The State of Cold Email 2026

Most cold email reports fall apart on definitions, so here are ours. A reply is any direct response from a prospect, counted once per prospect, with out-of-office and automatic replies removed. A positive reply is an expression of interest. We do not report open rate, and we explain why below. Every number is reply-based unless labelled otherwise.

The data covers 5,000+ campaigns from 400+ accounts over six months of 2026, 40M+ emails sent to 6M+ prospects. Crucially, 100+ of those accounts are agencies, teams that run cold email every day, across many clients and verticals, at volumes an in-house team rarely touches. That cohort is the point of this report. Where an in-house team runs one motion, an agency has run a hundred, killed the losers, and kept what works.

So we split the data two ways throughout: the broad average, and the agency cohort. We treat the agencies as the clearest available read on what a well-run cold email operation actually achieves in 2026, and the gap between them and the average is the story.

A reply rate is a vanity number if it doesn't connect to pipeline. Agencies get paid on meetings, so they track the whole funnel, and so should any leader judging a program. Positive replies are a fraction of total replies, most cold responses are “not interested,” “unsubscribe,” or “wrong person,” and only a portion of positives become meetings.

Roughly 1 in 5 replies is a positive one, and about half of positives turn into a booked meeting for a well-run agency. Leaders who manage only to reply rate optimise the top of a funnel that leaks everywhere below it.

The largest, most fixable cause of lost cold email is missing authentication. Fully authenticated domains, SPF, DKIM and DMARC on an enforcement policy, are around 2.7x more likely to land in the inbox, and full authentication makes 85-95% placement achievable. Yet only about 7.6% of domains actually enforce DMARC.³

That gap is the opportunity. Age compounds it: warming a domain for six to twelve months before critical campaigns can lift placement by up to 30 points versus a new one. And the providers now enforce at the door, Gmail rejecting non-compliant mail at the SMTP level in late 2025, and Microsoft rejecting non-compliant bulk senders outright since May 2025.

Two truths sit under every reply rate. First, B2B data rots: contact data decays around 2.1% a month, 22-30% of a list per year, and most providers ship data only about 50% accurate against the 97% the best sources hit.⁵ On a decaying list a healthy bounce rate under 2% slips past 5% quietly, and every point of bounce drags reputation down with it.

Second, even a clean list loses to a well-timed one. Teams moving from static lists to signal-based targeting, funding, exec hires, tech installs, hiring posts, report response lifts of 3-5x and sales cycles 30-50% shorter.⁶ The honest caveat, and agencies say this loudest: signal-data quality is wildly uneven, and more signals is not the answer. A smaller list hit on a real trigger beats a huge list hit at random.

The second email is the most under-worked asset in the average sequence: only 58% of replies come from the first send, the other 42% arrive later.⁸But 2026 flipped the old “just add touches” advice. As inboxes tighten, aggressive sequences now damage sender reputation faster than they used to, and the reputational cost shows up as worse deliverability on the next campaign.

Agency data shows returns bending after the third or fourth touch. The discipline isn't more follow-ups, it's knowing exactly where the curve turns negative and stopping there.