What 287 Cold Email Campaigns Actually Show About B2B Outreach

Since 2019, Clique Outreach has run cold email across twelve markets. Stripped to campaigns that contacted at least 200 people, that is 287 campaigns, 534,652 people contacted and 1,170 replies that asked for pricing, a call, samples or a next step.

This article was rebuilt in September 2026. Every figure in it was re-read straight off the sending platform, and several tables that used to sit here are gone, because the data did not support them. There is a section near the bottom saying exactly what we could not measure and why. The methodology is at our benchmark methodology page and the campaign data is in the Cold Email Benchmarks tool.

The Spread Matters More Than the Average

The single most useful thing in this dataset is not the average. It is how far apart the campaigns are. Same architecture, same infrastructure, same people writing the copy, and the results differ by more than ten times.

A sixteen fold gap between the bottom quarter and the top tenth is not explained by send times or word counts. In every case we have looked at, the campaigns at the bottom had a list that was too broad. The ones at the top were built around a signal that told us which week to write and what to say. That is the whole game, and everything else is rounding.

The Follow-Up Sequence

A frequent question is whether follow-ups actually move the needle or just annoy people. This one we can answer. Below is the reply split across our four largest campaigns, 134,000 people contacted and 1,854 replies between them.

So follow-ups recover about a quarter of all replies, consistently between 22 and 26 percent on every campaign we measured. That is worth having and it is not worth building a nine step sequence for. The second follow-up already produces a fifth of what the first one does. We run two or three steps and stop, and the sequence stops the moment somebody replies.

If you take one number from this article, take that one. Sending once leaves about a quarter of your interested replies uncollected. Sending six times does not get you a bigger share, it gets you complaints.

Bounces and Warmup

Across those 287 campaigns, 5.2 percent of emails sent hard bounced. That is the cost of building lists from providers rather than from people who handed you their address, and it is the number to watch, because a bounce rate climbing past single digits is the first sign a domain is in trouble.

Every programme we run gets a three week warmup before volume, and we do not bill for it. We are not going to put a percentage on what that buys, because we have never run the controlled version of the experiment, and inventing a figure is how the last version of this article went wrong. What we can say is that we have never shortened it and never wanted to.

What We Cannot Tell You

An earlier version of this article had tables on all three of these. They are gone. Here is why, because the reasoning is more useful than the numbers were.

Best day and time to send. We cannot measure it. Every campaign we have ever run sends Monday to Friday inside working hours, on 98 of the 100 campaigns we checked. There is no variation in the data, so there is nothing to compare. Anyone showing you a day of week table needs to have deliberately sent on Saturdays to have earned it.

Open rates and subject line length. We cannot measure these either, on purpose. Open tracking means loading a pixel from every email, which is one of the things that gets a domain flagged, so we have never switched it on. Every campaign in this dataset records zero opens. Any agency quoting you an open rate after Apple Mail Privacy Protection is quoting you a number their own tooling made up.

The ideal email length. We checked this properly, across 45 campaigns of a thousand people or more, and there is no clean curve in it. Emails under 60 words and emails of 90 to 120 words both did well; the 60 to 89 band did worst; the ordering goes up and down. Only the 90 to 120 band has enough people behind it to read at all. So we will tell you what we do, which is land between 90 and 120 words, and we will not pretend we have proved it is optimal.

What This Means for Your Program

Fix the list before you touch the copy. The gap between our worst and best campaigns is a list problem every time. A signal that tells you which week to write beats any amount of rewriting.

Run two or three touches, then stop. Follow-ups recover about a quarter of replies. A fourth and fifth touch collect complaints, not conversations.

Judge the programme on replies a month, not on a percentage. Rates at this volume are fractions of one percent and they sound alarming out of context. Across 72 documented months our median is 29 interested replies a month and the best single month was 128. That is the number that means something.

Watch the bounce rate. Ours sits at 5.2 percent. If yours is climbing toward double figures, stop sending and fix the list, because the domain is next.

Be suspicious of tidy benchmark tables, including ours. The old version of this page had six of them and most were not real. If a figure does not come with a denominator and a date, it is decoration.