What email warmup tools do, and what the mailbox providers say about it

Your sending platform has told you the domain needs warming, there is a monthly line item attached, and every page ranking for this question is published by a company that sells warmup.

Here is what is checkable. The vendors describe the mechanism openly and it is not subtle. Google and Yahoo do not mention the practice anywhere in their sender guidance, in either direction. And the tools that measure the result have been reported returning opposite verdicts on the same domain.

What a warmup tool does, in its vendors' own words

The category is unusually candid about the mechanism, which makes this easy to establish.

Smartlead: its AI sends, opens and replies to warm-up emails, even moving them from spam to the inbox. Mailwarm describes its emails getting out of the spam folder, opened, marked as important, and getting a reply. Warmbox says its service sends realistic everyday emails from your inbox, then removes them from the spam folder, opens and bookmarks them and replies to some of them, against a network it puts at over 35,000 private inboxes.

Instantly is the most specific. It describes a deliverability pool of over 1,000,000 real email accounts, a system that uses a private network of headless browsers, and an option that makes that system spend time scrolling through your warmup emails to emulate human-like reading.

So the product is a network of accounts, some of them driven by browser automation, generating opens, replies and spam-folder rescues against a schedule. That is what you are buying. Nothing in this paragraph is contested by anyone selling it.

What the mailbox providers say about it

Nothing, as far as we can find, and that absence took some checking.

The words warm up, warmup and warming appear nowhere on Google's email sender guidelines, nowhere in its thirty-question sender FAQ, and nowhere in its list of the top ten reasons senders have problems. They appear nowhere in Yahoo's sender best practices or its sender FAQs. They appear nowhere on Microsoft's outbound spam protection documentation, updated 21 May 2026, though that page governs mail leaving a Microsoft 365 tenant rather than what Microsoft asks of senders mailing Outlook.com. Microsoft's sender-facing requirements page is the one document in this set we could not retrieve.

That is a genuine absence rather than a permission. No provider we read says warmup is allowed, and none says it is prohibited. The practice sits outside the documented rules, which is a different position from being inside or outside them, and any page telling you that Google endorses warmup or that Google banned it is telling you something no Google page says.

What the providers do document is the thing warmup is often confused with.

Google publishes a ramp, and it is not this

Google's sender guidance asks you to start with a low sending volume to engaged users, and to slowly increase the volume over time. Its list of common sender problems carries the same instruction as an item in its own right: ramp up slowly.

That is a behaviour, and it is free. It means sending to people who want to hear from you, in small numbers at first, and increasing gradually while watching complaints. Google never calls it warming up. The category does, which is how a paid service and a sending habit ended up sharing a name on every page that ranks for this.

The metric Google actually watches is defined just as plainly: the spam rate is the percent of your messages that are delivered to an engaged recipient's inbox and then marked as spam by that recipient. Google asks you to keep it under 0.10% and never to reach 0.30%.

Read that definition against the mechanism above and one thing follows arithmetically. Warmup traffic enters the denominator of that rate and never the numerator, because the accounts in the pool are never going to mark you as spam. A rate improved that way is improved by dilution. That is inference from Google's own definition rather than a claim Google makes, and it is worth holding lightly, but the arithmetic is not complicated.

Two clauses that sit nearby, and name nothing

Two pieces of Google policy live close to this mechanism, and neither of them mentions warmup.

The Gmail program policies, last updated 13 November 2025, state that you are not allowed to automate the Gmail interface, whether to send, delete, or filter emails, in a manner that misleads or deceives users. The Workspace developer policy names applications that use multiple accounts to abuse Google policies, bypass Gmail account limitations, circumvent filters and spam, or otherwise subvert abuse or safety restrictions.

Whether a pool of accounts opening and replying to each other's mail falls inside either sentence is not something Google has answered in writing about this category. We are quoting the two clauses because a reader deciding whether to run this on a business-critical domain should read them, not because they resolve the question.

What the vendors who address limits concede

The vendors that address limits at all do it on their own pages, which makes those sentences the most reliable part of the category's marketing.

MailReach states that warmup alone is not enough, and lists what still matters: authentication through SPF, DKIM and DMARC, list quality, and relevant content. Instantly documents that its warmup is automatically disabled when your accounts bounce too many warmup emails from other users in its pool, which is a plain statement that the pool itself degrades and that the service switches off when it does.

Every item on that concession list is free and under your control. Authentication is an afternoon of DNS records. List quality is measurable before you send, and bad addresses produce the bounces that get sending reputations damaged in the first place, which is why verification sits upstream of everything else.

When two tools disagree about one domain

When we looked at deliverability testing, one sender at 20 to 30 emails a day, on an authenticated eight-month-old domain, had GMass reporting that everything leaving it was going to spam while Warmup Inbox reported the same mail landing in the primary inbox.

That is the state of measurement in this category, and it is why the question "did warmup work" is harder to answer than it looks. When two tools return opposite verdicts on one domain, neither can be your evidence that a subscription is doing something. What can actually be tested is narrower than the dashboards suggest.

If it is already switched on inside your sending tool

You may not be buying a warmup subscription at all. Instantly bundles warmup into every Outreach plan and Smartlead includes a warmup pool on every tier, so for a lot of senders this is a setting rather than a purchase.

That is a smaller decision than it looks. On a plan that bundles it, leaving it on costs you nothing extra, and nobody has measured either state, so there is no evidence to act on in either direction. The expensive version is leaving it on and believing it is protecting you, because that belief is what stops people fixing authentication, cleaning the list and slowing the ramp.

If the warmup is a separate charge rather than a bundled toggle, the question is worth asking properly: what did the vendor measure, on what sample, against what control. Nobody in this category has answered that in public.

What to do at 30 to 100 emails a day

Set up SPF, DKIM and DMARC properly, which costs nothing and is the only part of this with a pass or fail. Ramp the way Google asks, starting with a low volume to engaged recipients and increasing it slowly. Clean the list before you send to it. Keep the volume low enough that a bad week does not produce complaints you cannot recover from.

Then decide about warmup on its own terms, knowing what it is: a paid network of accounts generating artificial engagement, unmentioned in Google's and Yahoo's sender guidance, sold in a market where the same 50 warmup emails a day cost $15 a month at Warmbox and $69 at Mailwarm, and where Mailwarm's $479 tier buys ten times that volume rather than a better result. It may help. Nobody outside the companies selling it has published evidence that it does, and we could not find a measured study from anyone without a product in the category, nor any mention of the practice in Google's or Yahoo's sender guidance.