How to Warm Up a New Sending Domain Without Getting Filtered

You bought a fresh domain on Monday, wired up the DNS on Tuesday, and sent 40,000 launch emails on Wednesday. Gmail put almost all of them in spam and Outlook rejected a chunk outright. Nothing was wrong with the list or the copy. The problem is that your domain had no history, and a domain with no history that suddenly sends 40,000 messages looks exactly like a domain someone registered that morning to run a spam campaign.

Warming up is the process of building that history on purpose. You send small volumes to people most likely to open and reply, then increase gradually while watching the signals mailbox providers watch. Done properly it takes about four weeks. Skipped entirely, the recovery takes longer than the warm-up would have, and some of the reputation damage sticks to the domain permanently.

Before you start

Four things need to be true before you send the first message. Skipping any of them turns the warm-up into a slow way of building a bad reputation.

Authentication is live and passing. SPF, DKIM and DMARC records published, tested, and returning a pass on a real message, not just present in DNS. Gmail and Yahoo both require authentication from bulk senders, so a warm-up on an unauthenticated domain does not build reputation, it builds nothing. Our SPF, DKIM and DMARC guide covers the records and how to read the results.

A subdomain strategy, decided now. Send marketing from a subdomain such as news.yourdomain.com and transactional mail from another such as mail.yourdomain.com. Subdomains carry partly separate reputations, so a bad campaign does not drag your password-reset emails down with it. Changing this later means warming up all over again.

A clean, engaged seed list. Somewhere between 200 and 2,000 addresses that have opened something from you in the last 90 days, or that belong to your own team and customers. If you are launching a brand-new list with no engagement history at all, warm up on your own staff and beta users first.

Somewhere to watch the results. Google Postmaster Tools is free, takes ten minutes to verify, and gives you domain reputation, spam complaint rate and authentication pass rates for Gmail specifically. Since Gmail is usually 40–70% of a consumer list, this is the single most useful dashboard you can have during a warm-up. Microsoft’s SNDS covers Outlook and Hotmail and is worth adding if your list skews B2B.

Step 1: Week one — 20 to 50 messages a day, to people who will reply

Start absurdly small. Twenty to fifty messages a day, sent to the most engaged slice of your seed list, spread across the working day rather than fired in one batch.

The goal in week one is not delivery volume, it is positive engagement signals. Opens help. Clicks help more. Replies help most of all, because a reply is the hardest signal to fake and mailbox providers weight it heavily. If you can get colleagues and friendly customers to reply to the first few sends, do it. It is worth more than another 500 messages.

Send real content people want. A warm-up sequence of throwaway “testing, please ignore” emails teaches Gmail that mail from your domain gets ignored, which is precisely the lesson you are trying to avoid.

Step 2: Weeks two and three — double every two or three days

From roughly day 8, increase volume by 50–100% every second or third day, and start adding less-engaged segments in order of how recently they last opened something.

A workable ramp for a list of 50,000 looks like this:

• Days 1–7: 50/day to your most engaged addresses
• Days 8–10: 200/day
• Days 11–13: 500/day
• Days 14–17: 1,500/day
• Days 18–21: 5,000/day
• Days 22–28: 15,000/day, then full volume

Split each day’s send across providers rather than doing all of Gmail on Monday and all of Outlook on Tuesday. Reputation is built per provider, and a domain that only ever mails Gmail has no standing with Microsoft when you eventually send there.

Step 3: Watch four numbers, ignore the rest

Warm-ups generate a lot of noise. These four numbers decide whether to increase, hold, or stop.

Spam complaint rate. Under 0.10% is healthy. Above 0.30% you are in the territory where Gmail starts filtering by default. This is the number that ends warm-ups.

Hard bounce rate. Under 2% is fine, over 5% means your list has decay you did not clean out, and high bounce rates on a new domain read as list purchasing. Our list cleaning guide covers how to find the risky segments before they cost you.

Open rate, tracked as a trend rather than a number. Apple Mail Privacy Protection inflates opens and makes the absolute figure close to meaningless, but the direction still tells you something. A gradual decline across a ramp usually means you are reaching into staler segments faster than the domain can carry.

Domain reputation in Postmaster Tools. It moves slowly and lags several days behind your sending, so treat it as a weekly check rather than a daily one. Anything below “Medium” after week two means stopping and diagnosing rather than pushing on.

Step 4: Know when to hold or restart

Hold volume flat for three days when complaints tick above 0.10%, when a provider starts deferring your mail with 4xx responses, or when Postmaster Tools drops your domain reputation a grade.

Go back to the previous step’s volume when deferrals persist for more than a day or two, or when open rates drop by more than about a third from the ramp’s earlier days.

Stop and restart from week one when you get a hard block, a 550 rejection naming a blocklist, or a provider refusing your mail entirely. At that point the warm-up has failed and continuing to send makes the recovery longer. The blocklist removal guide covers what to do next.

Which platforms make warm-up easier

Your sending platform decides how much of this you run yourself.

Mailgun is the most direct fit if you need a dedicated IP, because automated dedicated IP warm-up is a listed feature on its plans rather than something you script yourself. Dedicated IP access begins at the 50,000-email volume level, and one dedicated IP is included at 100,000-plus volumes, with extra IPs at $59 per IP per month. Its Free plan covers 100 emails a day and Basic is $15 a month for 10,000 emails, with extra emails from $1.80 per 1,000. The real weakness for a warm-up is log retention: Basic and Free keep only one day of logs, and Foundation ($35/month, 50,000 emails) keeps five. Diagnosing a week-two deferral pattern on those tiers often means the evidence has already rolled off. Thirty-day retention arrives at Scale, $90 a month.

Amazon SES takes the opposite approach: its managed dedicated IP option hands the entire warm-up schedule to AWS, which ramps the IP pool automatically as your volume grows. It is also the cheapest way to send at genuine scale by a wide margin. The costs of that are real, though. There is no campaign builder, no templates worth the name and no support you have not paid extra for, so you are writing code or bolting on another tool. New SES accounts also start in a sandbox with tight sending limits until you request production access, which is its own small waiting game on launch week. Check the SES pricing page directly: the managed dedicated IP fee sits alongside per-thousand send charges and the total depends heavily on volume.

Postmark is the best place to watch a warm-up happen, because its bounce records carry the full SMTP response instead of a summarised category, and 45 days of message retention is standard on every plan including Basic. That combination is exactly what step 4 needs. Its Free tier is 100 emails a month, Basic is $15 a month for 10,000 emails with 5 custom sending domains, and Pro at $16.50 adds inbound processing and retention up to 365 days. Dedicated IPs start at $50 a month per IP but require a Pro plan and 300,000-plus monthly sends, so most warm-ups here happen on Postmark’s shared pools. It also bans cold outreach and enforces strict content rules, and it is not a marketing platform. You will run something else alongside it for campaigns.

MailerLite is the sensible answer for most small senders, because on its shared pools the IP reputation is already established and you are only warming the domain. Its free plan covers 250 subscribers and 2,500 emails a month, Comfort starts at $12 a month and Power at $25, with a dedicated IP available as an add-on on Power and Enterprise. The trade-off is control: you inherit whatever the shared pool’s neighbours do, you cannot see the IP reputation you depend on, and dedicated IP access is gated behind the higher tier. Support is email and chat only, with no phone line.

Brevo suits a slow ramp for a different reason: its free plan caps sending at 300 emails a day, which is almost exactly the volume ceiling you want for the first ten days anyway. Dedicated IPs sit on its Enterprise tier, and its Professional plan includes three hours a year of deliverability specialist time, which is genuinely useful if a warm-up goes sideways. Paid plans started at around $9 a month for 5,000 emails at our July 2026 pricing audit, but Brevo’s current rates load dynamically and we could not read them from the page. Confirm on Brevo’s pricing page before you budget. The daily cap that helps in week one becomes the constraint in week four, and its email builder is less polished than MailerLite’s.

Where people go wrong

• Warming up the root domain. Every future sending stream then shares one reputation. Decide the subdomain split first; retrofitting it costs you the whole warm-up.
• Sending filler content. “Test email, please ignore” builds a record of mail nobody engages with. Providers cannot tell a test from a campaign, and the ignored-mail signal is the one you are trying to avoid.
• Ramping on a schedule instead of on signals. The day counts above are a starting shape, not a rule. A domain showing 0.2% complaints on day 12 should not be doubling on day 13 because a table said so.
• Buying warm-up services that mail robot inboxes. Networks that auto-open and auto-reply between fake accounts produce engagement that increasingly does not correlate with real inbox placement, and providers have been getting better at spotting the pattern. The reputation you build is with a pool of bots.
• Restarting on a new domain after a failure. Registering yourbrand-mail.com after burning yourbrand.com moves the same sending behaviour to a domain with even less history. Fix the cause instead.
• Treating it as one-off. A domain that stops sending for two or three months cools off. Coming back to full volume after a quiet quarter needs a short re-ramp, not a cold restart.

What success looks like

You are done when three things hold at full volume for a full week. Complaint rates stay under 0.10%. Postmaster Tools shows domain reputation at High or Medium with authentication passing on essentially all mail. And your open rates at 15,000 a day look roughly like your open rates at 500 a day, meaning the ramp did not outrun the reputation.

After that, the work is maintenance rather than warm-up: consistent volume, regular list hygiene, and suppression of anyone who stops engaging. The deliverability guide covers the ongoing reputation work, and the deliverability-focused platform roundup ranks tools on inbox placement rather than feature count.

Pricing verified 12 August 2026 against the providers’ own pricing pages: Mailgun, Postmark and MailerLite plan rates and add-ons read directly from their live pricing pages. Brevo’s plan prices render dynamically and could not be read; the figure quoted is from our July 2026 audit. Amazon SES managed dedicated IP pricing was not verifiable from AWS’s page in this session and is linked rather than quoted. Vendor prices change. Confirm on the provider’s own page before you budget.

Sources

1. Google Postmaster Tools — accessed 2026-08-12
2. Amazon SES — Pricing — accessed 2026-08-12
3. Brevo — Pricing — accessed 2026-08-12