Cold Email Deliverability: The 70% of Outbound Nobody Talks About

Deliverability is roughly 70 percent of whether outbound works, and it is the 70 percent nobody wants to talk about because it is boring infrastructure, not clever copy.

Every outbound post you have read is about subject lines, personalization, and the perfect five-line sequence. That is the 30 percent. The other 70 percent decides whether any of those words land in a primary inbox or rot in spam, and it is governed by DNS records, sending reputation, and volume discipline. Our outbound engine spends most of its engineering on the unglamorous layer, because the best email ever written has a zero percent reply rate from the spam folder.

Here is the full deliverability stack we run, in the order it matters: domains, authentication, inbox rotation, send-time behavior, and human reply handling. None of it is secret. Almost nobody does all five at once, which is exactly why it works.

Think of outbound as two multiplied numbers: the percentage of sends that reach a human, and the percentage of reached humans who reply. Teams pour effort into the second number and assume the first is 100 percent. It is not. A cold domain with no authentication and aggressive volume can land under half its sends in the inbox, which quietly halves every result before the copy gets a vote.

Google, Yahoo, and Microsoft now enforce bulk-sender rules that demand spam complaints under 0.3 percent and bounce rates under 2 percent. Cross either line and placement collapses across the whole sending domain, not just one campaign. This is why deliverability is a standing discipline, not a setup step you do once and forget.

Disposable by design

Domains: separate, and warmed for 21 days

Rule one: never send cold volume from your primary brand domain. One bad week torches the reputation your billing emails and renewals depend on. We buy separate sending domains, usually close variants of the brand, and treat them as disposable infrastructure that absorbs the risk.

Rule two: warm every new domain for at least 21 days before a single prospect send. Warming means low, steadily increasing volume to real, engaged inboxes so the domain builds a sending history that looks human. Start at 5 to 10 sends per inbox per day and ramp over three to four weeks. Skipping this is the single most common reason a campaign dies on day one, and there is no shortcut, because reputation is a function of time plus consistent behavior.

Authentication is how inbox providers verify the mail actually came from who it claims. Three records do the work, and all three are non-negotiable.

• SPF lists which servers are allowed to send for the domain, so a spoofed sender fails the check.
• DKIM cryptographically signs each message, proving it was not altered in transit and genuinely came from the domain.
• DMARC ties the two together and tells providers what to do on failure. We set the policy to reject, not the lazy 'none' most senders leave in place after warmup.

Fully authenticated domains see roughly 2.7x higher likelihood of inbox placement than unauthenticated ones, and since Microsoft tightened enforcement in May 2025, leaving DMARC on a permissive policy is an active liability for any mail bound for Outlook addresses. The records take an afternoon to configure and pay rent every day after.

No single inbox should carry meaningful cold volume, because per-inbox sending caps are what keep reputation healthy. We spread the load across many inboxes and let boring arithmetic keep every one of them under the line.

1. 01
   Set the inbox cap

   Hold each inbox to about 40 cold sends per day. Past that, engagement signals dip and providers start treating the inbox as a bulk sender.

2. 02
   Do the math for your volume

   A 10,000-send month is roughly 500 sends per business day. At 40 per inbox, that needs about 13 active inboxes with headroom, so we run 40 inboxes across 4 domains to stay comfortably under caps and survive any one inbox getting flagged.

3. 03
   Rotate and rest

   Sends rotate across the pool, and inboxes that show any reputation dip get pulled and rested while the rest carry the load. No single point of failure takes down the campaign.

Forty inboxes across four domains for a 10k-send month is not over-engineering, it is the floor for predictable placement. The same scoring and segmentation that powers our data scraping layer also tells us which inbox pool to assign to which audience, so high-value accounts never share a sending reputation with bulk volume.

Look human, reply fast

Send-time behavior and the human layer

Real people do not fire ten emails at exactly 9:00:00. Robotic, on-the-minute sending is a pattern filters learn fast, so our cadence tool randomizes send time per inbox per day, with natural gaps and business-hours weighting. The goal is to look like a busy human, not a cron job, because that is what the spam filters are actually scoring.

The last mile is human, and it is the part most automated stacks skip. Positive replies are triaged by a person within 20 minutes during business hours. The infrastructure protects the inbox, the human protects the brand, because nothing kills a hard-won reply faster than a delayed, obviously templated answer. A real, fast, specific response is where deliverability work finally turns into pipeline.

The best email ever written replies at zero percent from the spam folder. Win deliverability first, then argue about subject lines.

Shivam Bindal

Deliverability is not a state you reach, it is a number you watch drift. A pool that lands 95 percent in primary inboxes one week can slide to 60 percent the next after a bad list, a spike in complaints, or one inbox provider changing its filters. We treat the sending pool like production infrastructure, with three metrics on a dashboard the team reads every day.

• Inbox placement: the share of sends landing in primary, not spam or promotions. Below 90 percent on a warmed pool is a fire, not a footnote.
• Complaint rate: the percentage of recipients marking mail as spam. Hold it under 0.3 percent or the provider rules start working against you across the whole domain.
• Bounce rate: invalid or rejected addresses, kept under 2 percent through list verification before send. A spike here is usually a list problem, not a sending problem.

When a metric drifts, the fix is almost always to slow down, not to push harder. We pull the affected inboxes, drop volume, and let reputation recover before scaling back up. Outbound teams that treat a placement dip as a reason to send more are the ones who lose a whole domain in a week. The discipline is closer to running a database than writing copy, which is exactly why the copy-obsessed crowd keeps losing the 70 percent.

List quality feeds directly into all three numbers, which is why deliverability and targeting are the same problem wearing two hats. A clean, well-scored list bounces less, complains less, and engages more, all of which lift placement. Send to a stale scraped dump and no amount of warming saves you, because the recipients themselves are telling the provider your mail is unwanted.

The foundation layer

Where deliverability sits in the bigger picture

Outbound is one channel inside a portfolio, and its job is to start conversations cheaply at the top of a motion that direct mail and paid ads reinforce further down. None of that sequencing matters if the email never arrives, which is why deliverability is the first thing we fix and the last thing we stop watching. It is the foundation the whole B2B SaaS growth operating system sits on.

If you take one thing from this: stop treating deliverability as setup and start treating it as a daily discipline with its own metrics. Watch placement, complaint rate, and bounce rate the way you watch reply rate. The teams that win outbound in 2026 are not the ones with the cleverest copy. They are the ones whose mail simply shows up.