply rate (even small sample can show directional signal)
Decision
- If tracking OFF improves placement: keep it off for cold outbound. Track replies. Track meetings. Open rates are a vanity metric with extra steps.
If you want the strategic version of this, we already wrote it: Open Tracking Is Becoming a Deliverability Tax. The 2026 Fix: Reply-First Sequences.
Minute 23-26: Link domain check (one bad link can poison the whole send)
What to check
- Are you using a tracking domain that differs from your sending domain?
- Are you using shorteners?
- Are you linking to Google Drive, Notion, Calendly, or a weird redirect chain?
Isolation test
- Version A: no links
- Version B: one link to your main domain
- Version C: one link to the “problem” domain (tracking, shortener, etc.)
If B lands and C dies, you found the bottleneck.
What not to do
- Do not “add more links to look legitimate.”
- Do not hide links behind shorteners. That screams “I am definitely not spam.”
Minute 26-28: Authentication quick check (fast, not nerdy)
You are not doing a full auth audit in 2 minutes. You are checking for obvious self-inflicted wounds.
Checklist
- SPF exists and includes the actual sender (Google Workspace, Microsoft 365, SMTP provider)
- DKIM signing is enabled on the mailbox provider
- DMARC exists (even
p=noneis better than nothing) - From domain matches the domain you authenticate (alignment matters)
Microsoft’s enforcement language is explicit for high-volume senders, and the penalty is rejection.
Source: Microsoft Tech Community
Also, if you send any kind of marketing mail, Yahoo has pushed hard on making unsubscribing easy because spam clicks damage everyone. Yahoo even built a Subscription Hub around user control.
Source: Yahoo Sender Hub - Subscription Hub
Cold outbound is different from newsletters, but the filtering systems do not care about your philosophy.
Minute 28-30: Sending pattern review (the silent killer)
You are looking for one thing: did your pattern change right before the drop?
Check these
- Did you double daily volume per inbox?
- Did you add 10 new inboxes and start them at full speed?
- Did you shift from “small batches” to “blast at 9:00 AM”?
- Did you start targeting a new geography or provider mix?
- Did you move from no-links to links in first email?
Fix order
- Reduce per-inbox volume for 72 hours
- Spread sends across the day
- Cut new recipients per day
- Remove links and tracking from first touch
- Then reintroduce variables one at a time
The decision tree: what to fix first (and what not to touch)
Outcome 1: Data bottleneck
You saw: hard bounces up, 5.1.1 dominates, verification looks weak.
Fix first
- Re-verify the next send segment
- Remove role accounts, catch-alls (or treat separately), disposable domains
- Tighten ICP so you stop emailing garbage
Do not touch
- Domains, warmup, copy templates, tooling
You do not have a deliverability problem. You have a list problem.
Chronic angle: this is where automation without quality checks burns teams. Chronic bakes enrichment into the workflow so personalization uses real facts, not fantasy: Lead enrichment
Outcome 2: Domain/auth bottleneck
You saw: policy bounces, auth failures, Microsoft-style rejections, provider-specific spam even on plain text.
Fix first
- SPF/DKIM/DMARC alignment
- Remove risky sending behaviors (tracking, link redirects)
- Reduce volume and ramp slowly
Do not touch
- Your offer, CTA, or subject line
When the mailbox says “no,” your copy does not matter.
If you need a clean explanation of DMARC alignment that doesn’t melt your brain: DMARC Alignment for Cold Email: A Non-Technical Explanation
Outcome 3: Tracking/link bottleneck (message shape)
You saw: plain-text lands, tracked version goes spam.
Fix first
- Turn off open tracking on first touch
- Remove links in first touch
- No images, no buttons, no HTML templates
Do not touch
- Sending domains, inbox providers, or CRM
You don’t need a new stack. You need fewer “marketing” artifacts.
Outcome 4: Sending pattern bottleneck
You saw: deferrals, throttling, sudden drop after scaling, inconsistent placement.
Fix first
- Cut volume per inbox
- Smooth schedule
- Ramp new inboxes like adults
- Segment by provider (Gmail-heavy days vs Microsoft-heavy days)
Do not touch
- Copy, for now
You’re getting throttled. Copy does not change rate limits.
Outcome 5: Copy/targeting bottleneck
You saw: placement fine, bounces fine, replies down.
Fix first
- Tighten ICP
- Simplify first email to one clear reason and one question
- Stop fake personalization
- Add a reply-handling system so replies do not rot
Reply handling is part of “deliverability” because ignored replies train the system that your mail is unwanted. Here’s the SOP: Reply Handling SOP
Chronic angle: scoring matters here. If you keep emailing low-fit leads, you farm spam complaints. Chronic prioritizes with dual fit + intent scoring: AI lead scoring
The practitioner checklist (copy/paste)
30-minute cold email deliverability troubleshooting checklist
- Segment symptoms: bounces vs spam vs reply drop
- Audit 50 leads: roles, duplicates, free domains, wrong titles
- Classify bounces: hard vs policy vs deferrals
- Seed test: plain-text vs production-like
- Spam sample: 10 real sends across providers
- Pixel A/B: 10 on, 10 off
- Link test: no link vs main domain vs tracking/shortener
- Auth check: SPF, DKIM, DMARC exist and align
- Pattern review: volume jump, batch timing, new inbox ramp
- Change one variable: re-test before “fixing everything”
“What not to touch” list (because panic kills domains)
When deliverability tanks, do not:
- Switch domains and keep the same behavior
- Add more HTML, images, or buttons to “look legit”
- Increase volume to “power through”
- Change five things at once
- Buy a new tool and call it a strategy
If you want to consolidate tools without breaking outbound, read: The 2026 B2B Outbound Stack: What to Consolidate First
Where Chronic fits (one paragraph, no fluff)
Most teams run deliverability triage with spreadsheets, five tools, and a prayer. Chronic runs outbound end-to-end till the meeting is booked. It keeps the system tight: ICP, enrichment, scoring, sequencing, pipeline. One workflow. Less tool spaghetti. More control.
- Build tight targeting with ICP Builder
- Keep personalization factual with Lead enrichment
- Write clean, plain-text-first outreach with AI email writer
- Prioritize high-fit leads with AI lead scoring
- Track outcomes in one place with Sales pipeline
And if you’re comparing stacks:
- Chronic vs Apollo
- Chronic vs HubSpot
- Chronic vs Salesforce
FAQ
What’s the fastest way to tell if deliverability is a data problem or a domain problem?
Classify bounces by SMTP code/message. A spike in 5.1.1 and “user unknown” is list quality. Policy/auth failures and widespread spam placement, even on plain-text seeds, points to domain/auth and sending behavior.
Should I turn off open tracking for cold outbound?
If your seed test shows tracked emails landing worse than plain text, yes. Also, open rates are increasingly unreliable because measurement depends on a tracking pixel and image loading behavior, which has been distorted by privacy changes like Apple Mail Privacy Protection. Source: Validity State of Email 2024 (PDF)
What bounce rate is “dangerous” for cold email?
If hard bounces climb above low single digits, you are feeding mailbox providers a signal that your list is low quality. For triage, the exact benchmark matters less than the direction and the bounce taxonomy. A jump from 1% to 4% is a bigger story than “what’s the ideal.”
Why does Outlook feel stricter than Gmail?
Microsoft’s own benchmark data shows Microsoft inbox placement is among the toughest. In Validity’s 2025 benchmark report, Microsoft is called out as the toughest mailbox provider with inbox placement around the mid-70s range. Source: Validity 2025 Email Deliverability Benchmark (PDF)
What’s one deliverability fix that usually backfires?
Changing multiple variables at once. New domain plus new copy plus new tracking plus new sending schedule gives you one outcome and zero diagnosis. You do not fix deliverability with chaos.
If Gmail is spamming me but Outlook is fine, what do I do first?
Run the isolation test: plain-text no tracking no links. If that lands in Gmail inbox, your issue is message shape (pixel, links, redirects). If even plain-text goes to spam, slow down sending, tighten list, and treat it as a Gmail-specific reputation problem. Then re-test before scaling.
Run the Triage. Then Ship the One Fix.
Set the timer. Run the checklist. Pick the bottleneck. Change one variable. Re-test.
Deliverability is not mysterious. It’s just unforgiving.