Two teams can run the identical sequence and get results that look nothing alike, because the sequence is the set of touches and the cadence is their timing. The words get all the attention. The timing does more of the work.
Four decisions define a cadence. Make them deliberately or inherit them from whichever template shipped with the tool.
Decision 1: how many touches
Not a number you copy. A number you can actually sustain at your account volume.
The arithmetic that constrains it: touches per account, multiplied by accounts in flight, divided by working days, must land under what a rep can do in a day alongside everything else. A twenty-touch cadence across 150 live accounts is 3,000 touches. It will not happen, so it will silently degrade into whichever touches are easiest, which is email.
Decide the number from capacity, then make each touch earn its slot.
Decision 2: over what window
The window matters more than the count, and it is the decision teams skip.
The same twelve touches over two weeks and over eight weeks are different plays. Compressed says this is time-sensitive and risks reading as pressure. Extended says this is a standing offer and risks being forgotten between touches.
Tie the window to the signal that put the account in the cadence. A funding announcement has a short window. A slow-burn fit signal has a long one. If the account entered with no signal at all, fix that before tuning the window; see the sales development process.
Decision 3: which channels, in what order
Most cadences are email with occasional calls attached. That is a channel preference, not a design.
Two rules:
- Front-load the calls. Attention is highest at the start, and a call is
the only touch that can become a conversation immediately. Cadences that save calls for the end use them as a last resort, which is how they get treated.
- Never run two of the same channel back to back unless the second
references the first. Two unconnected emails read as a mail merge, because that is what they are.
HubSpot's framing is useful here: a sequence is a defined series of touches over a set period. Both halves are design decisions, and the channel mix is where most of the differentiation actually lives.
Decision 4: what ends it
The one nobody writes down, and the reason CRMs fill with accounts in a cadence that ended months ago.
Three exits, all explicit:
- Positive: they replied and the account moves to qualification.
- Negative: they said no. Record the reason, remove them, and honour it.
- Exhausted: the touches ran out. This is the one to define. What happens
next? Nurture, re-entry after a fixed period, or a genuine close-out.
Without a defined exhaustion exit, reps keep half-worked accounts alive because removing them feels like giving up, and the working list quietly becomes a graveyard.
A worked shape
Not a template to copy. An example of the four decisions made together, for an account that entered on a hiring signal with a roughly three-week window:
| Day | Touch | Note |
|---|---|---|
| 1 | Call | The signal is freshest today |
| 1 | References the call attempt, not a fresh pitch | |
| 3 | Call | Different time of day |
| 5 | A view or a light comment, not a connect-and-pitch | |
| 8 | New angle, not a bump | |
| 11 | Call | Last call attempt |
| 15 | Explicit close-out, easy to reply to |
Seven touches, fifteen days, three channels, and a defined end. The close-out email earns replies more often than anything in the middle, because it asks for a decision rather than attention.
The failures that look like message problems
- No signal at entry, so the timing has nothing to be timed against.
- Every touch the same channel, which caps the reply rate regardless of
copy.
- Bump emails that add nothing ("just floating this to the top"). A bump
spends a touch to say you are still there.
- No exit, so nothing ever leaves and the reported list size stops meaning
anything.
- Personalisation only in the first line, bolted onto a generic body.
Readers detect the seam.
How to tell if it is working
Not by reply rate alone. Reply rate rewards volume of easy touches.
Look at meetings held per hundred accounts entered, and at where accounts die in the cadence. If most die after touch two, the entry criteria or the opener is wrong. If they die at the end, the cadence is too long for the signal.
Related: the first ten seconds of a cold call for the call touches, and pipeline velocity for what this feeds.
Where this sits
Cadence design is the Motion pillar of the Tenbound Pipeline Architecture Standard. Motion is where the work gets sequenced and run, and it is the pillar most often confused with Message. Better copy inside a badly timed cadence does not fix the cadence.