Domain Warming Is Not IP Warming
Most Amazon SES senders treat domain warming and IP warming as the same exercise. They are not. IP warming builds trust in a sending infrastructure address. Domain warming builds trust in the identity that appears in the From header and the MAIL FROM envelope. Mailbox providers track both signals independently, and a clean, well-warmed IP does nothing for a domain that has never established a positive sending history. If you switch to a dedicated IP pool without warming the sending domain simultaneously, or if you launch a brand-new domain and expect shared SES infrastructure to carry its reputation, messages will be routed to spam or rejected regardless of how carefully you manage your IP.
This distinction matters more in 2026 than it ever has. Microsoft enforced strict domain-level authentication requirements for high-volume senders from 5 May 2025, permanently rejecting non-compliant traffic with a 550 5.7.515 error rather than routing it to junk. Google retired the familiar High, Medium, Low and Bad domain reputation tiers from Postmaster Tools v1 on 30 September 2025, replacing them with a compliance-focused view in v2 that surfaces spam rate, authentication health, delivery errors and feedback loop data. Yahoo applies equivalent domain-level scoring. The industry has converged on a clear position: domain identity is the primary trust signal, and warming it is a discipline in its own right.
How Mailbox Providers Score Domain Reputation
Gmail scores your sending domain based on a rolling view of spam complaint rate, authentication compliance, delivery errors and engagement patterns. Until 30 September 2025 this was visible as a simple four-tier label in Postmaster Tools v1. After v1 retired, that label was removed and replaced by the v2 compliance dashboard, which surfaces spam rate, SPF, DKIM and DMARC compliance status, encryption coverage and delivery errors. The underlying scoring still exists and still drives inbox placement, but senders now need to read several signals together rather than relying on a single reputation indicator. Keep your Gmail spam rate well below 0.10 percent during warm-up: Google's own guidance places the acceptable ceiling at 0.10 percent and treats anything approaching 0.30 percent as cause for filtering.
Microsoft's enforcement is blunter. From 5 May 2025, domains sending more than 5,000 messages per day to Outlook.com, Hotmail and Live.com addresses that fail SPF, DKIM or DMARC alignment receive a permanent 550 5.7.515 rejection. This is not a temporary deferral. The message is dropped, and resending will not help until the underlying authentication is corrected. Microsoft made a deliberate policy decision to move from an initial junk-routing period to outright rejection, citing a desire to remove ambiguity for both senders and recipients. For a warming domain, the practical consequence is that authentication must be airtight before the first message is sent, not retrofitted after complaints arrive.
Yahoo applies domain-based reputation scoring through its own feedback loop and complaint-rate measurement. Like Gmail, Yahoo weighs engagement signals, complaint rate and authentication alignment. Register for Yahoo's Complaint Feedback Loop before you begin warming so that complaint data flows back to your SES suppression logic from day one.
Pre-Warm Checklist
No ramp schedule will save a domain that is not properly configured from the outset. Work through the following before sending a single production message.
Authentication records
Verify your sending domain in the SES console and enable Easy DKIM with RSA-2048 signing. SES generates three CNAME records; add all three to your DNS and confirm verified status before proceeding. Publish an SPF record that includes amazonses.com as an authorised sender. Configure a custom MAIL FROM domain, such as mail.yourdomain.com, so that the envelope sender aligns with your From domain. Without a custom MAIL FROM, SPF will pass on amazonses.com rather than your own domain, which weakens DMARC alignment and signals to mailbox providers that you are relying on shared infrastructure.
DMARC at p=none with reporting
Publish a DMARC record at p=none with an rua address pointing to an inbox you will actively monitor. The none policy does not protect your domain from spoofing, but it enables aggregate reporting from Gmail, Microsoft and Yahoo, giving you daily visibility into which sources are passing or failing authentication during warm-up. Do not start at p=reject or p=quarantine: if a forwarding path or legitimate third-party sender is misconfigured, you will not discover the problem until mail starts disappearing. Move to a stricter policy only after reviewing several weeks of aggregate reports and confirming clean alignment across all sources.
List hygiene before day one
Validate your list with an SMTP verification tool before any warm-up send. Imported lists, particularly those sourced from CRM exports, event registrations or older acquisition programmes, routinely contain 10 to 25 percent invalid addresses. A single high-bounce send during warm-up is enough to stall progress before it has begun. Remove role accounts, hard-bounce addresses from previous providers and any address that has not engaged in the past 180 days. The warm-up list should contain only your most reliably engaged subscribers.
Segmenting Your Warm-Up List in SES
Domain reputation is built on positive engagement signals. Mailbox providers observe whether recipients open, click, reply or move messages out of spam. They also observe whether messages generate complaints, hard bounces or immediate deletions. Starting your warm-up with your least engaged contacts produces the worst possible signal pattern at exactly the moment when the domain has no positive history to offset negative signals.
Segment your list before you begin. In SES, use configuration sets and message tags to separate your most engaged subscribers from the rest of your list. Define engagement tiers based on opens or clicks in the last 90 days, then 180 days, then beyond. Send only to the 90-day engaged tier during weeks one and two. Use SES event destinations to route bounce, complaint and delivery events to a data store or monitoring tool so that you can observe per-segment behaviour in real time. This separation also ensures that a problem with one segment does not contaminate the metrics of another.
Week-by-Week Ramp Schedule
The volumes below assume a target production level of several hundred thousand messages per month. Scale the absolute figures up or down proportionally for your programme, but do not compress the timeline. Domain reputation builds on consistency and positive engagement over time, not on raw volume alone.
Days 1 to 3: 250 to 500 messages per day
Send to your highest-engaged segment only. Monitor bounce rate and complaint rate after each send. Go/no-go criteria: bounce rate below 2 percent, complaint rate below 0.05 percent, no 550-class errors from any mailbox provider.
Days 4 to 7: 500 to 1,000 messages per day
If days one to three passed cleanly, double the volume. Continue sending only to 90-day engaged subscribers. Watch for 421 temporary deferral codes from Outlook and Yahoo, which indicate that the domain is being throttled pending further reputation data. A moderate number of deferrals is normal at this stage; SES will retry within the delivery window. A persistent 421 pattern across an entire send is a signal to hold volume flat rather than increase it.
Week 2: 1,000 to 5,000 messages per day
Gradually expand into your 180-day engaged tier. Increase daily volume by no more than double every two to three days. Positive engagement from the first week should begin establishing a baseline signal with Gmail and Outlook. Check the Gmail Postmaster Tools v2 spam rate dashboard: if your spam rate is visible and below 0.08 percent, proceed. If it is not yet visible, the domain has not reached sufficient Gmail volume for data to appear, which is normal at this stage.
Weeks 3 to 4: 5,000 to 25,000 messages per day
You are now sending enough volume to Microsoft that its domain-level authentication checks will be actively applied. Ensure all authentication records are passing correctly. If you see a spike in 550 5.7.515 rejections, stop, fix the authentication issue and restart from the previous volume level. Do not increase volume while rejections are occurring. Introduce transactional messages alongside marketing sends if your programme includes them: transactional mail typically generates stronger engagement signals and helps build reputation faster.
Weeks 5 to 8: 25,000 messages per day to full production volume
Continue doubling volume every three to five days while monitoring all metrics. By week six you should be seeing stable, low complaint rates and minimal deferral patterns. Begin introducing less-engaged segments cautiously, watching for complaint rate movement after each expansion. The ramp is complete when you have reached full production volume and maintained a bounce rate below 5 percent and a complaint rate below 0.10 percent for at least two consecutive weeks.
Signals That Your Warm-Up Is Stalling
Bounce rate spikes
Hard bounces during a warm-up usually indicate that your list contains invalid addresses not caught during pre-send validation. AWS recommends keeping your bounce rate below 5 percent, with account review triggered above that threshold and sending paused above 10 percent. During a warm-up, treat anything above 3 percent as a stop signal. Suppress hard-bounced addresses immediately and investigate whether the segment being sent to meets your engagement criteria before resuming.
Complaint rate creep
AWS publishes two complaint rate thresholds: 0.10 percent as the recommended ceiling and 0.50 percent as the level at which sending may be paused. During warm-up, these thresholds are more dangerous than at full production volume because low absolute send counts mean a small number of complaints produces a large percentage figure. Keep complaint rate below 0.05 percent during weeks one and two. A creeping complaint rate at this stage almost always indicates that the warm-up segment is not as engaged as assumed, or that message content, subject line or sender identity is unfamiliar to recipients.
Deferral error codes
A 421 error is a temporary deferral: the receiving server is asking SES to try again later. This is common early in the warm-up and SES handles retries automatically. A 550 error is permanent. If you are seeing 550 errors from Gmail, the domain may be on a blocklist or your content is being flagged. If you are seeing 550 5.7.515 errors from Microsoft, your authentication records are not passing. Treat any 550 pattern as a hard stop requiring diagnosis before any further volume increases.
Open rate drops
A significant drop in open rate within a single mailbox provider's ecosystem, while other providers remain stable, usually indicates that messages have been moved to the spam folder at that provider. During a warm-up this is a sign that the domain has accumulated enough negative signal to trigger filtering. Pause volume to that provider's address space, investigate complaint rate and authentication status, and restart at a lower volume after at least a week of no sending.
Monitoring During the Warm-Up
Set up SES event destinations through configuration sets to stream bounce, complaint and delivery events to Amazon CloudWatch, an SNS topic or a data store before your first send. CloudWatch allows you to set alarms on bounce rate and complaint rate thresholds so that you receive notification before reaching AWS review levels rather than after. This is not optional during a warm-up: volume changes are frequent enough that a single problem campaign can tip metrics into dangerous territory within hours.
Enable the SES account-level suppression list so that hard-bounced addresses and complaint sources are automatically suppressed without requiring manual intervention. If you are using SES's Virtual Deliverability Manager, inbox placement rate data by sending domain becomes available, giving you a direct signal of warm-up progress beyond what bounce and complaint rates alone can show.
Register your domain with Google Postmaster Tools v2 and monitor the spam rate and compliance dashboards daily during weeks one through four. Check Microsoft's Smart Network Data Services for Outlook-side data on complaint rates and filtering decisions. For Yahoo, monitor your Complaint Feedback Loop data. Together these three sources give you a mailbox-provider-specific view that SES metrics alone cannot provide.
Recovering from a Warm-Up Stall
If bounce rate, complaint rate or provider-side filtering signals indicate that the warm-up has stalled, pause all sending immediately. Do not attempt to push through the problem by increasing volume or changing content without first diagnosing the cause. Common causes include a segment that was less engaged than assumed, authentication misconfiguration not caught during pre-send checks, list hygiene issues that only became apparent at higher volumes, or content too similar to known spam patterns.
After pausing for at least five to seven days, clean the list again, remove any addresses from the problematic segment, verify that all authentication records are passing across SPF, DKIM and DMARC, and restart at the volume level from two stages before the stall. Do not restart at the volume where the stall occurred. A short pause does not erase all accumulated reputation: mailbox providers do not reset domain history during a brief interruption, but they will not tolerate a resumption of negative signals.
Common Mistakes to Avoid
Sending to unengaged lists early in the ramp is the single most frequent cause of warm-up failure. Senders who believe that high-volume sending to uninterested recipients will establish reputation faster are operating on the opposite of correct logic. Mailbox providers weigh complaint rate and engagement quality, not raw volume.
Skipping the custom MAIL FROM configuration undermines DMARC alignment from day one. Without it, SPF aligns to amazonses.com rather than your domain, and some mailbox providers treat the mismatch as a trust signal against the domain being warmed.
Warming a subdomain such as mail.yourdomain.com while leaving the root domain's DMARC record unpublished causes aggregate report data to be missing or misattributed. Publish DMARC at the root domain with rua reporting before warming any subdomain beneath it.
Ignoring soft bounces during the ramp stores up hard bounces for later. An address that soft-bounces repeatedly across multiple sends is effectively a hard bounce for reputation purposes. Remove persistent soft-bounce addresses from your sending list during the warm-up.
Keeping Reputation Healthy After Warm-Up
Completing the warm-up schedule is not the end of domain reputation management. Mailbox providers expect consistent sending cadences and will begin to discount reputation built during an active warm-up if volume drops sharply afterwards. If you anticipate a significant gap in sending, reduce volume gradually before the gap rather than stopping abruptly.
Continue monitoring bounce rate, complaint rate and provider-side signals with the same alerting infrastructure set up during warm-up. The disciplines that protect a warming domain are the same ones that protect a mature domain: clean lists, engaged segments, proper authentication and automated suppression of bounces and complaints. Set alerts to fire before you reach SES review thresholds, not at them. Domain reputation can degrade faster than it was built, and continuous monitoring is what gives you the time to act before a recoverable problem becomes a serious deliverability hole.