Outbound • Deliverability • Email Infrastructure
Cold emails going to spam? It’s rarely “just the copy.”
When outreach suddenly drops into Spam (or replies fall off a cliff), most teams do the same thing: rewrite subject lines, rotate templates, and crank volume to “force it through.”
That’s how you turn a small reputation wobble into a deliverability incident. The fix is boring on purpose: identity, behavior, and audience quality — in that order.
Also searched as
Skip to diagnosis →No gimmicks, no “guaranteed inbox.” Just a disciplined system that prevents spikes, enforces pacing, and protects your brand domain while you scale outbound responsibly.
On this page
Diagnose what’s actually happening (before you “fix” it)
“Going to spam” is a symptom, not a diagnosis. Different failure modes require different fixes — and guessing is how teams accidentally train providers to distrust them.
Scenario A: Spam placement
Messages land in Spam for many recipients. Replies drop sharply. Often triggered by bounces, complaints, or erratic send patterns.
Fix focus: behavior + list quality + identity alignment.
Scenario B: Promotions / Other tab
You’re not “blocked,” but you look like marketing: heavy link patterns, template sameness, tracking artifacts, broad targeting.
Fix focus: simplify content + tighten audience.
Scenario C: Throttling / deferred delivery
Providers accept messages but slow you down. You see delays, queued sends, or inconsistent delivery timing.
Fix focus: pacing, caps, and consistent windows.
The mental model
Deliverability is a reputation system. Providers don’t “read your intent” — they observe your outcomes. They watch a few buckets of signals:
Identity
SPF/DKIM/DMARC alignment, domain consistency, reply behavior.
Behavior
Volume ramps, spikes, send windows, stop-on-reply discipline.
Audience quality
Bounces, complaints, “delete-without-reading” patterns.
Content patterns
Template sameness, heavy linking, tracking, spammy phrasing.
Why Google Workspace and Microsoft 365 flag outbound patterns
Providers don’t hate cold email. They hate unpredictable senders who generate negative recipient signals. A “new-ish” inbox blasting a broad list looks the same as actual abuse — even if your intentions are clean.
What teams misdiagnose
- “We need better copy.” Copy helps — but reputation and list quality usually dominate.
- “Let’s rotate domains weekly.” Random rotation can look evasive and prevents stability.
- “Blast once a week.” Spikes are a classic risk signal. Consistency beats bursts.
- “Track everything.” Marketing-style links/pixels can add friction to cold deliverability.
What “good” looks like in practice
- Steady daily sending inside a consistent window (not random time-of-day chaos).
- Slow ramps with hard daily caps per inbox (especially on newer mailboxes).
- Low bounces + strict stop-on-reply to avoid irritation signals.
- Simple emails that read like a person wrote them — not a campaign engine.
A practical warning
If your system allows accidental spikes, duplicate sends, or “send outside the window,” you don’t have an outbound strategy — you have a reputation roulette wheel.
Deliverability improves when outbound behaves like a controlled system: deterministic pacing, state-aware sending, and immediate brakes when signals turn negative.
Provider reality
What the big inboxes reward
- Google: stable volume + clean identity; hates spikes and “campaign feel.”
- Microsoft: strict throttling when patterns look automated or lists are noisy.
- Yahoo/AOL: punishes sloppy authentication + complaint-y audiences fast.
Translation: “more clever” rarely helps. More predictable almost always does.
Fix plan (no heroics, no “send more to recover”)
The fastest recoveries come from removing chaos, lowering risk, and letting trust rebuild through consistency. Here’s the sequence that keeps you from making it worse.
Phase 1
Stabilize (today → 48 hours)
- Reduce volume immediately; eliminate spikes and long gaps.
- Audit bounces; remove risky segments and validate your lists.
- Confirm SPF + DKIM + DMARC alignment for the sending domain.
- Stop follow-ups to anyone who replied (even negative replies).
Phase 2
Rebuild trust (week 1 → week 3)
- Ramp gradually with firm caps per mailbox and a fixed send window.
- Tighten targeting until signals stabilize (fewer sends, better recipients).
- Simplify templates: fewer links, fewer “marketing” artifacts, more human structure.
- Prefer predictability over cleverness. Providers reward boring consistency.
Where LeadBadger fits
LeadBadger is built to make deliverability a behavior, not a hope: enforce send windows, pace mailboxes, stop on reply, and prevent “distributed system” mistakes like duplicate sends.
Enforced defaults
- Fixed send windows (no “oops” sends)
- Mailbox pacing + hard caps
- Stop-on-reply + dedupe
Deliverability checklist (the stuff that actually moves the needle)
Use this as your “don’t panic” guide. The goal isn’t perfection — it’s removing obvious risk signals, then staying consistent long enough for trust to compound.
Identity & setup
- SPF present and correct for your sending source.
- DKIM enabled and passing (not “sometimes”).
- DMARC set with alignment (start at monitor if you’re new).
- Reply handling is real and consistent (no dead inboxes).
Behavior & pacing
- Fixed send window (same hours, weekdays, predictable cadence).
- Gradual ramp per mailbox with hard daily caps.
- Stop-on-reply enforcement (no “oops we followed up anyway”).
- No duplicate sends from multiple workers/tools/manual retries.
Audience quality
- Validate emails; remove risky segments if bounce rates climb.
- Tight ICP targeting: fewer sends, better fit, higher reply quality.
- Respect negative replies; don’t push follow-ups into irritation signals.
Content patterns (keep it human)
- Minimize links early; one link max is a solid cold default.
- Avoid heavy tracking artifacts that make you look like a campaign engine.
- Lower template sameness; vary structure naturally, not gimmicks.
- Write like a person with a point, not a sequence with a quota.
If you only do one thing
Remove spikes. Spikes are the easiest way to train providers that you’re unpredictable. Stability buys you time; time lets reputation recover.
Rule of thumb
If your outbound can’t enforce pacing, send windows, and stop-on-reply automatically, your “deliverability strategy” is just you hoping nothing breaks this week.
Red flags
Patterns that correlate with spam placement
- “Nothing for days → huge send day” cadence.
- Rising bounces (especially role accounts + risky domains).
- Follow-ups continuing after replies (even negative replies).
- Templates with lots of links + tracking while reputation is fragile.
Safe defaults
A boring ramp example
You’re not trying to “win today.” You’re trying to become a sender providers can predict.
Hard rule: no spikes. No “catch up” days.
FAQ
Short, practical answers. If you want a deeper diagnosis, run a controlled test — don’t guess in production.
Is it normal for cold email to hit Promotions?
Yes. Promotions isn’t automatically “bad.” If replies are healthy and spam placement is low, don’t over-optimize for tabs at the expense of stability.
Should we pause everything if spam placement rises?
Reduce volume and remove risky segments immediately. Full pauses can help, but fixing bounces and spikes usually matters more than dramatic stops/starts.
What’s the biggest deliverability mistake?
Spiky sending: nothing for days, then a blast. It’s a reputation red flag and a common “systems” failure when tooling isn’t enforcing pacing.
Do separate domains really help?
They can reduce brand risk. The real win is discipline: slow ramp, hygiene, and consistent sending. A separate domain doesn’t fix chaos.
Can we scale fast without getting burned?
You can scale, but you need controls: caps per mailbox, state-aware sending, and brakes on negative signals. Most “burns” are uncontrolled growth, not bad luck.
Do links and tracking hurt deliverability?
They can. For cold outreach, fewer links and fewer marketing artifacts tend to be safer. Keep emails simple until trust stabilizes.
Next steps: turn deliverability into a system
If you want this to stay fixed, don’t treat deliverability like a one-time cleanup. Treat it like reliability engineering: predictable behavior, enforced constraints, and fast response to signals.
1) Stabilize
Remove spikes, lower volume, fix bounces.
2) Rebuild
Ramp slowly, keep windows consistent.
3) Scale
Add volume only when signals stay healthy.
Want help implementing it?
See the platform live, or talk through your stack and constraints. We’ll keep it practical and system-first.
You don’t need more templates. You need fewer failure modes.