Sales Cadence: The Six Variables That Define One, and Where Cadences Fit

Two of the largest sales engagement platforms sell the same object under different names. Outreach calls it a Sequence and defines it as "a series of touch points built to drive deeper engagement with a Prospect", offering five step types: auto email, manual email, phone call, generic task and LinkedIn task. Salesloft calls the same object a Cadence and markets it as multi-channel, sorting steps by channel, time zone and account. Buyers move between the two tools and keep using whichever word they learned first, which is why a conversation about cadences frequently involves two people describing identical machinery and disagreeing about it.

So it is worth pinning the term down before arguing about how many touches belong in one. This page defines what a cadence is, sets out how multi-channel cadences are actually structured, states our own position on cold outbound once, and gets specific about where a cadence is genuinely the right tool.

What a cadence is

A cadence is an ordered, scheduled set of touchpoints aimed at one contact, running across one or more channels, with defined spacing and a defined exit condition.

Six things have to be specified before a cadence exists as anything more than an intention.

What a cadence has to specify

• ✓Yes: Channels: which of email, phone, LinkedIn, SMS or physical mail are in play
• ✓Yes: Touches: how many of each, and in what order
• ✓Yes: Spacing: the delay between touches, usually counted in business days
• ✓Yes: Automation: which steps fire automatically and which create a task for a person
• ✓Yes: Exit conditions: what removes a contact, and whether the cadence has an end date
• ✓Yes: Owner: the named person whose task queue the manual steps land in

The exit condition is the variable that changes the character of the whole thing. Almost every cadence exits on a reply, which means each successive step is delivered only to the people who received the previous one and chose not to answer. Some tools extend the exit to a booking or a click. A cadence with no end date is a subscription the contact never agreed to.

Two adjacent words get used interchangeably and mean different things.

Cadence or sequenceOwned by sales

• Targets one named contact at a time
• Ordered steps with fixed spacing
• Mixes automated sends and human tasks
• Exits on reply, booking or completion
• Measured per step and per contact

CampaignOwned by outbound or marketing

• Targets a list, sent as a batch
• Measured in aggregate against the list
• Usually a single channel
• Exits are list-level: reply, bounce, unsubscribe
• Measured per campaign, against other campaigns

Drip or nurtureOwned by marketing

• Targets subscribers who opted in
• Triggered by time or by behaviour
• Almost entirely automated
• Exits on unsubscribe or a goal event
• Measured on engagement over months

The distinction that matters commercially is permission. A nurture programme runs on people who asked. A cadence to a stranger runs on people who did not, and the number of touches in it is a decision about how much of somebody else's attention you are willing to take without an invitation.

How multi-channel cadences are structured

The standard shape is a mix of automated email and manual tasks, spread over two to four weeks, alternating channels so that no single channel carries consecutive touches.

Three structural choices do most of the work.

Spacing. Delays are counted in business days, and a cadence built on calendar days will fire its second touch on a Sunday within a week of going live. Spacing usually widens as the cadence progresses, on the reasoning that early touches establish presence and later ones are checking for a change in circumstance.

The manual to automatic ratio. This is the constraint nobody costs properly. Automated email steps scale to whatever your sending infrastructure supports. Manual steps do not, because each one becomes a task in a queue owned by a human being. A twelve-touch cadence with five manual steps, running against two hundred active contacts, generates a thousand tasks over the life of the cadence. If that arrives in one person's queue at a rate they cannot clear, the manual steps get skipped silently and the cadence you are measuring is not the cadence you designed. Work out the task load before the touch count.

Channel sequencing. Multi-channel cadences alternate deliberately so that a contact who ignores one channel encounters a different one. The complication is that channels have different social rules. An email in an inbox is a normal intrusion. A second LinkedIn message lands directly under the first one in the same thread, so it reads as a nudge whatever the campaign structure calls it, which is why we do not run LinkedIn re-touches on people who did not reply. Structure and templates for the multi-channel case, where it is appropriate, are in multi-channel templates.

Our position, stated once

We run one message per campaign for cold outbound. No thread replies, no bumps, ever.

The reasoning is structural rather than a claim that follow-ups produce no replies. Because the exit condition is a reply, every touch after the first is delivered exclusively to people who already saw a message from you and declined to answer it, which is the population most likely to file a spam complaint. The reply lift from those touches is measured inside the cadence, and the reputation cost is paid by the sending domain across every campaign running on it, so the unit of measurement and the unit of damage are different objects and no per-cadence report can show you the trade. We would rather not spend a budget we cannot see the meter for. The full argument, including the deliverability thresholds involved and the costs the position imposes on us, is in email sequence software.

What replaces the follow-up is a new campaign built on a genuinely different angle. The same person is reachable, the message arrives as its own message with its own subject line, and it gets judged on its own premise rather than being read as a reminder about the one they ignored. That requires suppression discipline: repliers, bounces and unsubscribes are excluded permanently, a lead pending in one active campaign is not enrolled in another, and a negative reply suppresses that person across the whole workspace. Cadence tooling handles a lot of that automatically, which is one of the honest costs of running the other way.

Where cadences are genuinely the right tool

The argument above depends on one condition: the recipient never asked to hear from you. Where they did, silence stops being an answer and multiple touches are simply competent follow-through.

Warm inbound. Somebody filled in a form, requested a demo, started a trial or asked a question. They initiated contact, so non-response is far more likely to be a busy week than a rejection. Not following up is the actual error here, and speed matters more than touch count, because the relevance window is short and closes fast.

Existing customers. Renewals, onboarding milestones, usage prompts, product changes, support follow-through. There is a contract and an expectation of contact.

Opted-in nurture. Somebody subscribed and can unsubscribe. The permission is explicit and the tooling for honouring it is mature.

Scheduled commitments. Confirming, reminding and rescheduling around a meeting the person agreed to. Nobody has ever objected to a reminder about a call they booked.

Here is what a warm inbound cadence looks like when it is built for speed rather than persistence. The point of the shape is that the useful touches are clustered at the front.

1. Within the hourAutomated acknowledgement

   Confirms receipt and offers a booking link. Fires on form submission with no human in the loop.

2. Same dayManual email from the rep

   References what they actually asked for. This is the touch that most often gets the reply, and it is the one most likely to be missed if the queue is overloaded.

3. Day 1Phone attempt

   While the request is still recent enough that they remember making it.

4. Day 3Second call plus a short email

   Different time of day from the first attempt. Same thread is fine here, because they asked.

5. Day 7Value touch

   Something relevant to the question they raised rather than a reminder that you emailed them.

6. Day 14Close the loop

   A short message stating you will stop and how to restart. Then the cadence ends and the contact returns to marketing.

Two details in that shape are worth copying. The cadence has an end date, so nobody stays in it indefinitely. And the last touch says explicitly that the follow-up is stopping, which is a courtesy and also reliably produces replies from people who genuinely were busy.

The handoff into a cadence like this is where most warm follow-up quietly fails, and the fix is a written response time and a minimum attempt count agreed between the functions rather than a better cadence design. That agreement is part of the same document as the lead definitions, covered in MQL versus SQL.

Cadence hygiene that applies whatever you run

Four rules travel across every situation above.

One active cadence per contact. Two cadences running against the same person is the most common cause of somebody receiving two unrelated messages from your company in one morning, and the tooling will not stop it by default.

Suppression is global. An unsubscribe, a hard bounce or a clear no should remove the person everywhere, not from the cadence that produced it. Check how your platform scopes this, because the default is often per-cadence. The scoping behaviour varies between tools and is worth checking before you buy one, which is one of the axes in sales engagement platforms.

Manual steps need a named owner and a realistic queue. A cadence whose manual steps are being skipped is reporting the performance of a different cadence.

Per-step reporting measures the step, not the programme. Attribution to the step that produced a reply is accurate as far as it goes. What it cannot show is anything happening outside the cadence, including inbox placement across the sending domain, which is exactly the measurement gap that shapes our own position above.

The short version

A cadence is an ordered, scheduled set of touchpoints to one contact across one or more channels, with defined spacing and a defined exit condition. Outreach calls the object a Sequence, Salesloft calls it a Cadence, and they are the same machinery. Six variables define one: channels, touch count, spacing, which steps are automated, exit conditions, and the person whose task queue the manual steps land in.

Multi-channel cadences alternate channels over two to four weeks with widening gaps. The constraint that gets missed is the manual step load, because task-based steps scale with human hours rather than with sending capacity, and an overloaded queue turns skipped steps into a cadence you are not actually running.

For cold outbound we run one message per campaign with no bumps, because every touch after the first is delivered only to people who already declined to answer, and the reply lift is measured inside the cadence while the reputation cost is paid across the whole sending domain. What replaces the follow-up is a new campaign with a genuinely different angle rather than a nudge in the same thread.

Cadences are the right tool wherever the person initiated contact, has a relationship with you, or opted in: warm inbound, customer lifecycle, opted-in nurture and confirming scheduled commitments. Give them an end date, a global suppression rule, one cadence per contact and a named owner for the manual steps.

Our clients get the cold half run this way, one message per campaign, with the angle carrying the work and billing tied to attended meetings that meet criteria agreed in writing before launch. You can see what a campaign would look like for your market.

Vendor terminology and product descriptions verified against the vendors' own pages in August 2026. Verify current terms with the vendor before relying on them.

Sources: Outreach sequence overview, Salesloft cadence automation