A cold email sequence should be three messages: an opener and two follow-ups, spaced three to five business days apart, each shorter than the last and each adding one genuinely new thing. Then stop.
That length is not politeness. The marginal reply from a fourth and fifth message is small, and the marginal *complaint* is not. Complaints are the metric that ends your sending — Google asks bulk senders to keep reported spam below 0.3% — and the senders who cross it are overwhelmingly the ones who kept bumping. A sequence is a budget, and the budget is your sending reputation.
What each step is for
Step 1 — the opener. Three to five sentences. A trigger or a narrow problem, one outcome, one ask with an easy exit. One link at most, no image, no attachment.
Step 2 — a different angle. Not "just following up". Pick a second reason the thing matters: a different stakeholder's problem, a specific example, the objection you expect. Shorter than step 1.
Step 3 — the close. Two or three sentences. Announce that it is the last one, make the no cheap to give, and leave the door open. "If this isn't the quarter, say so and I'll close the loop" gets more replies than any amount of urgency, and a no is a useful outcome: it removes the person from your list before they remove themselves with the spam button.
| Step | Day | Length | Adds |
|---|---|---|---|
| 1 | 0 | 3–5 sentences | The trigger and the ask |
| 2 | +3 to +5 business days | 2–4 sentences | A new angle or a concrete example |
| 3 | +4 to +7 business days after step 2 | 2–3 sentences | A graceful exit |
Spacing, and why not shorter
Three to five business days is long enough that the reader has cleared their inbox and short enough that they remember the first message. Under two days reads as pressure, and pressure is what gets reported.
Spacing also has a mechanical side. If every contact in a 500-person campaign gets step 2 exactly 48 hours after step 1, your mailbox emits a visible wave. Randomise the interval within a range, keep sends inside working hours in the recipient's time zone, and randomise the delay between individual messages. Even spacing is machine-shaped, and machine-shaped sending patterns are one of the things that correlate across a group of domains.
Same thread or new thread?
Reply in the same thread for follow-ups. The reader gets the original context in one place, the message can be genuinely short because it does not have to re-establish anything, and you are not competing with your own earlier email in their inbox.
Two caveats. First, do not quote the entire previous message beneath each follow-up — by step 3 the quoted history is longer than the content, and that bulk is what pushes an HTML message toward Gmail's clipping threshold of roughly 102 KB. Second, never fabricate a thread: a "Re:" on a conversation that never happened is a deceptive subject line, which CAN-SPAM prohibits, and it is the fastest way to turn a mild non-response into a report.
A worked sequence
Step 1 — Subject: the two SDR openings in Berlin Hi Priya — saw you're hiring two SDRs in Berlin, which usually means ramp time is the thing on someone's mind. We built the onboarding tracker a similar-size team used to get new reps to a first booked meeting in three weeks instead of seven. Worth 15 minutes next week, or is this the wrong quarter? — Sam, MailPilot team
Step 2 — same thread, +4 business days Hi Priya — one thing I should have led with: the part teams tell us matters most is the first-week checklist, because that's where new reps stall rather than in month two. Happy to send the checklist itself if that's more useful than a call. — Sam
Step 3 — same thread, +5 business days Hi Priya — last one from me. If ramp isn't the priority this quarter, just say so and I'll close the loop. If it is, I'll send the checklist and get out of your way. — Sam
Step 2 adds a specific detail and offers a *smaller* ask than step 1, which is the move most sequences miss. Step 3 asks for a no explicitly. Nothing in any of the three has an image, an attachment or more than one link.
Stop conditions that have to work
Configure and then actually test these:
- Reply received. The whole sequence stops, not just the next step.
- Out-of-office. Does not count as a reply, and does not fire the next step while they are away.
- Reply from a colleague at the same company stops the sequence for your contact too.
- Hard bounce. Suppressed immediately and globally.
- Unsubscribe or an "not interested" reply. Suppressed across every campaign and every mailbox, immediately. CAN-SPAM allows 10 business days; there is no reason to use them.
- A bounce or complaint spike pauses the whole campaign for review.
Test each by hand before a live send. The expensive failure is a follow-up to someone who already declined.
What not to do
Do not write seven steps. If a sequence needs seven messages to work, the targeting is wrong. The extra steps buy a small number of replies and a complaint rate that costs you the domain.
Do not repeat the pitch. "Just bumping this to the top of your inbox" adds nothing and signals that you have nothing to add.
Do not escalate the tone. "I've reached out three times" makes the reader responsible for your sequence.
Do not use a new subject line to look like a new conversation. Changing the subject to evade their mental filter is the same manoeuvre as a fake "Re:".
Do not add a calendar link in step 1. It asks for commitment before you have earned attention and it adds a third-party domain to the message.
Do not run a sequence on an unwarmed mailbox. No structure survives a sending identity with no history.
Do not judge the sequence on open rate. Image pre-fetching by Apple Mail Privacy Protection and corporate proxies means recorded opens include machines. Judge on replies, and watch bounces and complaints per domain.
Do not assume no replies means bad copy. A sequence with no replies and no bounces is usually a sequence in the spam folder, and your sequencer will report it as fully delivered.
Where MailPilot fits
MailPilot tells you which of those two it is. It measures real inbox placement per provider through a seed-inbox network — including whether Gmail filed the message under Promotions instead of the inbox — warms every mailbox over OAuth on Gmail and Outlook or over SMTP/IMAP for SendGrid, Amazon SES, Mailgun, Postmark and custom SMTP, checks SPF, DKIM and DMARC per sending domain, and tracks per-mailbox reputation daily.
Ready to reach the inbox every time?
Automated email warmup across real peer mailboxes. Live inbox placement monitoring. Free 7-day trial - no credit card required.
Start free trial