A brand new sending domain carries no reputation, so Gmail, Yahoo and Microsoft treat its mail as suspect until you give them a reason not to. Push your full list through it on day one and you'll earn throttling, bulk-foldering and outright blocks that can take weeks to unwind. Warming is the deliberate process of building that reputation before you need it, and getting the ramp and the audience right is what keeps your first real campaign in the inbox.
Key takeaways
- A new domain has to earn reputation from zero, and you can't transfer it from an old domain or a previous ESP.
- Start low, roughly 50 to 100 messages per major provider, then grow by 20%-50% per day once volume is substantial rather than doubling blindly.
- Send to your most engaged subscribers first (opened or clicked in the last 30 days), then widen the window in stages instead of one jump.
- Google asks bulk senders to keep the spam rate under 0.10% and never reach 0.30%; treat a 2% hard-bounce rate as the line where you hold volume.
- Full warmup takes four to eight weeks depending on volume and engagement, and reputation can keep fluctuating for up to 120 days after that.
What does warming a sending domain actually build?
Warming builds a track record of wanted mail that mailbox providers attach to your domain. Providers score senders on spam complaints, hard bounces, spam-trap hits, authentication and engagement, then decide whether your mail reaches the inbox, the spam folder or gets rejected before delivery.
The reputation you're building lives on the domain, not just the IP. Microsoft's warm-up guidance for Dynamics senders is blunt about the consequence: you can't transfer reputation from one domain to another, or from your old IPs to new ones, so a fresh domain starts from zero every time. Gmail in particular scores the domain reputation coupled with the IP reputation, which is why moving ESPs does not let you skip the work.
Why can't you skip warmup on a brand-new domain?
Because providers read a sudden spike from an unknown domain as exactly the pattern spammers produce. A domain that has never sent, then dumps 200,000 messages in an hour, looks identical to a compromised or throwaway sender, and the filters respond accordingly.
Google's bulk sender guidelines tell you to "start with a low sending volume to engaged users, and slowly increase the volume over time," and to "avoid introducing sudden volume spikes if you do not have a history of sending large volumes." Since February 2024 those guidelines apply to anyone sending 5,000 or more messages a day to Gmail accounts, and once you cross that line Google keeps treating you as a bulk sender regardless of later volume, so the reputation you build now is the one you live with.
What volume should you start at, and how fast can you ramp?
Start small, near 50 to 100 messages per major provider on the first days, then let engagement decide the pace. Postmark's domain warmup guide suggests doubling volume each day while numbers are tiny, then slowing to a 20%-50% daily increase once volume becomes substantial, with most senders reaching dependable full-volume deliverability in three to six weeks.
Higher targets take longer. Microsoft puts full warmup at four to eight weeks depending on target volume and engagement. The exact numbers matter less than the shape: a smooth curve the provider can predict, not a staircase of doublings. A workable ramp toward roughly 100,000 a day looks like this.
| Phase | Approx. daily volume | Audience | What to confirm before advancing |
|---|---|---|---|
| Days 1-3 | 500 to 1,000 | 30-day engaged | Bounces under 2%, complaints under 0.10% |
| Week 1 | 2,000 to 5,000 | 30-day engaged | Opens strong, no provider throttling |
| Week 2 | 10,000 to 15,000 | Add 60-day cohort | Bounces stable, no blocklist hits |
| Week 3 | 20,000 to 30,000 | 60-day cohort | All metrics stable seven days running |
| Week 4+ | 50,000 to 100,000 | Full active mix | Stable at least 14 days before each step |
Consistency compounds the effect. Microsoft notes that sending less than weekly during warmup stretches the whole process out, because providers need a steady signal to trust.
Which subscribers should receive the warmup sends?
Your most engaged subscribers first, then widen the window in stages. These are your recent opt-ins and your regular openers and clickers, the people least likely to bounce or hit "report spam," and their positive signals are what teach the provider your mail is wanted.
Microsoft's plan is specific: send to people who opened or clicked in the past 30 days during weeks one and two, expand to the 60-day cohort in weeks three and four, and keep anyone who hasn't engaged in 90 days out of the send for the first six weeks. When you do add colder segments, add them in chunks of about 15% of your existing volume so a batch of stale addresses can't tip the reputation you've built.
This is also where list quality decides your fate. Warming into a list full of dead or mistyped addresses spends your fragile new reputation on hard bounces, so it pays to verify the list before the first send. If a specific cohort keeps dragging complaints up, the same engagement lens tells you which segment is quietly damaging your reputation so you can pull it from the ramp.
What signals tell you to slow down or hold?
Bounces, complaints and provider-specific delays are your dashboard, and any one of them going wrong means hold, not push. The rule is simple: never raise tomorrow's volume until you've read today's performance by receiving provider, because Gmail can be clean while Microsoft is deferring you.
Watch these thresholds. If you cross one, hold the current volume for another seven days or cut back by 25%-30% until the numbers normalize, rather than continuing the ramp.
| Signal | Healthy during warmup | What it means when it breaks |
|---|---|---|
| Spam complaint rate | Under 0.10% | At 0.30% Google treats you as ineligible for mitigation |
| Hard bounce rate | Under 2% | List quality problem; stop widening the audience |
| Provider deferrals (421) | Clears within 72 hours | If timeouts pile up, tighten the engagement window |
| Bulk-foldering | Clears after a few engaged sends | Keep sending to engaged only, do not add cold segments |
Microsoft's guidance expects some bulking and 421 delays early at Yahoo, AOL, Gmail and Comcast, and the correct response is to keep sending to engaged recipients rather than panic-pausing, which resets the signal you're trying to build. Persistent timeouts, though, are a cue to tighten the window, meaning send only to people who engaged more recently.
How is warming different from ramping?
They're related but not the same, and conflating them leads operators to stop too early. Klaviyo separates ramping from warming: warming is the period in which you establish a reputation as a legitimate sender on new dedicated infrastructure, while ramping is the gradual volume increase that aids that process and applies to every new sender, even on shared infrastructure.
The practical distinction is who needs which. A brand new domain or a newly registered branded domain needs a full warm, roughly 30 days of clean sending. A sender moving onto an established shared pool still needs to ramp volume, because dumping a large list into a new account looks like abuse even when the underlying IPs are already trusted. Klaviyo notes that reputation can keep fluctuating for up to about 120 days while providers validate you, so the ramp discipline outlasts the formal warmup.
Do you need a dedicated IP, or does the shared pool warm your domain?
Most senders don't, and forcing a dedicated IP too early makes warmup harder, not easier. A dedicated IP needs enough consistent volume to hold its own reputation, and a low-volume sender can't generate the steady signal a cold IP demands, so the ESP's shared pool of already-warm IPs does the IP half of the job while you warm the domain half.
The threshold is volume. Klaviyo's warming requirements recommend a dedicated IP only once you consistently send over 1,000,000 emails per month, and they define a successful warm as about 30 days holding bounce rate under 1%, unsubscribe rate under 0.30% and spam rate under 0.01%. Below that volume, a branded (dedicated) sending domain on shared IPs gives you domain-level control without the burden of warming an IP you can't keep busy.
| Setup | Best for | Warmup burden |
|---|---|---|
| Shared IP, shared domain | New or low-volume senders | Ramp volume only; IP reputation is inherited |
| Shared IP, branded domain | Most established brands | Warm the domain; IPs are already warm |
| Dedicated IP, branded domain | Over ~1M sends/month | Warm both the IP and the domain |
Trade-offs and what to watch
The real risk in warmup isn't going too slow, it's engineering a cliff at the end. If you warm exclusively on your most engaged 30-day cohort for the full period and then switch to your whole list overnight, providers see the audience shift as a fresh spike and can pull you back into filtering right when you thought you were done. Widen the window in stages so the transition to your real audience mix is gradual, not abrupt.
The second trap is treating warmup as a one-time event. Reputation decays if you go quiet, so a domain warmed and then left idle for a month sends its next campaign closer to cold than warm. Keep a consistent cadence to your engaged base even between promotions.
The last thing to watch is authentication, because none of the ramp discipline matters if your mail can't authenticate. SPF, DKIM and DMARC alignment have to pass before the first warmup send, since a domain that fails alignment doesn't get to build reputation at all, it gets rejected. Set those up, confirm the From header aligns with SPF or DKIM, then start the ramp.