Cold email deliverability: the levers that matter, ranked

Most of what's written about cold email deliverability is about the message. Most of what actually decides whether your email reaches the inbox is about the sender and the list. That gap explains why teams spend a week rewriting subject lines and see nothing change.

In rough order of importance: authentication comes first and is pass-or-fail. Then complaint rate and how recipients engage, both of which come down to who you sent to. Domain age and how gradually you build up volume matter moderately. Subject-line wording and spintax barely register.

What decides inbox placement, ranked

That ranking is our reading of the published guidance rather than an official formula (nobody outside the mailbox providers has one). But the guidance points one way. The two levers at the bottom get most of the attention, and they're the two a copywriter can help with. The three at the top have nothing to do with the message. Complaint rate and engagement are both measures of whether you picked the right people to write to, which makes deliverability largely a prospecting problem in a technical costume.

Authentication: SPF, DKIM and DMARC

These three DNS records work as a gate, not a dial.

SPF lists which servers are allowed to send email for your domain. DKIM adds a signature so the receiver can check the message wasn't altered. DMARC tells receivers what to do when those checks fail, and sends you reports about it.

Gmail's rules are the clearest statement of how much this matters. Since February 2024, Google has required every sender to use SPF or DKIM, and anyone sending more than 5,000 messages a day to Gmail addresses needs SPF, DKIM and DMARC all in place. Yahoo's sender requirements are similar. Setting them up usually takes under an hour. Afterwards, send a real test email and check the headers, or use a public checker, rather than assuming your provider did it correctly.

Build volume up slowly

A common approach is to start a new domain at around ten emails a day and build up gradually over six to eight weeks, which is roughly what the chart shows. There's no official schedule; the principle is simply that a brand-new domain sending at full volume looks exactly like a spammer.

This part is mostly waiting, which is why people skip it, and skipping it tends to backfire. A domain pushed to full volume on day one can end up filtered within weeks, and then the replacement domain has to start again from zero.

As for warm-up tools that trade emails between accounts in a network to fake engagement, we'd be cautious. They manufacture exactly the kind of signal mailbox providers are trying to measure honestly, and the published guidance is all about real recipients' reactions. A slow ramp with real, relevant emails is the part that holds up.

Send from a domain you can afford to lose

Don't send cold email from your main domain.

A reputation problem on the domain that also sends your invoices, password resets and support replies is a much bigger problem than a stalled campaign. Buy a similar domain, set up authentication properly, point it at your website, and treat it as replaceable.

If it gets burned, you retire it and start another. That's a normal cost of outbound, and a very abnormal cost of running your company's email.

What does spam complaint rate measure?

A complaint is someone clicking "report spam". Mailbox providers take it seriously, and the bar is lower than most people assume: Gmail asks senders to keep the spam rate reported in Postmaster Tools below 0.3%. That's three complaints per thousand emails.

The way to keep complaints down isn't clever wording. It's the list. People report email as spam when it's irrelevant to them, and relevance is decided before you write a word, when you choose which accounts are worth contacting. A team sending six well-researched emails a day has a deliverability strategy, whether or not they call it one.

A checklist before you send

The opt-out row isn't just good manners. In the US, the FTC's CAN-SPAM guidance requires commercial email, cold email included, to give recipients a clear way to opt out and to honour it promptly, along with an accurate sender and a physical postal address. Other countries have their own rules, some of them stricter.

The last row catches people out. Sending from an address nobody reads means real replies go unanswered, and that's a wasted conversation at best.

What should you do if your cold emails land in spam?

Assume the domain is the problem rather than the message.

We'd pause sending from it for a couple of weeks. Carrying on at lower volume from a domain with a complaint history mostly prolongs the problem, because whatever got you filtered is still happening.

Then check authentication properly, on a real email, rather than trusting a settings page that says everything's green. A DKIM record that was never published correctly is an easy thing to miss.

Then be honest about whether the list caused it. If the last batch was bought and sent unfiltered, the domain may not recover in any useful timeframe, and the better move is a new domain plus a different way of choosing who to write to. Warming up a new domain and pointing it at the same list just repeats the problem more slowly.

What deliverability work won't fix

A bad list. No technical setup rescues email sent to people with no reason to hear from you, and the complaint rate catches up with you eventually.

Volume as a strategy. Everything above helps a reasonable volume arrive reliably. None of it makes an unreasonable volume safe, and adding more sends doesn't add more replies at the same rate.

Open-rate worries. Apple's Mail Privacy Protection preloads images, tracking pixels included, even for email nobody opens, and some corporate security filters open links to scan them. So opens can be inflated by software. Judge your emails by replies instead.

And the fixes people reach for first: rewriting the copy, running it through a spam-word checker, or adding spintax. All three work on the bottom of that first chart and leave the top untouched.

The deliverability work that actually works is upstream and unglamorous: send to fewer people, chosen for reasons you could defend. Revtive's prospecting agent is built around that constraint rather than around volume, and it never sends anything on its own.