Sales Cadence

There was a time when one message with no reply was the end of it. Send a hundred emails, get two replies, and the other ninety-eight were filed as uninterested. The rep went off to build the next hundred.

Take the deals that did reply apart later, though, and an odd pattern shows up. Most of the ones that closed weren't the ones that answered the first message. They answered on the third or the fourth. The day the first email arrived the contact was closing their quarter; the day the second arrived they were on vacation; the day the third arrived their manager happened to ask them to look into a new tool. Unread and uninterested look like exactly the same silence.

A sales cadence exists to stop that silence from being read as a no. How many times to reach out, how many days apart, which channels to mix — all of it goes on paper in advance. The reason for writing it down isn't memory, it's feeling. Sending a fourth message to someone who hasn't replied is unpleasant for anyone, so the point is to have made that decision earlier, when it wasn't unpleasant yet.

Building one takes five steps.

1. Grade the list. Thirty companies that match every criterion and three hundred that match half don't get the same cadence. The first group gets written one at a time; the second gets a template, and the effort you save on the second group is what pays for the first.
2. Fix the count and the span. Eight to twelve touches across two to three weeks is the common shape. Crowd them into the first three days and you earn irritation; do two in two weeks and you get forgotten.
3. Mix channels. Alternate between a cold email, the first email sent with no prior contact, a cold call, the first phone call with no prior contact, and a message on LinkedIn. Stay on one channel and twelve touches equal zero for anyone who doesn't watch that channel.
4. Make each touch say something different. If day one named the problem, day five brings a comparable company's result and day nine asks one short question. Resending the same sentences doesn't count toward the number.
5. Set an exit condition and a record. The last touch says you're closing the file, and then you actually close it. Record which touch produced the reply. That record is the material for revising the cadence.

Built out, one looks like this: email day 1, call day 3, LinkedIn day 5, email day 8, call day 12, final email day 16. Six touches across three weeks. Nothing about those specific dates is magic — the value is that the fourth and fifth touches exist at all, since those are the ones an unwritten cadence quietly drops.

Read the numbers by touch position, not in aggregate. If fourteen of twenty-two replies arrived between the third and the fifth touch, a cadence that stops at five was missing all fourteen of them.

It fits when the list is large and the customers resemble each other. If you can ask three hundred companies between fifty and two hundred people the same question, sharpening one cadence and running it across all of them produces far more meetings than writing each one from scratch.

Three situations are a poor fit.

1. Account-based marketing (ABM), where you pick a few companies and go deep. With eight targets, approaching each differently beats fixing an order. The cadence survives only as a skeleton while the content gets rewritten each time.
2. Someone who has already shown interest. Running a twelve-touch cadence at a person who opened the pricing page three times is knocking at a door that's already open. One reply within the day works far better.
3. Markets where the decision cycle is set by institution. A customer whose budget is allocated once a year has nowhere to put a three-week sequence. What's needed there isn't more touches but knowing when the next budget window opens.

The limits are real too. Once the order is written down, people move toward completing the order, and having sent all twelve starts to look like a result in itself. And mixing channels doesn't reduce the total volume the other person receives. When three companies each run a twelve-touch cadence at the same person, that person gets thirty-six messages in a month. What produces a reply in that situation isn't the count — it's the first two lines, which is the part a cadence can't schedule for you.

"This list runs on a five-touch cadence, and the thirty best-fit accounts go to eight."

Different sequences for different grades of list.

"We were stopping at three and the replies cluster at four and five. Adding two."

Revising the cadence from the record.

"The cadence is running but the copy is identical every time. That's one touch, not twelve."

Pointing out that only the count got filled in.

At Pagedeck, Ravi and Sofia build cadences together. Ravi writes several versions of the first two lines; Sofia runs them and logs the reply rate by touch position. Last quarter, switching the third touch from email to a phone call raised replies on its own — half the list turned out to be people who barely read email.

The record lives in one place: which list ran which length of cadence, and which touch produced the answer. Without it the only basis for revising a cadence is instinct, and instinct pushes in one direction only, which is adding touches.

A common belief is that enough volume eventually produces an answer, and learning that replies cluster at the third and fifth touch reinforces it. But that statistic came from sequences where each touch said something new. Send the same email twelve times and the reply rate falls with every touch, and a spam complaint blocks the messages you send afterward to other people at the same company. What gets increased isn't the count but the number of distinct angles.

The other mistake is treating a cadence as a tool configuration. Enter the intervals into an automated sender and it feels finished. What actually has to be decided here isn't the intervals but what each touch says. A cadence with only the intervals filled in amounts to saying the same thing six times, and after that the list can't be used again. A list that has been worn out once doesn't recover its reply rate after six months of rest.