Reply data by sequence step from four live 2026 cold email campaigns, 91,000+ sends. Where follow-ups stop paying and why we run four to five, not nine.
Short answer: Four to five emails. In four live 2026 campaigns, 91,000+ sends, the first email produced between a seventh and two-fifths of all human replies (14-39%), follow-ups 1-3 produced 41-68%, and the steps after follow-up 3 produced a fifth to a third where they were tested, falling to almost nothing by the end of the two small campaigns. We run four steps by default, five when the list is big enough to read step 5, and a sixth only while the fifth is still producing. Page one of Google will tell you four to nine, five, or that nobody knows, so the step-level data is below.
I run a lead generation agency. We are paid per qualified meeting, not per email, so I have no reason to send a ninth email to a list I will need again in six months. That is my bias. Keep it in view.
Three of the pages ranking for this question, as of October 2026:
Four to nine, or at most five, or nobody knows. Two of the three sell the software that sends the emails; BounceZero sells list verification, as of October 2026. None of them is paid by the reply. None shows replies by step on a named campaign. I can.
Read this before the table. These are human replies of any sentiment: interested, not interested, a question, "take me off your list". They are not positive replies and they are not meetings. The step-level split for positive replies is not in the campaign analytics I can pull, so I will not guess at it. Autoreplies and out-of-office are excluded. No open data, because we do not track opens.
| Campaign | Sends | Replies | Rate | 1st email share | FU 1-3 share | FU 4+ share |
|---|---|---|---|---|---|---|
| A — P1 EU/Turkey, 7 steps | 46,201 | 360 | 0.78% | 24% | 44% | 32% |
| B — P1 Caribbean, 7 steps | 2,196 | 22 | 1.00% | 14% | 68% | 18% |
| C — PR Noir ICP1, 4 steps | 38,254 | 123 | 0.32% | 32% | 68% | — |
| D — P1 Indian Ocean, 8 steps | 5,017 | 41 | 0.82% | 39% | 41% | 20% |
P1 is P1 Communications, a UK hospitality PR agency prospecting hotel and travel brands. PR Noir is a US PR agency prospecting jobs and employment brands.
Three things are true in every row. The first email is never the majority of replies. Follow-ups 1-3 are the biggest block, by two points in campaign D and by a wide margin elsewhere. After follow-up 3 the answer depends on the campaign, and that is the whole argument.
The big one. 8,227 contacts entered the first email; 46,201 emails went out across seven steps; 360 people replied, 4.4% of the cohort.
Reply rate per email sent, by step: 1.07%, 0.88%, 0.81%, 0.65%, 0.68%, 0.76%, 0.48%.
Follow-up 5 produced 45 replies on 5,953 sends, 0.76%. Better than follow-ups 3 and 4. Only follow-up 6 fell away, at 0.48%. On this list, with this copy, the sixth email was worth sending and the seventh was marginal. This is the campaign that contradicts the claim that every follow-up does worse than the last. It is roughly flat for six emails.
PR Noir's first ICP. 11,551 contacts, a colder list than A, 0.32% overall. Per step: 0.34%, 0.33%, 0.33%, 0.28%. Follow-ups 1 and 2 matched the first email per send. 68% of replies came from follow-ups.
It was built as a four-step sequence, so I cannot tell you what a fifth would have done. That is the honest limit of this one.
967 hotel contacts in one region. First email: 16 replies, 1.65%. Follow-up 1: 11 replies, 1.95%. Those two emails produced 66% of all replies.
Then 4, 2, 7, 1, 0, 0. Follow-up 4 had a blip of seven. Follow-ups 5-7 produced one reply on 1,474 sends, 0.07%. Three emails to about 490 people each, for one reply. They spent sender reputation on a small list we wanted to keep clean and bought nothing.
Campaign B, 2,196 sends and 22 replies, is too small to read on its own. It is here because I checked the pattern on a second segment, and follow-up 1 was again the strongest single step, 9 replies of 22.
So the rule we run now: four steps by default. Five when the list is large enough that step 5 has sends you can read, which in practice means a few thousand contacts. Six and beyond only while five is still producing, checked weekly. Not nine.
Campaign A's first email ran three subject lines on the same list, same body, same weeks:
| Variant | Sent | Human replies | Rate |
|---|---|---|---|
| 0 | 3,141 | 42 | 1.34% |
| 1 | 2,543 | 26 | 1.02% |
| 2 | 2,543 | 20 | 0.79% |
Best beat worst by 1.7x. Follow-up 5, the sixth email, added 45 replies from 5,953 sends. If you have a list of that size and a week to spend, spend it on the opener's subject line, not on a sixth email. Cold email for PR agencies goes further on what to put in that subject line.
"Just bumping this" is a repeat, not a follow-up. Every step needs its own job.
A follow-up 1, written for a PR agency prospecting a hotel group. Same thread, no new subject line:
[First name], one more angle. Most hotel groups we speak to are well covered in the trade press and absent from the titles their guests read on a Sunday. If [Hotel] is the same, that gap is the thing we fix. Worth a short call, or is this the wrong quarter?
The brackets are yours to fill. The angle only works if the gap is real for that hotel, so check before you send.
Pay per qualified meeting, no retainer, no minimum spend, nothing charged for a no-show. The number is quoted on the call once we have seen your deal size. Why that model and not a retainer is in pay-per-meeting vs retainer.
Get four things in writing, from us or from anyone else:
If you want the definition we use, it is written out in what counts as a qualified meeting.
Then run your own deal size through the ROI calculator before the first call, so you know what a meeting is worth to you before anyone tells you what it costs.
Or run the free GTM teardown on your own site for the ICP, cold email and numbers specific to your company.