Opens in a new tab

Deliverability

Warming up a new sending domain in six weeks

A realistic ramp with engagement tiers, the numbers that tell you to slow down, and an honest note on why every schedule you find online, including ours, is a heuristic.

Mailnaptic team

Published 20 October 2026

Last reviewed 2 October 2026

9 min read

Domains / news.example.com

Warm-up status (sample data)

Level

3/10

Health score

82

Healthy

Est. cap

~160/day

Active today. Advances to Level 4 in the next warmup cycle.

A new domain has no history. Mailbox providers see mail from a name they have never met, and some of them limit how much they accept until you prove people want it. Microsoft’s warmup guidance puts it plainly: it takes four to eight weeks to reach maximum deliverability, depending on target volume and engagement, and longer if recipients do not seem to want the mail.

So you ramp up. The question is how. Most guides give you a table of numbers. We will give you one too, but the numbers are the least important part. What carries the plan is who you send to in each week, and what you do when the feedback turns bad.

Before you start

Warmup assumes your authentication is already in place. If SPF, DKIM and DMARC are not set up, do that first with our setup walkthrough. Google requires SPF or DKIM for every sender, and SPF, DKIM and DMARC for anyone sending more than 5,000 messages a day to Gmail accounts.

What warmup is and is not

Warmup is a sequence of small, well-received sends that lets providers build a picture of you. It is not a trick. Google’s own sender guidance says to start with a low sending volume to engaged users and slowly increase it, and says nothing more specific. There is no official table of daily volumes from Google or Yahoo. Every schedule you read, ours included, is somebody’s experience.

That has a practical consequence. If a schedule says "1,000 on day nine" it cannot be right for everyone, because the real input is how recipients react. Complaints, bounces and deferrals are the dial. The calendar is only a guess about where the dial will let you go.

Three ways to read the same ramp

Week

Omnisend sample ladder

Our six-week sample

Mailnaptic level cap (est.)

1

250 per day

300 per send

L1 to L3: ~50 to ~200 per day

2

500 per day

700 per send

L3 to L4: ~200 to ~400 per day

3

1,000 per day

1,500 per send

L5: ~800 per day

4

2,000 per day

3,000 per send

L6: ~1,600 per day

5

not covered

5,500 per send

L7: ~3,200 per day

6

not covered

whole list

L8 and up: ~6,400 per day and more

Sample data. Omnisend's ladder is from their published article, a four-week plan. The middle column is our own example, not a benchmark. The Mailnaptic column lists the estimated capacity per warm-up level shown on the Domains page; which level you sit at in a given week depends on how often you send, so the pairing by week is only illustrative.

The flat ladder, and where it breaks

Omnisend publishes a four-week sample: 250 emails a day in week one, 500 in week two, 1,000 in week three and 2,000 in week four, starting with existing subscribers and adding customers and new signups later. It also suggests no more than one campaign a day and three a week. It is a fair outline and easy to follow.

Where it breaks is the flatness. It does not say what to do when week two goes badly. It also tops out at 2,000 a day, which suits a list of a few thousand and leaves a 50,000 contact list with no plan for week five. If you stick to it mechanically, you will either stall or jump. The better habit is to treat the ladder as a ceiling you step toward only when the numbers say go.

Engagement tiers: the part that matters

Microsoft’s guidance gives the clearest tier structure we found. In weeks one and two, send to people who opened or clicked in the past 30 days. In weeks three and four, widen to 60 days. During the first six weeks, do not send to anyone who has not opened or clicked in the past 90 days. When you add an older segment, add it in chunks of about 15 percent of existing volume.

The reasoning is simple. Recent openers rarely complain and rarely bounce, so early reputation data looks good. Old addresses are where spam traps and abandoned mailboxes live. See list hygiene and spam traps for why that matters.

Weeks 1 and 2

Opened or clicked in 30 days

Newest opt-ins and your regular openers. Least likely to complain or bounce.

Weeks 3 and 4

Opened or clicked in 60 days

Add the next ring in small steps, no more than about 15 percent of current volume per step.

Weeks 5 and 6

Opened or clicked in 90 days

Broaden to the 90 day ring only if complaints and bounces are flat.

Not yet

Silent for more than 90 days

Keep them out for the first six weeks. Run a win-back later.

Windows from Microsoft Learn’s warm-up guidance for marketing senders (30, 60, 90 days; 15 percent steps). Mapping weeks to rings is how we read that guidance.

A six-week plan you can copy

Here is the sample ramp for a list of 10,000 contacts who have opened or clicked within the last 90 days. Treat the figures as a starting shape and rewrite them for your list size. The principles are what to keep.

  1. Week 1, about 300 per send, two sends. Newest opt-ins and 30 day openers only. Send a welcome or best-content email people will want, at a steady time.
  2. Week 2, about 700 per send. Same tier, more of it. Check numbers before each send.
  3. Week 3, about 1,500 per send, three sends. Move to the 60 day ring in a 15 percent step.
  4. Week 4, about 3,000 per send. Keep the 60 day ring. Resist adding anything new this week.
  5. Week 5, about 5,500 per send. Open the 90 day ring if everything has been flat.
  6. Week 6, the full 10,000. Everyone who opened or clicked in 90 days. Anyone silent for longer still waits.

Consistency counts. Microsoft notes that the steadier you are with volume, frequency, complaint and bounce levels the faster you build reputation, and that sending less often than weekly slows it down. So a Tuesday and Thursday rhythm beats one big blast and two weeks of silence.

Sample ramp. Not a benchmark.

Example send size, 10,000 recently engaged contacts (sample plan)

Wk 1

Wk 2

Wk 3

Wk 4

Wk 5

Wk 6

Week 1: 300 per send, 2 sends

300

Week 3: 1,500 per send, 3 sends

1,500

Week 6: the full engaged list

10,000

Sample ramp for 10,000 recently engaged contacts. The shape is the point: small, then roughly doubling, then the whole engaged list. Not a benchmark and not a promise of inbox placement.

Mixing in transactional mail

If you already send order confirmations, password resets or receipts, keep sending them. Real, wanted, one-to-one mail is the best reputation signal you have, and it tends to be opened. Whether to send it from the same domain as your marketing is a design choice: shared domains share reputation, which helps early and hurts if marketing goes sideways. We lean toward keeping the domain shared during warmup and splitting only if you have a reason. That is our judgement, not a rule from the sources above.

Mailnaptic offers transactional email on Pro and above, and we have not verified here whether warm-up caps apply to it the same way as to campaigns and automations. The Domains page lists the caps for campaigns and automations. Check what your account shows before you plan around it.

One more caution: do not fake it. Sending marketing mail dressed up as a receipt to build volume is the kind of thing that creates complaints. See our Black Friday sending plan for why warmup should be finished well before a peak.

Example week 1 (sample)

Tuesday

Marketing: newest opt-ins and 30 day openers

300 sends

Wednesday

Transactional mail keeps flowing, if you already send it

Order and account emails

Thursday

Marketing: second group of the same tier

300 sends

Friday

Postmaster Tools, complaints, hard bounces

Read numbers

Example only. Names a plausible first week for a store with order emails; yours will differ.

The feedback you read each week

You need three numbers after every send. Complaints, hard bounces and deferrals or delays.

  • Complaints. Google asks senders to keep the spam rate shown in Postmaster Tools below 0.10 percent and to avoid ever reaching 0.30 percent. It says that rates above 0.1 percent can hurt inbox delivery for bulk senders. Yahoo’s published line is below 0.3 percent. For how to read the Gmail view see Google Postmaster Tools explained.
  • Hard bounces. No provider publishes a single warmup threshold we could verify. Treat this as a rule of thumb only: a rising bounce rate in the first weeks points to a dirty source, so stop adding segments if hard bounces creep above roughly 2 percent and find out where those addresses came from.
  • Deferrals and delays. Microsoft describes 421 delays at AOL, Microsoft and Comcast that retry for 72 hours before bouncing as a 5XX. That is not necessarily a problem if the mail is delivered. If messages time out in significant numbers, tighten your engagement window.

Expect some turbulence. Microsoft’s guidance says to expect some bulking, meaning mailbox providers filtering your marketing mail as bulk or spam, and some blocking, and that the answer is to keep sending to engaged subscribers rather than stop.

Empty graphs

Gmail’s numbers cover Gmail recipients only. If a graph is empty or missing, treat that as no information, not as good news.

When to slow down

Slow down means repeat last week’s volume and tighten the tier. It does not mean stop. The diagram below is the rule we would follow.

Weekly decision

Send to this week’s tier

Same time, same cadence

Wait 24 to 48 hours, read the numbers

Complaints, hard bounces, deferrals

Complaints near 0.1 percent or bounces climbing?

Gmail asks for under 0.10, and never 0.30 or higher

Yes: repeat last week’s volume, tighten the tier

Drop the oldest ring, check list source

No: add the next step (about 15 percent)

Move up a tier or a volume step, not both

Decision flow we use. The 0.10 and 0.30 percent figures come from Google’s sender guidelines; the 15 percent step comes from Microsoft Learn.

What Mailnaptic warmup does

Once a domain is verified, the Domains page shows a Warm-up status card. Warm-up starts automatically with your first campaign send. The card shows a level from 0 to 10, a health score, and an estimated daily cap for campaigns and automations.

The level labels read roughly like this: Level 1 about 50 a day, Level 2 about 100, Level 3 about 200, Level 4 about 400, Level 5 about 800, Level 6 about 1,600, Level 7 about 3,200, Level 8 about 6,400, Level 9 about 12,800, and Level 10 about 25,600. Each level doubles the one before, up to a platform daily cap that is 50,000 a day by default. Those are estimates in the app, not guarantees. The card also tells you when you are active: if you sent in the last 24 hours it says you advance to the next level in the next warmup cycle, and if you did not it asks you to send a campaign today.

Two things we want you to understand. First, the cap is a ceiling, not a recommended schedule. The tiers in this article decide who you send to; the cap only prevents you from going too fast. Second, the health score moves the cap. Per the interface, the score comes from bounce rate, complaint rate and open engagement, with 70 and above healthy, 40 to 69 fair, and below 40 possibly pausing progression. Below 80 you get a recovery plan button.

Domains / Warm-up status

Health score 70 and above

Green bar. Keep climbing.

Healthy

Health score 40 to 69

Amber bar. Hold volume, find the cause.

Fair

Health score below 40

Red bar. Stop adding segments.

May pause progression

Below 80

AI recovery plan is offered on the domain card.

Recovery plan button

From the Mailnaptic Domains page: the score is built from bounce rate, complaint rate and open engagement.

What the Domains page shows, with the bands as written in the app. Layout simplified.

Why every schedule is a heuristic

No provider publishes the rules. Google says gradual and engaged. Yahoo says nothing on ramps. Microsoft gives windows and 15 percent steps. M3AAWG’s published documents list sender best practices and no warmup table that we could find. Everything else is vendor experience, including the sample numbers on this page.

Three things change the real answer. Your list source: Microsoft’s guidance puts opt-ins first, and mail to people who did not clearly ask for it takes longer to warm, if it warms at all. Your cadence: weekly or better. And your engagement: a list where most people open warms differently from one where few do. Use the schedule to plan, and the feedback to decide.

And if you must send a big campaign before week six? Send to your most engaged group only and wait. A smaller campaign that lands well is worth more than a full one that goes to spam. Silent subscribers can be handled later with a win-back campaign.

For the wider picture of what a small sender should do in its first month, see our 30 day plan for a small business. For the rules the major providers enforce, see Gmail and Yahoo sender requirements.

Try it in Mailnaptic

Verify your domain, send your first campaign to your most engaged contacts, and watch the warm-up level and health score on the Domains page. See the deliverability page for how the reputation tools fit together.

Sources