For about fifteen years, deliverability advice was the vegetables of email marketing. Sensible, worthy, and quietly ignored by everyone chasing this month's revenue number.
That's over.
Since November 2025, Google has treated deferral and rejection as one escalating ladder. Keep failing the sender requirements and the temporary 4.7.x responses nobody was reading become permanent 5.7.x rejections. Microsoft went further and went earlier. Since 5 May 2025 it has rejected non compliant bulk mail outright from Outlook, Hotmail and Live with a 550 5.7.15 error rather than routing it to junk.
Rejected is not the spam folder. It means the message never entered the mailbox at all.
Plenty of email programmes were built off a 2022 checklist and never revisited. This is the gap between what got set up then and where the goalposts sit now.
The rules, plainly
Google and Yahoo set the baseline in February 2024, and both publish it in their own sender guidelines. It applies to any domain sending 5,000 or more messages a day to personal accounts.
You need SPF, DKIM and DMARC configured and aligned. You need one click unsubscribe implemented as a header, not just a link in your footer, honoured within 48 hours. And you need to keep your spam complaint rate under 0.3%.
Microsoft's published requirements are authentication focused, and they enforce them with rejection rather than filtering. Yahoo won't publish a volume threshold at all. Its guidance defines a bulk sender as one sending significant volume and declines to put a number on it, which means you cannot count your own sends and rule yourself out.
One detail that catches brands out, and it's worth confirming against Google's own sender guidelines before you rely on it: bulk sender classification is widely understood not to reset when your volume drops. Plan on the assumption that once you're in, you're in. You don't get to have a quiet January and revert to being a small sender.
0.3% is not your target
This is the most common misreading of the rules, and it's the expensive one.
Google's guidance is to stay below 0.10% and never reach 0.30%. Those are two different numbers doing two different jobs.
0.30% is the enforcement line. Cross it and you're into bulk spam foldering territory.
0.10% is where degradation starts. Google's own FAQ says, in the present tense, that rates above 0.1% already harm delivery for bulk senders. Between the two figures your deliverability quietly rots while your dashboard still says you're compliant.
Manage to 0.1%. Treat 0.3% as the point where you've already lost.
Worth knowing as well: your ESP's internal health thresholds are not the same as the mailbox providers' enforcement lines. Klaviyo publishes its own account health bars, and its complaint threshold is far stricter than Google's. Those are a vendor's internal guardrails, not a provider's rule, and confusing the two means managing to the wrong number entirely.
How the number is measured, which changes what it means
Postmaster Tools reports your complaint rate, but there are two quirks in how it's calculated that make the figure less reassuring than it looks.
It counts only DKIM signed mail. And it measures against messages that actually reached an engaged recipient's inbox.
That second one is the sting. If a chunk of your mail is already being auto filtered to spam, fewer messages reach an inbox, so fewer complaints are possible, so your reported rate can look healthy while your delivery is falling apart.
A low complaint rate alongside falling open rates is not good news. It's often the same news, arriving in disguise.
Recovery takes far longer than the damage
This next part is the one that should change your behaviour.
Breach the threshold and you don't just get filtered. You lose access to Google's delivery mitigation route until your spam rate has come back down and stayed down for a sustained period. Only then can you ask for help, and there's no published timeline for actual inbox recovery after that. It depends on your wider sending history.
So a single bad send, or one badly judged campaign to a stale segment, buys you weeks of degraded performance on the sends that follow. During Q4, that's the whole quarter.
You do not fix deliverability reactively. There is no fixing it reactively. You either have headroom before the busy period or you don't.
What poor placement actually costs
Deliverability arguments fail because they're made in percentages and decided in pounds. So convert it.
A brand sending to 80,000 subscribers, four campaigns a month, earning £0.08 per email delivered. At 90% inbox placement that's roughly £23,000 a month from campaigns. Drop to 65% placement and the same sending earns about £16,600.
Nothing changed. Same list, same offers, same copy, same agency doing the same work. Six and a half grand a month gone, and every diagnostic anyone runs will point at creative, because the open rate fell and open rate is what gets looked at.
That's £78,000 a year sat inside a technical problem that a morning's work would have surfaced.
Run the same sum with your own numbers before you decide this is a job for next quarter.
Sunsetting, and the maths that makes it obvious
Every brand knows they should sunset unengaged subscribers. Almost nobody does it at the right scale, because cutting the list feels like cutting revenue.
Run the numbers instead of the feeling.
Say you've got 100,000 subscribers and 30,000 of them haven't opened or clicked in a year. Sending to all 100,000 at 65% inbox placement puts you in front of 65,000 people.
Cut the 30,000. Now you're sending to 70,000 at 90% placement, which is 63,000 people, and every one of them is somebody with a demonstrated interest in your brand. Near identical reach, dramatically better engagement signals, and complaint rate headroom going into your busiest period.
The dead weight isn't neutral. It's actively suppressing delivery to the people who do want to hear from you.
A workable policy: no opens, clicks, site activity or purchases in six months goes to a re engagement sequence. No response to that goes to suppression. Review quarterly. Suppress hard bounces permanently and automatically, because re sending to an address that has already hard bounced is about the clearest list hygiene signal you can send. As a working figure, most ESPs flag a bounce rate above 2% as an account health problem, and that's a sensible internal line to hold even though it isn't a mailbox provider's published threshold.
Shared IPs and who you're standing next to
Most brands on Klaviyo and similar platforms send from a shared IP pool. Your reputation is partly determined by the other senders in that pool, which is fine when the pool is well managed and less fine when it isn't.
The thing that actually protects you is domain reputation rather than IP reputation, and that one is entirely yours. Which is the argument for sending marketing mail from a dedicated subdomain rather than your root domain.
Two reasons, and the second one is the one that keeps business owners up at night.
First, it isolates your marketing reputation from your transactional mail. If a campaign goes badly, your order confirmations and password resets still arrive.
Second, it protects your root domain, which is what your invoices, your supplier emails and your customer service replies go out on. A brand that torches its root domain reputation with a bad send doesn't just lose campaign revenue. It starts having conversations about why the accounts team's emails are landing in spam.
If you're on your root domain right now, moving is a project rather than a switch. Subdomains need warming, and warming takes weeks. Which puts it firmly in the category of things to do in a quiet month rather than a panic.
The awkward Q4 conversation
Sunsetting before Black Friday feels insane. It's the correct call, and it's the recommendation your previous agency probably never made, because "we're cutting a third of your list six weeks before your biggest sale" is a difficult meeting and "your list grew again this month" is an easy one.
You're about to triple your sending volume into a period of maximum scrutiny. Do it on a list of people who want your emails, or do it on a list padded with 30,000 profiles who'll never open and might complain.
One of those options protects the sale. The other risks it to keep a vanity number intact.
Where to start this week
Check whether SPF, DKIM and DMARC are all passing and aligned on your sending domain. Confirm one click unsubscribe is present as a header. Open Postmaster Tools and look at your complaint rate against 0.1%, not 0.3%. Then look at what proportion of your list hasn't engaged in twelve months.
Four checks. Between them they'll tell you whether you're compliant, quietly degrading, or already in trouble.
Most brands find out they're the middle one.
And if you want a quick test of whoever currently runs your email: ask them what your Gmail complaint rate was last month and what your sunset policy is. Both answers should arrive without anyone going away to check.
If the reply mentions open rates, you've learned something.