How long should a cold outreach sequence be?
By Jānis Plūme, Founder, Outbound Pros · 2026-08-06
Quick answer
A cold outreach sequence should be exactly as long as the number of genuinely distinct reasons you have to contact the prospect, plus one same thread bump. For most B2B offers that is four to six touches, and the binding constraint is angles rather than days. Every touch past your last real angle says some version of bumping this to the top of your inbox, and those touches do not merely underperform, they spend the familiarity the earlier ones built. Count your angles first. If you have three, you have a three angle sequence, whatever the template said.
Designing email and LinkedIn sequences for 36 active B2B clients since 2024.
How long should a cold outreach sequence be?
A cold outreach sequence should run for as many touches as you have genuinely distinct reasons to contact the prospect, plus one same thread bump, and then it should stop. For most B2B offers that lands at four to six touches. The number is an output, not a target.
Every published answer here is a number of days or a number of touches, and both are the wrong unit. Seven touches across twenty one days is a container, not a plan. What fills it is how many separate things you have to say, which no template can know.
We call it the Angle Budget. One angle per touch, and when the budget is empty the sequence is finished, whatever slot number you stopped on.
What is an angle, and how do you tell two of them apart?
An angle is a distinct, defensible reason this person should care right now. Not a new subject line, not a new opening sentence, not a rearrangement of the same claim. What separates two angles is what the prospect learns.
The Angle Audit. Write the reason for each touch in one sentence, without mentioning that it is a follow up. Put the sentences in a column. Any two sentences that could be swapped without changing what the prospect learns are one angle, not two.
| Reason behind the touch, in one sentence | Angle? | Why |
|---|---|---|
| You are hiring three sales people and the ramp cost is what surprises finance | Yes | Observable fact, with a consequence |
| Everyone in your category shows up in AI answers for this and you do not | Yes | Different evidence, different implication |
| We work with companies like yours and get great results | No | Every touch could say it |
| Just floating this to the top of your inbox | No | The only reason is that the last one failed |
Most teams who believe they have seven angles have three. In our own process that gets caught at the approval round: clients approve messaging and lead lists before anything sends, and the approval document carries the reason for each touch on one line. Written out in a column in front of the person who knows their market, two of them usually merge before the client says a word.
Supply differs by motion. WideNET, our high volume motion, tests angles across the full addressable market, so it carries a full budget by design. Spearhead, our signal triggered motion on the hottest slice, runs shorter: the trigger is the angle, and a second one that good is rare.
Why is touch count the wrong unit?
Touch count is an output of the angle count and the channel structure, not an input you get to choose. Two sequences with five touches behave completely differently depending on what is inside them.
| Sequence A, five touches | Sequence B, five touches | |
|---|---|---|
| Real angles | 5 | 2 |
| Touches referring back to a previous message | 0 | 3 |
| What the prospect sees at touch five | A fifth reason | A fourth reminder |
| What the sequence spends | Attention | Goodwill |
Both look identical in every dashboard, because length gets reported as a count and a count is what the sending tool stores. It is also why the open web contradicts itself here: five touches is right for the team with five angles and wrong for the team with two, and the article cannot tell which is which.
How many emails in a sequence is too many?
One more than your last real angle. The touches after that point do not fail neutrally, and that is what most sequence length advice misses.
A sequence with eight slots and four real angles produces four touches that say something and four that say some version of bumping this to the top of your inbox. The second group is not free, it spends the familiarity the first group paid for, and a prospect who was mildly warm at touch four is irritated at touch seven.
We grade this at the sequence level, because the sequence is the unit we test and the unit we kill, and the kill line sits at 0.5% positive on sends. That judgement runs off standing monitoring desks reading the fleet daily, which is why a sequence that has gone quietly flat gets caught in a week instead of a quarter. More than once I have watched a re engagement variant bolted onto an exhausted sequence come back below the base it was extending. The paired figures for that account sit in our own reports and not in the verified table this site publishes from, so it stays an operator observation rather than an evidenced claim, and it is named here as the gap it is.
The full threshold ladder is campaign management rather than cadence, and our sibling property allboundpros.io covers it once it publishes.
Where does the same thread bump fit, and when do you break the thread?
A same thread bump is a follow up sent inside the original email thread, and it earns its place twice, after which you break to a new subject line carrying a new angle. Threading gives context for the first follow up or two without making the recipient search for it.
Past that, a thread with four unanswered messages stacked underneath is a visible record of being ignored, which makes the fifth harder to answer rather than easier. Breaking it is not a reset trick. It is an acknowledgement that the previous angle did not land. If the new thread says the same thing as the old one, keep the thread and save yourself a touch. Spacing is a separate decision, worked through on touch timing and spacing.
Does a channel switch buy you an extra touch?
Yes, and it is the one genuine exception to the angle rule. A LinkedIn touch and an email touch are recognised as separate events by someone who ignored the first one, so a channel boundary lets a single angle land twice without reading as a repeat.
The allowance is narrower than it sounds. One angle, two channels, once. Whether you can use it at all is arithmetic: divide your prospect count by your daily LinkedIn ceiling and see whether the channel can reach the people you want to escalate onto, which is what sequence capacity planning works through. What the DM actually says belongs to our sibling property's DM sequence method, the structure of the switch is on email and LinkedIn sequence structure, and the commercial case for running two channels at all sits with the agency that runs both of them.
When should you stop following up with a prospect?
Stop when you have run out of angles, when the prospect has given you a date, or when the sequence as a whole has hit your kill threshold. Those are three different stops and teams routinely apply the wrong one.
- Out of angles. The per prospect stop, and the right answer more often than the other two.
- They gave you a date. Not a stop, a diary entry. "Not now, ask me in Q1" is the most valuable reply after a yes, and a system that files it as a removal loses the people who told you when to return.
- The campaign hit the kill threshold. A decision about the sequence rather than the person. Kill it, keep the list, rebuild the angles.
Do breakup emails still work?
A breakup email works as a final angle when the angle is real, and fails as a manipulation when it is not. The version that works is a genuine last contact with a genuine last reason. The version most people send is a guilt lever with a deadline nobody enforces.
If you write that you are closing the file, close it. The same sender reappearing six weeks later teaches the recipient that nothing you write is literally true, which is expensive for a sender identity you will use for years. And a reply asking to be removed rather than contacted later is a compliance decision rather than a cadence one: in the US the governing text is the CAN-SPAM Rule at 16 CFR Part 316.
Should enterprise sequences be longer than SMB ones?
Yes, and the reason is the buying cycle and not persistence. More genuinely changes at an enterprise account over six months than at a twenty person company over six weeks, so more real angles exist and wider gaps are defensible. Touching several people at one account is a different design problem with different arithmetic, and it is not the question this page answers.
Where this rule breaks, and who should ignore it
The Angle Budget is a rule about the unit of measurement, not a measured optimum. We have no controlled test showing a five angle sequence beats a nine touch one on matched lists, and neither does anyone publishing a recommended number. What we have is a mechanism you can check in ten minutes and a default shape, tighten early and widen late. The one measured number this site owns is about motions rather than lengths, published with its confound on the multichannel lift benchmark. I would rather name the gap than fill it.
Four situations where this will not help you.
Your problem is the offer. If three genuinely different angles produced close to nothing across a few thousand sends, the sequence is not what is broken, and more touches buy more silence at greater expense.
You run founder led outbound to under a hundred named accounts. Each contact there is a judgement call and not a sequence. Write to people individually.
You are following up on inbound or a warm introduction. Someone who filled in a form has already handed you the angle. Everything here assumes cold.
Your constraint is capacity and not angles. If follow ups already go out late, that is a scheduling problem with a different fix. Shortening a sequence frees capacity the same afternoon, while adding mailboxes takes weeks of warm up first, and warm up does not compress. That is the capacity arithmetic, not this page.
Count your angles, then put them on a calendar
The Sequence Cadence Builder takes your angle count, your channels and your window, and returns a day by day interleaved email and LinkedIn touch plan with real dates. Free, no signup, and it will produce a four touch plan if that is what your angles support.
It will not pad the sequence to hit a recommended number, because there is no recommended number.
If writing the angles is the part you have no capacity for, that is the job the agency that writes and runs these sequences does inside the group. We publish no prices anywhere in this group, because scope varies enough that a number without context misleads.
Frequently asked questions
How many touches should a cold email sequence have?
As many as you have distinct, defensible reasons to make contact, plus one same thread bump. For most B2B offers that is four to six, but the number is a result and not a target. Two teams with the same list and the same offer should end up with different sequence lengths if one of them did more research.
Should every follow up have a new angle, or can I repeat the offer?
Every follow up should carry a new angle, with one exception: a same thread bump straight after the opener, which catches the group who meant to reply and did not. Repeating the offer in different words is the first angle at a lower response rate. Which touch carries the personalization budget is an angle architecture question rather than a count question, and this page owns the count.
Does a LinkedIn message count as one of my angles?
No. A channel switch lets one angle land twice without reading as a repeat, so the LinkedIn touch usually carries an angle you already had. That is an allowance, not a licence to double the length. What the message actually says belongs to our sibling property linkedpros.io, which publishes DM sequence copy as its own subject.
How many days should sit between the touches?
Tighten early and widen late: the first follow up within a few days of the opener, later touches a week or more apart, because each tests a new angle instead of reminding anyone of anything. The right intervals depend on deal size and buying cycle, so we publish no fixed pattern.
Does a longer sequence damage deliverability?
Indirectly, yes. Replies, complaints and unsubscribes feed sender reputation, so a sequence that keeps sending after it has run out of things to say generates the signals mailbox providers read as unwanted mail. The mechanical side includes one click unsubscribe, specified in RFC 8058, and domains, mailboxes and warm up are a separate discipline covered by the parent agency rather than on this site.
MultichannelPros is part of the Outbound Pros group, operated by Jānis Plūme.
Last updated: 2026-08-06