How to write a cold email follow-up.
Most cold replies don’t come from the first touch. They come from touches 2, 3, and 4 — and most founders write all of them badly. The follow-up is its own craft, with its own structure, its own length constraints, and its own anti-patterns. Here’s how each touch in a 5-touch sequence should be written, what the breakup email should actually say, and the templates that hold up.
TL;DR
- 60-70% of replies come from follow-ups, not the first touch. If you only send touch 1, you’re leaving most of the pipeline on the table.
- Touch 2 is the most-replied-to message in a sequence. Send 3-4 business days after touch 1.
- Each follow-up should add new information, not re-pitch. The same message louder doesn’t work.
- Follow-ups get shorter, not longer. Touch 1 = 60-90 words. Touch 5 = 15-30 words.
- The breakup email (final touch) outperforms touches 3 and 4 on reply rate. Use it.
- Send follow-ups in the same thread as the original (keep
Re:subject). Threaded follow-ups outperform new-thread follow-ups on touches 2-4.
Why follow-ups matter more than touch 1
A common founder mistake: spend an hour crafting the perfect cold email, send it, get a 1.5% reply rate, conclude that cold email doesn’t work. The actual math:
- Touch 1 reply rate: 1.5-3% on typical B2B cold
- Touch 2 reply rate (3-4 days later): 2-4% — often higher than touch 1
- Touch 3 reply rate (7 days later): 1.5-3%
- Touch 4 reply rate (10-14 days later): 1-2%
- Touch 5 — the breakup (3-5 days later): 2-4% (yes, often higher than touches 3 and 4)
Cumulative across a 5-touch sequence: typical total reply rate is 6-12%, of which 60-70% comes from touches 2-5. If you cut after touch 1, you’re running at 25-30% of the achievable reply rate. The math is overwhelming: write the follow-ups.
The mechanism is simple. The first message arrives during a moment of inbox-skim attention. Most prospects who would be interested don’t reply on the first read — they intend to come back to it, forget, or simply prioritize other things. The follow-up is what catches them at a moment when they have the attention budget to respond.
Touch 2 — the most important message in the sequence
Touch 2 is sent 3-4 business days after touch 1 (so Tuesday→Friday, or Wednesday→Monday — never Friday→Monday because the weekend gap diffuses the signal). It is sent in the same thread, so the recipient sees it as a reply to the original message.
The structure that works:
- Opening: a sentence that acknowledges the prior message without re-explaining it. “Following up on this — ” or “Wanted to bump this to the top of your inbox” both work. Avoid “Just checking in” (signals the operator has no new information).
- New information: this is the critical move. Touch 2 should add something — a different angle on the value prop, a relevant case study, a specific observation about the prospect’s company you missed in touch 1, or a different question. The same message repeated is not a follow-up; it’s a nag.
- CTA: lighter than touch 1. If touch 1 asked for a 15-minute call, touch 2 asks for a 5-minute answer to one question. The lower-friction ask increases conversion.
- Length: 40-70 words. Shorter than touch 1.
Working example, in plain English:
Following up on this — saw you also hired two AEs in the last 30 days, which usually means you're feeling the lead-gen crunch we talked about. Quick one: are you running outbound in-house or with an agency? I can share what's worked for two other Series-B teams in your ICP and you can use it or not.
Touch 3 — the proof point
Sent 7 days after touch 2. The recipient has now seen you twice. They’ve either filed you away as not-now or they’re still on the fence. Touch 3 is where you offer a low-cost proof point — a case study, a specific number, a piece of social proof — that earns the next read.
Structure:
- One-line acknowledgment.
- A concrete number or named customer. “We’ve been doing this for [Company X] and got them from 0 to 8 meetings/week in 6 weeks.”
- An offer to send the case study or share a quick walkthrough. The CTA is still light — “want me to send it?” rather than “book a call.”
- Length: 30-50 words.
The named-customer reference is the highest-leverage element. If you don’t have a real customer to cite, cite a vertical or a generic outcome (“we’ve seen this pattern at 4 of the last 6 Series-A SaaS teams we worked with”). If you have a real customer with explicit permission, use the name.
Touch 4 — the pivot
Sent 10-14 days after touch 3. This is the lowest-reply-rate touch in the sequence — by this point you’ve either won the prospect’s attention or you’ve lost it. Touch 4’s job is to pivot the conversation away from the original ask and into a different one.
Pivot options that work:
- Pivot to a different stakeholder: “Is this something your head of growth would care about more than you?” This re-uses the message at a different recipient and bypasses the silent-no from touch 1-3.
- Pivot to a different ask: instead of a meeting, ask for a 2-sentence reply. “Even a no would help — are you running outbound at all right now?”
- Pivot to a content asset: link to a deep-dive article or a tool that helps them with the problem regardless of whether they buy. Reduces ask to zero and re-engages on value.
The CTA in touch 4 should be the easiest in the entire sequence to respond to. A yes/no question typed in 5 seconds outperforms any framing that requires the recipient to schedule something or open a deck.
Length: 25-40 words.
Touch 5 — the breakup email (the surprise winner)
The breakup email — touch 5, sent 3-5 days after touch 4, framed as the final message in the sequence — consistently outperforms touches 3 and 4 on reply rate. This is the most counterintuitive finding in cold copy and the one most teams underuse.
The structure:
- Explicit framing: name it as the last message. “Last note from me” or “closing the loop on this.”
- No re-pitch: do not summarize the value prop. The recipient has seen it four times.
- A simple question: ask the prospect to confirm they’re not interested, or to flag the right person. Make “no” the easiest response.
- Length: 15-30 words.
Working example:
Closing the loop on this — should I assume outbound isn't a priority for you right now? Happy to take the hint either way.
Why this works: the breakup creates a small social pressure to respond. The recipient who’s been ignoring the thread now has a clear exit (say no, you go away) or a chance to engage. Both outcomes are cheap to provide. The reply rate is typically 2-4%, and roughly half of those replies are positive or soft-pass — neither of which would have surfaced without the touch.
The follow-up cadence — timing matters
The spacing between touches is part of the signal. Too close together reads as nagging; too spread out and the prospect forgets the original context. The empirical pattern that works:
| Touch | Send when | Length | Job |
|---|---|---|---|
| Touch 1 | Day 0 | 60-90 words | Earn the open and the first read |
| Touch 2 | Day 3-4 | 40-70 words | Add new information, lower the ask |
| Touch 3 | Day 10-11 | 30-50 words | Offer a concrete proof point |
| Touch 4 | Day 21-25 | 25-40 words | Pivot — different ask or different person |
| Touch 5 | Day 25-30 | 15-30 words | Breakup — make “no” easy to send |
The threading question — same thread or new thread?
Touches 2-4 should go in the same email thread as touch 1, keeping the Re: subject prefix. This produces:
- 5-15% higher open rates than a new subject on touches 2-4 (recipient processes it as an active conversation, not a new cold message).
- Lower spam-folder risk (a reply within an existing thread is statistically a more trusted pattern than a new outbound).
- Better in-inbox visual continuity for the recipient — they can scan the thread and remember context.
The exception: touch 5 (the breakup) sometimes performs better as a new subject, because the explicit “last note” framing benefits from arriving as a distinct message rather than buried at the bottom of a long thread. Test both.
Anti-patterns to avoid
- “Bumping this to the top of your inbox” — fine occasionally. Used as the entire content of touch 2, it’s lazy and reads as such.
- “In case you missed this” — implies the recipient missed something. They didn’t; they chose not to reply. The framing is condescending.
- “Following up on my last 3 emails” — counting your own follow-ups out loud signals the operator is nagging.
- Re-pitching the same value prop — the recipient saw it last week. Either add new information or shorten the ask.
- Aggressive guilt-trip closes — “Should I take your silence as a no?” works as the breakup (touch 5). Used at touches 2-3, it reads as petulant.
- Apology framing — “Sorry to bother you again” signals you know the message isn’t worth their time. If you believe that, don’t send the touch.
The 5-touch cap and why you stop
Five touches is the operating cap. Touches 6+ produce diminishing reply rates and rising spam-complaint rates. The recipient who hasn’t responded after 5 touches over 25-30 days isn’t a near-term opportunity. Move them to a nurture cadence (see nurture cadences) and re-engage in 60-90 days with a new angle or trigger.
The spam complaint rate climbs steeply at touch 6 and beyond. Every touch past 5 is contributing more to your sender reputation damage than to your pipeline.
Where this fits
This chapter is the per-touch craft layer for the multi-touch sequence covered in email sequencing. Sequencing covers the cadence math and per-touch reply curve; this chapter covers what to write inside each touch. The two are paired — the structure of the sequence determines what each follow-up’s job is, and this chapter handles the execution.
The natural next read is CTA architecture, which covers the specific ask in each touch and why the ask should get lighter as the sequence progresses.
Allston Labs operates the full sending estate as a service.
We provision domains, configure the entire authentication record set, run warmup, and monitor reputation across providers. The stack lives under your entity. The engineer on call lives in your Slack.