Mailbox warmup and inbox placement are the two halves of getting cold email seen: warmup builds a new mailbox's sending history slowly before launch, and placement testing measures whether mail actually lands in the inbox at Google, Microsoft and Yahoo. You warm every mailbox before real volume, watch its health every day, and pull back the moment placement slips. Inside Artemis, the system behind the AGP growth machine, all three of those steps run on every mailbox.
Nerdy Joe is owned by Agency Growth Partners (AGP), so this is our own system and we are not neutral about it. Below is exactly how it works, a set of criteria you can use on any provider or a DIY setup, and the cases where it is not worth it.
This post picks up where cold email sending infrastructure for agencies leaves off. Domains are bought and authenticated, mailboxes are connected. Now they need to earn trust.
Why warmup and placement get their own stage
A new mailbox on a new domain has no history. To a mailbox provider, a stranger that suddenly sends hundreds of emails looks the same as a spammer. Google's own sender guidelines tell senders to avoid sudden volume spikes, start with low volume and increase it slowly, and to increase more slowly the more email they send.
Placement is the scoreboard. A campaign can show near-zero bounces and still be landing in spam at one provider, which quietly kills replies. The only way to know is to test where mail lands, by provider, on an ongoing basis.
In AGP's sequence (Advise, Build, Fill, Close, Partner), warmup sits at the end of Build. Nothing moves to Fill until mailboxes are ready.
How warmup works inside Artemis
Each mailbox is warmed slowly before launch. Warmup exchanges email with other inboxes, so the new mailbox builds a normal pattern of sending, receiving and replies.
The ramp starts at about 10 a day and increases gradually. Progress is tracked per mailbox out of 40 steps, so you can see exactly where each one stands instead of guessing from a start date.
| Warmup element | How it runs in Artemis |
|---|---|
| Start point | About 10 warmup emails a day |
| Ramp | Gradual increase over time |
| Progress tracking | Per mailbox, out of 40 steps |
| What warmup does | Exchanges email with other inboxes to build history |
| Before launch | Every mailbox is warmed before real volume |
Why tracking by step beats tracking by date
Two mailboxes created on the same day will not warm at the same pace. One may hit a bump at a provider; another may move cleanly. Step-based progress lets the growth engineer launch the ready ones and hold the slow ones, instead of flipping the whole pool on at once.
It also stops the most common mistake in DIY setups: launching on a calendar date because "three weeks have passed," whether or not the mailbox is actually ready.
Where warmup fits in the launch timeline
Warmup is the reason a new engine does not send on day one. Here is how it lines up with AGP's published timeline:
| When | What happens |
|---|---|
| Week 1 | Strategy: ideal buyer, offer and markets defined |
| Weeks 1 to 3 | Build and warm: domains, mailboxes and warmup ramps run in parallel |
| Weeks 3 to 4 | Launch, with first replies within days |
| Month 2 onward | Follow-ups compound while mailboxes stay monitored |
Time to launch is about 3 to 4 weeks. Running strategy and build in parallel with warmup is what keeps that window short without rushing mailboxes that are not ready.
Daily mailbox health monitoring
After warmup, monitoring does not stop. Artemis checks every mailbox daily and shows four things for each one.
- Health score. A single number that summarizes how the mailbox is doing.
- Daily send count against limit. For example, 23 of 50 on AGP's sample dashboard.
- Warmup progress. Where the mailbox sits in its 40 steps.
- Inbox placement. Where its mail is landing.
The value of daily checks is speed. Placement problems tend to start small at one provider and spread. A mailbox that slips on Tuesday and gets noticed on Tuesday is a minor fix. One noticed at the end of the month is a burned domain.
Google gives the same advice for its own channel: as you gradually increase volume, monitor delivery with Postmaster Tools. Artemis applies that habit per mailbox, every day.
Placement tests at Google, Microsoft and Yahoo
Artemis measures inbox placement at Google, Microsoft and Yahoo with placement tests. Splitting by provider matters because each one filters differently. A blended number can hide a serious problem at one provider behind strong results at the other two.
Here is what that view looks like on AGP's sample dashboard. These figures are from a sample, not a guarantee and not an average across clients.
| Provider | 14-day placement (AGP's sample dashboard) |
|---|---|
| 97% | |
| Microsoft | 95% |
| Yahoo | 94% |
What "90%+ healthy" means
AGP treats 90% or higher placement as healthy. In plain terms, at that level roughly nine in ten test messages land in the inbox rather than in spam at that provider.
Three things follow from using a line like that:
- It is per provider. A mailbox at 97% at Google and 82% at Microsoft has a Microsoft problem, and the fix targets that.
- It is a trend, not a moment. The sample uses a 14-day window. One bad test is a signal to watch; a falling two-week line is a signal to act.
- It sets the response. Above the line, keep sending. Below it, reduce volume and find the cause.
Placement is also only half the story. It tells you mail is being seen. Whether it gets replies depends on the list and the message, which is why signal-based cold email personalization is the next stage.
Resting mailboxes and automatic throttling
When a mailbox starts to slip, meaning it starts landing in spam, Artemis drops its volume automatically. If it keeps sliding, it rests: it comes out of campaign volume while the rest of the pool carries the load.
AGP's sample dashboard shows a panel of 48 mailboxes: 41 healthy, 6 warming and 1 resting. The resting one shows a health score of 71 and 84% placement, below the 90% line. Again, this is a sample panel used to show how the view works.
| Mailbox state | What it means | What the system does |
|---|---|---|
| Warming | Still building history | Warmup only, no full campaign volume |
| Healthy | Health and placement in good range | Sends under its daily limit |
| Slipping | Starting to land in spam | Volume drops automatically |
| Resting | Health or placement too low | Pulled back so it can recover |
The design principle is simple. You never solve a placement problem by sending more from the mailbox that has it. You lower the pressure, keep the engine moving with the healthy mailboxes, and bring the weak one back when it recovers.
This is a good place to see how AGP's growth machine works as a chain: data quality lowers bounces, infrastructure spreads load, warmup builds history, monitoring catches slips, and throttling protects the pool.
What breaks placement after warmup
Warmup gives a mailbox a clean start. These are the things that undo it.
Bad data
High bounce rates tell providers you do not know who you are emailing. Inside Artemis, contacts are enriched and verified before anything sends, and every address is verified before it is sent to.
Spam complaints
Google asks senders to keep their reported spam rate below 0.10% and to never reach 0.30%. Yahoo asks bulk senders to stay below 0.3%, and notes it calculates spam rate based on mail delivered to the inbox. Irrelevant emails to the wrong people are the fastest route to complaints.
Ignoring unsubscribes
Yahoo asks bulk senders to honor unsubscribes within 2 days. In Artemis, unsubscribes are honored automatically, and the reply inbox sorts them so nobody gets emailed again by accident. Our post on AI reply handling and follow-up covers how replies get sorted.
Volume spikes
Adding a big new list and doubling sends overnight looks exactly like the spike Google warns against. Grow volume by adding warmed mailboxes, not by raising caps on the ones you have.
Who watches all of this
Each AGP client engine has a dedicated growth engineer who builds, runs and tunes it, and deliverability is part of that job. The growth engineer decides which mailboxes are ready to launch, reads placement by provider, and acts when a mailbox rests.
The client sees results in a live dashboard that tracks sends, replies, interested prospects, meetings and deals. Mailbox health is the engineer's daily work so it does not become the client's problem.
How to judge any warmup and placement setup
Use these checks on any provider, any sending tool, or your own build. For a comparison of standalone tools, see our roundup of email warmup tools, and for the base setup steps see the cold email infrastructure guide.
- [ ] Every new mailbox is warmed before it sends real volume
- [ ] Warmup starts low and ramps gradually, not straight to full volume
- [ ] Warmup progress is tracked per mailbox, not by calendar date
- [ ] Health is checked daily, per mailbox
- [ ] Daily send count against limit is visible per mailbox
- [ ] Placement is tested at Google, Microsoft and Yahoo separately
- [ ] There is a stated health line (for example, 90%+) and a stated response below it
- [ ] Volume drops automatically when a mailbox slips
- [ ] Weak mailboxes can rest without stopping the campaign
- [ ] Lists are verified before sending, and unsubscribes are honored automatically
If a provider cannot show you placement by provider for one named mailbox, they are guessing. If they talk only about open rates, they are measuring the wrong thing.
Who this is not a fit for
If you send a few dozen emails a week from one inbox, a full warmup and placement system is more than you need. Steady, low, relevant sending and a clean list will carry you.
It is also not a fit if the goal is to push volume no matter what. Artemis is built to slow mailboxes down when they slip, and some buyers find that frustrating. If you want an engine that keeps blasting through a placement drop, this is the wrong design.
Finally, warmup cannot rescue an offer nobody wants. If replies are low while placement is healthy, the problem is upstream, in the audience or the offer.
Want this built for your agency?
If you want warmup, placement testing and the rest of the AGP machine run for your agency by a dedicated growth engineer, book a call with David. You will get a straight answer on whether it fits.
Sources
Frequently asked questions
How long should you warm up a new mailbox before cold emailing?
Long enough that it has a steady history of normal sending before real volume starts. In AGP's timeline, building and warming run through weeks 1 to 3, with launch in weeks 3 to 4, and warmup progress is tracked per mailbox rather than by the calendar.
What is a good inbox placement rate for cold email?
AGP treats 90% or higher as healthy in its sample dashboard. Below that, the right move is to find which provider is slipping and reduce volume on the affected mailboxes, not to push harder.
What is the difference between deliverability and inbox placement?
Deliverability often just means the email was accepted and did not bounce. Inbox placement asks the harder question: did it land in the inbox or in spam. A mailbox can deliver nearly everything and still place badly.
Should you keep warmup running after a campaign launches?
In most setups, yes. Ongoing warmup and daily monitoring help a mailbox keep a steady pattern while real sends go out, and they make it easier to spot a mailbox that starts to slip.
What does resting a mailbox mean?
Resting means pulling a mailbox out of real campaign volume for a while so it can recover, while the rest of the pool carries the load. It is the response to a falling health score or placement, before the problem spreads.
Can warmup fix a bad list or bad copy?
No. Warmup builds a sending history, but high bounces from unverified lists or spam complaints from irrelevant emails will undo it. Clean data and relevant copy are what keep placement up after launch.