Sales Cadence: Design for Relevance, Not Persistence

Most sales cadence advice is a template. Day 1, email. Day 2, call. Day 4, LinkedIn. Day 7, email. Copy it into the sequencing tool, press start, and wonder why it underperforms. The template is not exactly wrong. It answers the wrong question: it tells you what to send without telling you why, and a cadence you cannot explain is a cadence you cannot fix.

A more useful definition: a sales cadence is a hypothesis about a buyer's attention. It is your best guess at how a particular kind of person, in a particular situation, notices outreach: how often you need to appear before you register, through which channels, and over how long before the moment passes. Seen that way, two things follow. There is no universal cadence, because different buyers have different attention. And every step has to justify itself against that hypothesis, instead of existing because the template had a slot.

This guide covers where the famous touch counts came from, the design rules that still work now that inboxes push back, four cadences built from four different buyer hypotheses, and how to measure a cadence without fooling yourself.

What a sales cadence is

A sales cadence (also called a sequence) is a planned series of contacts with a prospect or customer over a set period, across channels such as email, phone and social messages. Each step has a channel, a timing and a purpose. Some are automated, some become tasks for the rep, and the whole sequence ends either when the buyer responds or when it runs out of steps.

Cadences are usually run inside a sales engagement platform. We covered what those platforms do well, and what their dashboards cannot see, in sales engagement. This guide is about designing what goes into them.

The touch-count folklore

Almost every article on cadences quotes a number: it takes eight touches to reach a buyer, most sales happen after the fifth follow-up, most reps give up after one or two. These numbers are repeated so often that they feel like settled fact. Most of them are not.

When you try to trace them, they tend to lead to one of three places: a vendor analysing its own customers' data with methods that are not published, a buyer or seller survey asking people to estimate something they did not count, or a citation to an organisation or study that cannot be found at all. Some of the most widely shared follow-up statistics fall into the last group.

Even the honest versions have a deeper problem: they measure when replies arrive, not what caused them. If a reply comes after the eighth touch, the analysis credits the eighth touch. It cannot tell whether that buyer would have replied to three better touches, or whether the eight touches that produced one reply also produced several complaints and a dozen people who will now ignore your company. A long cadence will always find some late replies. That is not evidence that length is what worked.

So treat touch counts as folklore with a kernel of truth. The kernel: many buyers do not respond to the first message, often because they never really saw it, so a single attempt is rarely enough. The folklore: that a specific number of touches is a target worth hitting.

A long cadence will always find some late replies. That shows when replies arrive, not what caused them, and says nothing about what the other touches cost.

The design rules that still work

The environment cadences run in has changed. Inbox providers now penalise senders whose mail people mark as spam, and buyers have become very good at recognising automated sequences. A cadence designed for persistence alone burns the very attention it is trying to win. These rules hold up.

1. Every step has one job. Introduce yourself, give one useful idea, ask one specific question, share one relevant example, close the loop. If you cannot name a step's job, delete it. "Just bumping this to the top of your inbox" has no job.

2. Assume the previous step was never seen. A buyer who finally notices your fifth message is reading it cold. It has to make sense on its own, which means a new angle each time rather than a resend of the first email with "following up" on top.

3. Front-load, then widen. Contacts close together early give you a better chance of catching someone during a week when they are paying attention; wider gaps later keep you present without becoming noise. The exact spacing matters much less than the principle.

4. Use each channel for what it is good at. More on this below. The short version: a channel earns its place in a cadence by doing something the others cannot, not because the template has a slot for it.

5. Write the stop rules first. Decide when the cadence ends before it starts: an explicit no, a request to stop, a reply of any kind, a meeting booked, or the last step reached. Then make the stops automatic. Most of the damage a cadence does comes from the steps that should never have been sent.

6. Keep the volume per sender sensible. Send from properly authenticated domains and mailboxes, ramp new ones gradually, and do not let one person's sequences send at a rate that looks like a bulk campaign. The details belong to whoever runs your email infrastructure, but the cadence design has to respect them.

Four cadences, four buyers

The value of the hypothesis framing is that it produces very different cadences for different buyers. The structures below are examples of shape, not benchmarks; the timings are reasonable starting points to test against your own results.

1. Cold outbound

The hypothesis: this person has never heard of you, has not connected any current problem to what you sell, and will not see most of what you send. You are starting from no attention at all.

An example shape, over about three weeks:

• Day 1: Call, and leave a short voicemail. Email the same day, referring to something specific about their company.
• Day 2: View their profile and send a connection request without a pitch.
• Day 4: Call at a different time of day.
• Day 6: Email with one useful idea for someone in their role, not a feature list.
• Day 10: Call, plus a short social message referring to something they posted or their company announced.
• Day 14: Email from a completely different angle, assuming nothing earlier was read.
• Day 19: A short close-the-loop email that makes it easy to say "not now".
• Afterwards: stop the sequence. Revisit only if something genuinely changes at their company.

Why it looks like this: the early steps are close together because most will miss. The angles change because each step is read cold. The calls matter because a conversation is the only touch that can turn into a meeting on the spot, so the cadence is partly a way of earning a pickup. Openers for those calls are in 75 B2B cold call hooks, and short emails that still say something in 100 sales emails under three lines.

2. Expansion inside an existing customer

The hypothesis: the opposite of cold. They know you, and you hope they trust you. Attention is not the problem; relevance and timing are. Over-contacting a happy customer with a cold-outbound sequence makes them feel like a target.

The shape: fewer contacts, each tied to a real reason rather than a clock.

• Open on an event: a usage milestone, a new leader joining, a team growing. Lead with what you noticed, not with an ask.
• Recap what they have already gained: something concrete, such as "your team ran 400 practice calls last quarter" (an invented example), so the expansion builds on proven value.
• Ask your contact for an introduction to the adjacent team, rather than contacting that team cold.
• Make a specific case for the new team, in their terms, not a copy of the original proposal.

There is no fixed calendar here, and that is the point. Expansion runs on the customer's calendar: budget cycles, renewals, reorganisations.

3. After a demo

The hypothesis: interest is highest right after a good demo and fades quickly. Meanwhile the person you met now has to sell your product internally to people who were not there. The cadence races against forgetting and, just as much, equips your contact to make the case without you.

The shape:

• Within a day: a recap of what they said mattered, answers to open questions, and something short they can forward to colleagues.
• Two or three days later: a proposed next step with two specific times.
• About a week later: one relevant example or piece of evidence, also easy to forward.
• Then a call: "who else needs to be comfortable with this?"
• Then a shared plan: move from follow-ups to a mutual action plan agreed with the buyer.

The most common failure here is "send me some information and I'll take a look", which ends many promising deals quietly. It has its own page: send me more information.

4. A customer at risk

The hypothesis: usage has dropped, a renewal looks uncertain, or they have gone quiet. They feel the product is not delivering, and they may already feel over-sold. A hard retention push confirms their suspicion that they are a number.

The shape: low pressure, diagnostic, with a real way out.

• Open with curiosity: "I noticed your team's usage dropped; wanted to check what's changed", not a save attempt.
• Find the real reason: price, missing value, a change inside their company, a product problem. The reason decides everything after it.
• Offer something that fits their situation: a smaller plan, a pause, or a specific fix, presented as fitting their reality rather than as a discount to keep them.
• If they are leaving, make it easy: a customer who leaves well sometimes comes back. One who feels trapped does not.

Act at the first sign of risk, not a month before renewal, but keep each contact gentle. Pressure tends to accelerate the decision you are trying to prevent.

What each channel is for

A multi-channel cadence is not the same message posted in three places. Each channel does a different job.

• Phone is for conversation, and the only channel where a meeting can be agreed on the spot. Even a voicemail helps the email that follows it.
• Email is for substance and for forwarding. It respects a busy buyer's time and leaves something they can act on, which is also why it is easy to ignore.
• Social messages are for familiarity. A relevant comment or a connection without a pitch makes your name recognisable before the call. Used as a pitch channel, they mostly annoy.
• Video or text messages, where appropriate, are pattern breaks: worth using occasionally, later in a cadence, when the usual channels have gone stale.

Personalisation that scales

Every cadence sits between two failures. Fully automated, identical messages are efficient and ignored, because buyers recognise them instantly. Fully handcrafted ones work well and cannot be sustained across a real prospecting list.

The practical answer is to be deliberate about where personalisation goes. Personalise the opening of each touch and standardise the rest. The first sentence decides whether the message is read, so spend your research there: one specific, accurate observation about the person or their company. Keep the useful idea and the ask consistent, and improve them over time.

A tight buyer hypothesis makes this faster. If you know what a VP of operations at a mid-sized logistics company is likely to care about this quarter, the personal first line takes a minute to write, not ten.

Measuring a cadence honestly

Step-by-step reply rates are the easiest numbers to look at and the most misleading, for the reason in the folklore section: later steps get credit for replies they may not have caused. Measure the cadence as a whole.

• Outcome per hundred prospects. Of every hundred people who entered the cadence, how many meetings were held, not just booked? This is the number that matters.
• Guardrails beside it. Negative replies, unsubscribes and spam complaints per hundred. A change that raises meetings and also raises complaints is not an improvement.
• Change one thing at a time. Shorten the cadence, change the second email, add a call. Compare against the previous version on the same kind of list.
• Test a shorter version. Run a cut-down cadence on part of the list. If it produces nearly as many meetings with fewer complaints, the extra steps were costing more than they earned.

Running it every day

A good cadence run occasionally is worth less than an average one run consistently. Jeb Blount makes this point in Fanatical Prospecting with what he calls the 30-day rule: the prospecting done in any month shows up in the pipeline over the following quarter. Stop running cadences for a month because you are busy closing, and the gap appears three months later, when it is too late to fill.

This is where a sequencing tool earns its keep. It removes the daily judgement call about whether prospect X is worth another touch, so the cadence happens whether or not anyone feels like it. That only helps if the cadence is worth running, which is what the rules above are for.

Where our tool fits

We build an AI practice tool, so read this with that in mind.

A cadence only earns the conversation. The rep still has to win it: the first thirty seconds when someone picks up a cold call, the next-step ask after a demo, the diagnostic conversation with an at-risk customer. SalesArmor lets reps rehearse those calls against a realistic AI buyer and get scored afterwards, so the conversations a good cadence produces are not lost in the first minute. See cold call practice.

Common questions about sales cadences

What is a sales cadence? A sales cadence is a planned series of contacts with a prospect or customer over a set period, across channels such as email, phone and social messages. Each step has a channel, timing and purpose, and the sequence ends when the buyer responds or the steps run out.

How many touches should a sales cadence have? There is no reliable universal number. Widely quoted figures usually come from vendor data or surveys that cannot be checked, and they measure when replies arrived rather than what caused them. Design each step to do one job, set clear stop rules, and test a shorter version against your current one.

What is a good sales cadence for cold outreach? A few weeks of contacts, closer together at the start and further apart later, mixing calls, short emails and social messages. Each email should take a new angle and make sense on its own, and the cadence should end with an easy way for the buyer to say not now.

What is the difference between a sales cadence and a sales sequence? In practice, very little. Both describe a planned series of outreach steps. Some teams use cadence for the overall rhythm and sequence for the specific steps configured in a tool, but most use the words interchangeably.

Should a sales cadence start with a call or an email? Either can work, and the best choice depends on the buyer. Calling first gives you the chance of an immediate conversation, and a voicemail can make the email that follows more likely to be read. Test both on your own lists rather than relying on a general rule.

How do you personalise a sales cadence at scale? Personalise the first sentence of each touch with one specific, accurate observation about the person or company, and keep the rest of the message consistent. A clear buyer hypothesis makes that first line quick to write. Check AI-generated openers before they send, because a wrong detail is worse than none.

How do you measure whether a sales cadence works? Measure the whole cadence rather than individual steps: meetings held per hundred prospects, alongside negative replies, unsubscribes and spam complaints per hundred. Change one thing at a time and compare versions on similar lists.

When should a sales cadence stop? When the buyer replies, books a meeting, says no, asks to stop, or the last step is reached. Decide these rules before the cadence starts and make them automatic, then revisit the prospect only if something meaningful changes at their company.

A note on sources

This guide deliberately publishes no figures for how many touches it takes to book a meeting, how quickly reps give up, which channel to start with or what reply rates to expect. The widely circulated versions of these numbers come from vendors analysing their own customers, from surveys of estimates rather than counts, or from citations that cannot be traced to any study. All of them measure when replies arrive, not what caused them.

The example cadences are shapes to test, not benchmarks. The 30-day rule is Jeb Blount's, from Fanatical Prospecting, paraphrased here. The framing of a cadence as a hypothesis about buyer attention, the design rules and the measurement approach are ours.