How Long Should a Cold Message Be? The Curve Is a U

“Keep it short” is the most common outbound advice. Checked against 37,375 messages it is only half right.

Contents
  1. Introduction
  2. The numbers
  3. Why the middle loses
  4. What this means in practice
  5. The caveat that belongs with this
  6. Next steps
  7. Goal

Introduction

"Keep it short" is the most common piece of outbound advice. We checked it against 37,375 sent messages and found something we had not expected: the curve is not a line. It is a U.

The numbers

Message lengthsentrepliesreply rate
under 200 characters6,6864676.98 %
200 to 399 characters5,0172795.56 %
400 to 599 characters12,2003813.12 %
600 to 799 characters10,4434033.86 %
800 to 999 characters1,840774.18 %
1,000 characters and up1,147786.80 %

Very short wins. Very long wins almost as much. And the middle, where most messages sit, loses clearly.

Why the middle loses

Our explanation: a message under 200 characters is recognisably a question. People read it to the end because it costs them nothing.

A message over 1,000 characters is recognisably work. If someone wrote that much, effort was obviously involved, and that creates obligation.

In between sits the range where a message looks like a template. Long enough to claim effort, short enough not to prove any. That is exactly where 22,643 of our messages sit, almost two thirds.

Good messages differ from bad ones in whether their length was a decision or an accident.

What this means in practice

Decide before you write which of the two messages you are writing:

  • The short question. Under 200 characters. One trigger, one question, no pitch. Works particularly well on LinkedIn.
  • The long case. Over 1,000 characters. Only for accounts you genuinely put work into. With a concrete reference, with numbers, with a thought the recipient has not had yet.

What you should not write is the message in between. It costs you the same send and returns half.

The caveat that belongs with this

The 1,000-character row rests on 1,147 messages. That holds up, but it is the thinnest row in the table. And long messages rarely go to arbitrary contacts, they go to selected ones. So part of the effect probably comes from the selection rather than the length.

We are saying so because the honest version is worth more than the catchy one.

Next steps

We are currently testing whether the effect holds when the long message is generated rather than written by hand. If it does, effort stops being an argument.

Goal

We want our software to pick the right length for each contact by itself. Happy selling, Axel

Find your customer before your competition does.

In real time. Signal-based. On autopilot.

Start for free  →
Keep reading