How to warm up a new sending domain

A brand new domain has no sending history, and mailbox providers treat unknown senders with suspicion by default. Warming is building that history on purpose, a little at a time, instead of finding out what an unfamiliar domain earns you on the day of your biggest send.

Key Takeaways

• Reputation attaches to your domain and to the sending IP separately. On a shared sending pool, the IP already has history from other senders. The domain is the part that’s yours to build.
• There’s no vendor-published day-by-day schedule for warming a domain. Volume ramps because each send has to earn the next one, not because a calendar says so.
• Send to your most engaged contacts first. Early opens and clicks are the signal that teaches a mailbox provider you’re wanted.
• AWS documents a specific 45-day warmup process, but it’s for a dedicated IP address, a different job most senders never need.

What warming actually protects

Two different things earn a reputation with a mailbox provider: the sending domain and the sending IP address. They’re scored separately, and which one you’re warming changes the job entirely.

If you’re sending through a shared IP pool, which is how most hosted email tools send by default, the IP already has a track record built by every account using it. You didn’t build that reputation and you can’t warm it yourself. What’s yours, and what a mailbox provider is actually meeting for the first time, is the domain: the From address, its DKIM signature, its DMARC policy. That’s what warming addresses.

Dedicated IP addresses are a separate job, and most senders on a shared pool never need it. If you lease one, it arrives with no history at all, and it’s the one part of this process AWS documents with actual numbers: see warming up dedicated IP addresses in Amazon’s SES developer guide. Confirm which one applies to you before you plan around either.

On EmailFig, that split is the product. Your sending domain, its DKIM and DMARC records, and your bounce and complaint counts belong to your account alone. The sending infrastructure a mailbox provider actually sees is shared with every other customer, which is exactly why your own domain’s reputation is the lever you control. BYOS accounts send through their own Amazon SES account instead, so the same split applies one level up: the domain is still what you’re warming, and the reputation stays with your AWS account rather than pooled with anyone else’s.

Why there’s no schedule to copy

Search for domain warming and you’ll find plenty of day-by-day tables: send 50 on day one, 100 on day two, and so on. None of them come from a mailbox provider, and none of them are wrong so much as unverifiable. Nobody outside Gmail, Yahoo or Microsoft knows the actual thresholds those systems apply, and the three don’t publish one.

What Gmail does publish is a shape, not a number. Its bulk sender guidelines, part of the wider rulebook covered in Gmail, Yahoo and Microsoft’s bulk sender rules, tell senders to “start with a low sending volume to engaged users, and slowly increase the volume over time,” and to avoid “sudden volume spikes if you do not have a history of sending large volumes.” No day counts, no percentages: volume grows because your metrics hold up, not because a schedule says day seven is the day for 500.

Amazon’s own guidance draws the same line, and only for the dedicated-IP case. AWS says establishing a positive reputation with some providers takes about two weeks and with others up to six, and that its own automatic warmup feature spreads a dedicated IP’s ramp over 45 days regardless of how much you’re sending. That’s a real, sourced number, and it describes an IP, not a domain. Applying a 45-day IP schedule to a domain you’re warming on shared infrastructure is stretching a fact past what it was measured on.

Start with the segment that already trusts you

The first send from a new domain should go to the people most likely to open it, because early engagement is the signal that teaches a provider the domain is wanted, not spam. A campaign that opens well from address one tells Gmail and Yahoo something concrete; a campaign blasted to your whole list, engaged and dormant together, tells them nothing except volume.

EmailFig’s segment builder ships a ready-made Engaged (30d) starter: subscribed contacts who opened any campaign in the last 30 days. Pick it as your audience for the first several sends from a new domain, then widen out as your metrics hold. See build a segment with the rule builder and warm up a new sending domain for the exact steps. It’s the same segment, read the opposite way, that defines a cold subscriber on an established domain.

A realistic ramp shape

Here’s the shape that follows from the guidance above.

Start with a send small enough that a bad outcome costs you little: your most engaged segment, not your full list. Watch how it lands before you send again. If opens are healthy and nothing bounced or complained at an unusual rate, the next send can go a bit further, either to a slightly larger slice of the engaged segment or to a second, only somewhat less engaged one. If a metric moved the wrong way, hold where you are or pull back, and don’t send the next planned increase on schedule anyway.

That’s the whole mechanism: each send has to earn the next one. In practice this tends to mean weeks, not days, before a domain is carrying full list volume comfortably, and a bigger list takes longer to work through than a small one. What matters is the trend across your sends, not a target day to hit full volume by.

The one instruction that shows up in every source that documents this, AWS’s included, is what to avoid: a sudden jump. Going from a cautious volume straight to your entire list in one send looks the same to a mailbox provider whether the domain is three days old or three years old, and it’s the single fastest way to get throttled or filtered.

How to tell it’s working

Two numbers matter more than anything else on the domain: your hard bounce rate and your complaint rate, both against the whole of what you sent.

Google’s bulk sender guidelines set the ceiling at a spam rate of 0.3% in Postmaster Tools, and recommend staying under 0.10% rather than treating 0.3% as a target to approach. EmailFig’s own Sending health card watches the same two numbers and flags an account as Needs attention once either climbs too high. During a ramp, don’t wait for either line to cross into that state before caring. A rate that’s climbing toward the budget, even while still under it, is the signal to hold your volume rather than increase it. See the Sending health card on your dashboard.

Opens and clicks are the second signal, softer but earlier. A domain that’s landing in the inbox shows engagement close to what the same list showed on your old sending address. A domain quietly landing in a spam or promotions folder instead shows a sudden drop in opens with nothing else obviously wrong, often before a bounce or complaint rate moves at all.

When a provider starts deferring or bulk-foldering mid-ramp

Two outcomes look different but call for the same response. A deferral is the receiving server asking you to slow down and try again later; delivery usually completes, just later than sent. Bulk-foldering is worse: the message arrives, but in Promotions or spam rather than the inbox, and nothing in your reports says so directly, only a drop in opens.

Either one, mid-ramp, means stop increasing volume rather than push through it. Hold your next send at the same size as your last successful one, or smaller, until the signal clears. Sending more into a provider that’s already hesitating is the mistake that turns a slow week into a much longer one.

While you wait, check the parts that are actually yours to fix. Confirm the domain still shows Verified, since a broken authentication record can look identical to a reputation problem from the outside: see why are my emails going to spam for the full checklist in order, and meet Gmail and Yahoo bulk sender requirements if the affected recipients are concentrated at one of those two. If the list behind the send was recently imported or hasn’t been reviewed in a while, that’s usually the actual cause: a warming domain has no reputation cushion to absorb a rough list, so problems that a mature domain would shrug off show up immediately on a new one.

Getting started

Warming is patience applied to a small, real signal: does this domain, sending to people who already want to hear from it, look like something worth trusting with more. Start there, watch the two numbers, and let the size of your sends follow what they tell you rather than a schedule borrowed from somewhere else. For the exact steps on EmailFig, see warm up a new sending domain.

If you’re choosing between sending through EmailFig’s managed infrastructure or through your own Amazon SES account first, the domain-warming work is identical either way: only the IP side of the reputation split differs. What Amazon SES actually costs covers what BYOS adds to the bill, and pricing has both plan families side by side.

Warming is one part of building a reputation; see email deliverability for people who aren’t experts for how it fits alongside authentication, bulk sender rules, and bounce and complaint thresholds.