A sequence is designed from what you have to say, not from a number of touches. Six to eight contacts across three to four weeks cover most B2B markets, provided every message brings something new. The rule that separates useful sequences from the rest: if a touch cannot stand alone, it only adds volume and it damages the ones before it.
Most sequences we inherit were designed backwards: someone decided the number of touches and the cadence first, then filled the boxes. You recognise the result at message three, the one that says “bringing my email back to the top in case it was missed” and brings nothing.
A sequence is designed the other way round. You start from what you have to say, and the number of touches follows. Here is how we build them, and the four mistakes that waste a sequence even when it is well written.
The principle: one touch, one reason to exist
The rule separating useful sequences from the rest fits in a sentence: every message must be able to stand alone. If a touch only makes sense because it follows the previous one, it exists to add volume.
The test is simple. Take message four of your sequence and imagine it is the first thing the prospect receives. If it stands up, it has a reason to exist. If it opens with “following up on my note”, it does not, and it damages the three before it by making them read as automated harassment.
That constraint naturally limits length: you do not have twelve different things to say to someone who asked you nothing. Six to eight, across three to four weeks, is the format that covers most B2B markets.
The trigger sets the opening
A sequence does not open with a presentation of your company. It opens with the reason you are writing now, to this person.
When entry comes from an observable event at the prospect, that reason is given: the event is the subject, and you are only the context. That is the whole principle described in signal-based selling, and it completely changes the first message.
Without a trigger, your opening has to carry a hypothesis about their situation, drawn from what you know of the segment. That is harder, weaker, and it demands a tight target: a hypothesis can only be right if the companies addressed genuinely resemble each other, which points back to why an ICP that is too broad kills growth.
Pick two or three channels, not five
The reflex is to add channels to increase coverage. Mostly it increases noise, and it dilutes the attention of whoever runs the sequence.
Choose channels where your target actually replies, and verify that on your own data rather than in an article. Each has a distinct function. Email carries reasoning: it is the only place you can develop an idea. The phone settles things fast, including negatively, which is valuable. Social builds familiarity before you write, and it works better as preparation than as solicitation.
Two channels held properly beat four skimmed. And a channel where you never get a reply should leave the sequence, even if everyone else uses it.
Pace, and the fixed-cadence mistake
A sequence does not run at a regular pace. The first messages sit close together, because the trigger is fresh and the window is closing. The last ones spread out, because the point is to stay present without weighing.
A structure that works: two touches in week one, two in week two, one in week three, a final one a fortnight later. The important part is not those numbers, it is the decay: a fixed three-day cadence for a month reads as automated, because it is.
Adapt the pace to the signal’s lifespan too. An event opening a window of several months, such as a funding round, allows spreading out. Repeat visits to your pricing page call for a reaction the same day.
What to personalise, and what not to
Surface personalisation, inserting the company name and sector into a template, fools nobody any more and costs time for nothing.
What matters is substantive personalisation, and it concerns only two messages. The first, because it has to name the precise reason you are writing. And the last, because a personalised closing message often gets a higher reply rate than all the others, simply because it no longer asks for anything.
Middle messages can be shared across a segment, provided that segment is narrow enough for the content to stay accurate. It is the same trade-off as for messaging: one promise per segment, one proof per role, which we detail in adapting your message by persona.
Knowing when to stop
A sequence has a planned end, and that end is a message, not silence.
The closing message says explicitly that you are stopping, sums up in one sentence what you thought you could bring, and leaves the door open. It costs thirty seconds to write and it produces a significant share of replies, because it reverses the burden: you are no longer asking for anything.
After that the account is not harassed, it goes back in the queue. It will come round again when a new trigger appears, and you will then have a legitimate reason to make contact. Following up indefinitely costs your deliverability, your reputation, and the right to come back later with a real reason.
The four recurring mistakes
The empty touch, the follow-up that brings nothing. It is the most frequent, and it turns a sequence into a nuisance.
The fixed cadence, which signals automation as reliably as a robot signature.
One sequence for every segment, which forces language general enough to touch nobody.
No ending, which leaves accounts in permanent follow-up and durably damages your ability to contact them again.
Where to start
Open your current sequence and apply the test to each message: could this one be the first? Delete the ones that fail.
In most sequences we open up, three or four out of ten survive. That is a much better sequence, it takes less work, and it returns more.
Frequently asked questions
- How many touches should a B2B sequence have?
- Six to eight across three to four weeks covers most cases. The number matters less than the content: a four-message sequence where each one brings something beats twelve empty follow-ups. The real ceiling is how much you actually have to say.
- Which channels should you combine?
- Two or three, chosen where your target actually replies, not where it is easy to send. Email carries reasoning, the phone settles things fast, social builds familiarity. Adding a channel where nobody answers does not increase coverage, it increases noise.
- Should every message be personalised?
- The first and the last, always. The middle ones can be shared across a segment, provided the segment is narrow enough for the message to stay accurate. Surface personalisation, of the insert-the-company-name kind, fools nobody any more and costs time for nothing.
- When should a sequence stop?
- At its planned end, with an explicit closing message that leaves the door open. Following up indefinitely costs your deliverability and your reputation for close to zero return. An unresponsive account goes back in the queue for a later trigger, it does not get harassed.