Most advice on cold email sequence length gives a number and moves on. The number matters less than three things you can write down and check: what each follow-up adds, what makes the sequence stop, and who approved those rules before the first email went out.
This post covers how to set follow-up count and timing, and how to treat replies and non-replies differently. It also covers how a reviewer can approve cadence as part of an outbound policy.
The short answer
No fixed count fits every list, but you can set one with a rule. A sequence is the ordered set of emails one contact receives; the glossary entry on sequence has the definition. Its length is a decision about how much contact you are willing to have with someone who has not answered.
Start from the base rate. Belkins' 2026 study covers 7.5M B2B emails sent in 2025, with reply rate defined as replies ÷ emails sent [[1]](#sources). Across industries, 0.45% of emails get a reply [[2]](#sources). Most recipients do not answer any single email, which is why teams add follow-ups. It is also why each extra one needs a reason.
The rule: send one email per distinct, sourced observation about the account. The first email uses one observation and each follow-up uses a different one. A Readiness Card needs at least three observations, each with a source and a date [[3]](#sources). An account with three such observations supports a sequence of three emails. An account with one supports one email.
Cost is not the limit. Sending, follow-ups and the mailboxes a plan includes use no credits [[4]](#sources). If the number of steps is capped, the cap is a judgment about the recipient, not a billing constraint.
What changed
What changes in this approach is where the cadence lives. If it sits only inside the sequence tool, set by whoever built the campaign, a reviewer may never read it.
A better place is the outbound policy. In Outlevo, nothing is sent outside a policy that a person approved [[5]](#sources). Policies set who can be contacted, on which channels, how often and with what budget [[5]](#sources). Follow-up count and spacing fall under "how often", so a reviewer can read and approve them before launch.
Every external action is also checked against the approved policy before it runs, and recorded in a ledger [[5]](#sources). If someone asks later why a contact received a fourth email, the answer is in the policy and the record, not in someone's memory.
Other tools handle approval differently. Reply.io's Jason AI lets you approve each message before it goes out, or run on autopilot [[6]](#sources). AiSDR's per-campaign managed service has you approve before anything sends, and replies can run in co-pilot mode [[7]](#sources). Approving each message and approving the rules those messages follow are different jobs. A cadence rule is easier to review once than to re-check in every message.
What each follow-up should add
A follow-up that says "just bumping this" adds nothing. Give each step one job and write the job next to the step.
- Email 1: the reason you are writing to this company, tied to the first observation about them.
- Email 2: a second observation about the company that the first email did not use.
- Email 3 (last): a third observation, a different proof point or a smaller ask. Or a short, plain close that tells the person this is the final note.
The rule above depends on having more than one fact to work from. If your own research stops at one fact per account, you have material for one email, and a long sequence will repeat it.
Keep the facts honest as well. Facts read from a website are kept only when their quote is found on the page [[3]](#sources). Guesses are labelled as hypotheses, never presented as findings [[3]](#sources). Hold follow-ups to the same standard. A step that states a guess as a fact is a worse email than a step that is not sent.
When to stop
Write the stop rules as plain conditions, not as a feeling about the list. A workable set:
- The contact replies, positively or negatively.
- The contact unsubscribes.
- The address bounces.
- The sequence reaches its last step.
Smartlead reports sent, reply, bounce, unsubscribe and category-change events back to Outlevo by webhook [[4]](#sources). Those are the events your stop rules depend on, so a reviewer should ask that the policy name each one.
Business email is covered by the law on commercial email too. The FTC's CAN-SPAM guide says: "The law makes no exception for business-to-business email." [[8]](#sources)
Replies and non-replies are different cases
Do not run one cadence for everyone. Split by what the contact did.
No reply. The contact continues down the sequence at the approved spacing until a stop rule applies. Do not shorten the gaps because the earlier steps went unanswered.
A reply. The sequence should not send the next scheduled step to someone who has answered. On LinkedIn, Outlevo's flow is explicit about this. It sends a connection invitation with a short note of at most 300 characters, then one message after the person accepts [[9]](#sources). Follow-ups wait for an acceptance, and a reply stops that lead's emails and the sequence [[9]](#sources). Your email policy should state the same rule in its own words, so a reviewer does not have to assume it.
An interested reply. Here the job changes from following up to answering. Outlevo's Calendly integration puts your booking link in the answer to an interested reply [[10]](#sources). On autopilot the answer goes at once; otherwise it waits in Replies for a person to send [[10]](#sources). Which of those two modes applies is a choice the reviewer should see written down.
A negative or out-of-office reply. These need their own line in the policy. A "not now" is different from an automatic reply, and both differ from "remove me". Decide in advance which ones end the sequence and which pause it.
Spacing
Set one gap per step, in days, and write the gaps into the policy where a reviewer can read them. Three points help when choosing:
- Make the gap long enough that the second email reads as a new message, not a repeat of the first.
- Make it the same for every contact on the sequence, so the reviewer approves one rule and not a pile of exceptions.
- Count from the actual send date, not the planned one. A mailbox that was held back changes the real schedule.
The last point matters at the start. Before a campaign sends, every mailbox must be connected and warmed up, and its domain must have SPF, DKIM and DMARC checked in the last 7 days [[4]](#sources). A new managed mailbox sends nothing until it is warm: at least 14 days, usually about 3 weeks [[11]](#sources). A sequence planned for a launch date can slip if the mailboxes are not ready.
Each added step is also more mail to send from the same mailboxes. The guide to daily limits per mailbox covers the per-mailbox side. The deliverability guide and the glossary entry on deliverability cover deliverability more broadly.
What a reviewer approves
Treat the cadence as a short checklist the approver can answer yes or no to:
- How many emails does one contact receive at most, and over how many days?
- What does each step add that the previous one did not?
- Which events stop the sequence? Reply, unsubscribe and bounce at minimum.
- What happens after an interested reply, and does a person send the answer or does it go out on its own?
- Which channels are in the sequence, and is the order stated? On LinkedIn, the follow-up message waits for an acceptance [[9]](#sources).
- Who can pause it? Kill switches pause a campaign, a channel or the whole workspace at once [[5]](#sources).
If the answer to any of these is "it depends on whoever set it up", the policy is not ready to approve.
What to do
- Pick a maximum and write it down. Choose the number of emails and the gaps between them, and put both into the outbound policy before launch.
- Give every step a job. If you cannot say what step three adds, remove it.
- Check your material. If an account has fewer observations than the steps you planned, shorten the sequence for that account rather than repeating the first email.
- Write the stop rules. Name reply, unsubscribe and bounce as stop events, and decide how negative and automatic replies are handled.
- Split replies from non-replies. Decide whether interested replies are answered on autopilot or wait for a person.
- Check the mailboxes first. Confirm warm-up and the SPF, DKIM and DMARC checks before the first send date [[4]](#sources).
- Review after real sends. Compare the replies and unsubscribes at each step. If a late step produces unsubscribes and no replies, shorten the sequence and approve the change as a policy edit.
If you are still deciding who to write to, the free ICP assessment is a starting point. It reads up to 5 public pages of a website [[12]](#sources). The how it works page explains the approval flow in full.
Sources
- Belkins, cold email response rates (2026)
- Belkins, cold email response rates (2026)
- Readiness Cards
- Smartlead integration
- How Outlevo works
- Reply.io Jason AI
- AiSDR pricing
- FTC: CAN-SPAM Act compliance guide for business
- LinkedIn integration
- Calendly integration
- Managed mailboxes integration
- Free ICP assessment