Most teams trying to improve email deliverability start with the wrong layer. They rewrite subject lines, swap out merge tags, change the call to action. All content-level fixes for an infrastructure problem.
Deliverability is an infrastructure problem before it is a content problem. Receiving servers decide inbox placement based on signals they verify before the email is even opened: where it came from, what domain reputation the sender has built, whether the message is authenticated. Content matters, but it matters last.
Here is the actual order of operations to fix and then improve email deliverability, top to bottom of the stack.
Step 1: Authenticate Every Sending Domain
Authentication is the single highest-leverage fix and the first thing to check. Without SPF, DKIM, and DMARC configured correctly, modern receiving servers (Gmail, Microsoft 365, Yahoo) will treat the email as suspicious by default.
- SPF tells receiving servers which IPs and services are allowed to send on behalf of your domain. Most SPF problems come from too many "include" statements (the 10-lookup limit) or unflattened records that drift over time.
- DKIM signs each outgoing email with a cryptographic signature that proves the email was not modified in transit. If your DKIM signature fails, mail servers know the message is either forged or relayed by something untrusted.
- DMARC tells receiving servers what to do when SPF or DKIM fails (none, quarantine, reject) and where to send aggregate reports. A DMARC policy of "none" still produces reports that show exactly who is sending mail claiming your domain.
The practical check: send a test email from each sending domain to check-auth@verifier.port25.com or use Mail-Tester. If SPF, DKIM, or DMARC fails on any sending domain, deliverability is broken at the foundation and no content fix will compensate. Authentication setup is week one work, not week six.
Step 2: Audit Domain and IP Reputation
Once authentication is clean, the next signal receiving servers check is reputation. Both domain reputation and IP reputation are tracked independently, and both have to be healthy.
Tools to check:
- Google Postmaster Tools, domain reputation, IP reputation, spam rate, feedback loop data, and delivery errors for any domain sending more than ~100/day to Gmail
- Microsoft SNDS. IP-level reputation data for any IP sending to Outlook/Hotmail accounts
- MXToolbox / MultiRBL, checks against 60+ public blocklists; if your IP or domain is on Spamhaus, SORBS, or Barracuda, that is the entire problem
If reputation is bad, the playbook depends on how bad. A single blocklist with a clear delisting process: submit the request, fix what flagged you, move on. Multiple blocklists or a sustained low Postmaster reputation: the domain is salvageable but it will take weeks of clean sending. A domain showing red across multiple sources for months: retire it.
Step 3: Match Infrastructure to Volume and Email Type
This is where most deliverability efforts go sideways. A sending setup that works at 100 emails/day will fail at 1,000/day, and a setup that works for transactional email will fail for cold outreach.
The basic rules:
- Cold outreach, multiple secondary domains (never the main business domain), real provider inboxes (Google Workspace or Microsoft 365), warmed up properly before any campaign. SMTP relays are the wrong tool here regardless of price.
- Transactional email, dedicated transactional ESP (Postmark, Mailgun, AWS SES) on a subdomain like
mail.company.com, separate from everything else - Marketing/newsletter, dedicated marketing ESP (Mailchimp, Klaviyo, Customer.io) on a separate subdomain like
news.company.com - Internal and warm outreach. Google Workspace or Microsoft 365 on the main domain
Mixing these on a single domain is the most common cause of "deliverability got worse and we don't know why." A cold campaign that generates spam complaints damages the reputation of every other email type sharing that domain.
Step 4: Warm Up Every New Sending Inbox
A new inbox or a new sending domain has zero history with receiving servers. No history means no trust, which means new senders default to spam folders or get rate-limited until trust is built.
Proper email warmup is the deliberate process of building that history: gradually increasing send volume over 2-4 weeks while generating positive engagement signals (opens, replies, marking as not-spam, marking as important). Warmup tools automate this with networks of inboxes that exchange mail with each other.
What kills warmup faster than anything: starting real campaigns mid-warmup. The volume jump from 30 warmup emails/day to 200 cold campaign emails/day is the exact pattern receiving servers flag as spam behavior. Finish warmup before any real send.
Step 5: Tighten List Hygiene
Bad lists are the most common cause of deliverability degradation on previously healthy domains. Three patterns matter:
- Bounce rates above 2-3% tell receiving servers your list is dirty. This includes both soft and hard bounces, but hard bounces are the worse signal because they indicate addresses that never existed.
- Spam complaint rates above 0.1% are the threshold Gmail uses to start filtering more aggressively. One complaint per 1,000 sends. Keeping complaints below this requires verified opt-in for marketing and tight ICP targeting for cold outreach.
- Unverified addresses, every list, every month, run through an email verification service (NeverBounce, ZeroBounce, Bouncer). Verification catches catch-all and dead addresses before they show up as bounces in your metrics.
A list pulled six months ago and emailed today will have a 5-10% decay rate just from natural turnover. Re-verify before any send to a list older than 90 days.
Step 6: Match Send Behavior to Inbox Norms
Receiving servers profile sender behavior. Patterns that look like a real human or a real business get rewarded; patterns that look automated and aggressive get filtered.
- Per-inbox volume caps: 30-50 cold emails per day per inbox is a safe ceiling for established Workspace/M365 inboxes. Pushing past this signals automation and damages reputation fast.
- Send time randomization: sequencers that send 50 emails between 9:00:00 and 9:00:30 get filtered. Real humans space sends out across hours.
- Reply tracking: when recipients reply, the reply has to actually arrive in the sending inbox. If replies bounce or get auto-archived, receiving servers learn the conversation is one-sided.
- Engagement signals: sustained low open rates (under 30% on a clean cold list) signal that mail is hitting spam. The fix is upstream, domain reputation, warmup, list quality, not new subject lines.
Step 7: Then, and Only Then, Optimize Content
Content does matter, but it matters once everything underneath is solid. The content rules are simpler than most "spam trigger word" lists make them sound:
- Plain text or near-plain text outperforms image-heavy HTML for cold and follow-up email
- One clear call to action per email, not three
- Short paragraphs, conversational tone, no marketing copy in the first paragraph
- Avoid spam trigger phrases ("act now," "limited time," "100% free"), but the modern filters care less about specific words and more about engagement patterns
- Personalization at the open and a specific reason for the email, not a merge tag stuffed into a generic template
Content optimization at the top of a healthy stack moves reply rates from 3% to 6%. Content optimization at the top of a broken stack moves nothing because the email never arrives.
Continuous Monitoring (Where Most Teams Stop)
Deliverability is not a project that finishes. The signals that determine inbox placement shift constantly: new policy updates from Gmail and Microsoft, blocklists adding and removing domains, changes in your own list quality, drift in authentication records.
What to monitor weekly:
- Google Postmaster reputation across all sending domains
- Bounce rate and spam complaint rate per domain and per campaign
- Blocklist status (MultiRBL or equivalent)
- Inbox placement test results (Mail-Tester, GlockApps) for representative campaigns
- Reply rates and engagement, segmented by domain
A domain showing reputation drift over two consecutive weeks needs intervention before it becomes a blocked domain. The cost of catching it early is hours; the cost of recovering a flagged domain is weeks or a full domain replacement.
The Common Failure Mode
Most teams improve email deliverability the wrong way around: they start with content tweaks, escalate to ESP changes, and only after months of degraded inbox placement realize that the actual problem was authentication, infrastructure, or warmup. By that point, the domain reputation is damaged enough that fixing the foundation alone takes weeks.
The right order is exactly what is laid out above: authentication, then reputation, then infrastructure match, then warmup, then list, then send behavior, then content. Skip a layer at the bottom and nothing above it can compensate.
At ScaledMail, the infrastructure layer is the entire job. Every cold outreach client gets dedicated sending domains separate from their main business domain, real Google Workspace and Microsoft 365 inboxes (not SMTP relays), authentication configured correctly on every domain, IP rotation, and continuous reputation monitoring. Warmup runs inside the customer's sequencer (Smartlead, Instantly, EmailBison, PlusVibe), where the engagement and send-pattern data lives. The cold email infrastructure we provision is built so the deliverability stack is solid before the first email goes out, and stays solid as volume scales. Book a call or see the setup if you want the foundation built right the first time.
