The Cold Email Follow Up Playbook: Schedule, Templates, and When to Stop

Most cold email campaigns are decided after the first send. The opener gets the credit, but in nearly every campaign we have measured, the majority of replies arrive on a cold email follow up rather than on email one.

That inverts how most teams spend their effort. Weeks go into the opener, subject lines get A/B tested to death, and then the follow ups are written in ten minutes as an afterthought. The sequence that produces most of the pipeline gets the least attention.

This is the playbook for the other way around: how many follow ups to send, on what schedule, what each one should say, the mistakes that quietly kill threads, and what to do when the sequence runs out.

Why the follow up wins the campaign

A non-reply to a cold email is almost never a no. It is a timing problem. Your message landed while the prospect was in a meeting, mid-quarter-close, or triaging 80 other emails, and it scrolled out of view before it was ever evaluated.

Follow ups convert because they re-enter the inbox at a different moment. Same offer, same person, different Tuesday. Across the sequences we look at, a reasonable working picture is that email one produces 35 to 45 percent of total replies and the follow ups produce the rest, with the first two follow ups doing most of that work.

There is a second effect that gets less airtime: follow ups filter for seriousness. The prospect who replies on touch four has seen your name five times and answered anyway. Those threads convert to meetings at a higher rate than fast replies to the opener, in our experience, because the interest survived repetition.

The cold email follow up schedule that works

The default below is four follow ups over 30 days. It front-loads touches while the opener is still in short-term memory, then widens the gaps as the odds per touch drop.

Touch	Day	Job of the message	Why this gap
Email 1	0	State the problem and the reason you picked them	n/a
Follow up 1	3	Add one new fact to the same thread	Long enough not to crowd, short enough that email 1 is still findable
Follow up 2	7	Proof point plus a smaller ask	A new inbox week, and a fresh chance at a quiet morning
Follow up 3	14	Different angle on the problem	Two weeks is enough for priorities to have shifted
Follow up 4	30	Permission to close the loop	A month later, budget and staffing may look different

Treat the days as defaults, not law. Three notes on running it:

1. Selling to executives at larger companies: stretch every gap by roughly 50 percent and cut follow up 3. Senior inboxes punish frequency more than they reward persistence.
2. Working a trigger event such as a funding round or a job posting: compress the front of the sequence to days 0, 2 and 5. The trigger has a shelf life and the schedule should respect it.
3. Replying mid-sequence changes everything: any reply, even a “not now,” ends the automated sequence immediately. From that point the thread is a conversation, and conversations run on response speed, not calendars.

Send every follow up in the same thread. A fresh thread throws away the context the prospect needs to say yes, and forces them to search their inbox to remember who you are. They will not.

What each follow up should say

The rule for every follow up: it must contain a reason to reply that did not exist in the previous email. New fact, new proof, new angle, or a smaller ask. If you cannot point at the new thing, the message is a bump, and bumps get deleted.

Here are five templates that hold up in the field, each with the reason it works.

Follow up 1, day 3: the new fact.

One thing I left out that might make this relevant now: [specific, checkable fact about their company, e.g. “you posted three SDR roles last week”]. Does that change the picture on [problem] for you?

It works because it proves you are paying attention to them specifically, and it hands them a question that takes one line to answer.

Follow up 2, day 7: proof plus a micro-ask.

[Peer company] had the same issue with [problem] and got to [specific result] within [timeframe]. Happy to send the two-line version of how they did it. Useful?

The ask is deliberately tiny. “Send me the two-line version” is a much easier yes than “book 30 minutes,” and a yes to anything restarts the thread.

Follow up 3, day 14: the corrected hypothesis.

Different angle in case my first one missed: most [role] teams we talk to are not actually struggling with [assumed problem], it is [adjacent problem] that hurts. Which is closer to true for you?

Admitting your first guess might be wrong is disarming, and the either-or question is easier to answer than an open one.

Follow up 4, day 30: permission to close.

Closing the loop on this. If [problem] is not a priority this quarter, say the word and I will stop here. If it is just timing, when should I check back?

The easy exit is the whole point. A surprising share of “not now, try me in Q4” replies come from exactly this message, and a dated deferral is a real pipeline asset.

The wrong-person variant, usable at any step:

If this landed with the wrong person, who owns [area] on your side? One name is all I need.

Referral asks have high compliance because they cost the reader five seconds and let them be helpful without committing to anything.

The mistakes that kill threads

The same handful of errors shows up in almost every sequence audit.

“Just bumping this up.” The classic. It adds zero information, restates a request the prospect already declined by silence, and tells them your time matters more than theirs. Same family: “floating this to the top of your inbox” and “in case you missed my last email.” Delete on sight.

The guilt trip. “I’ve reached out three times now” is an accusation, not a message. Nobody has ever booked a meeting out of shame.

The fake re: and the fake forward. Subject-line tricks that fake an existing relationship earn the open and lose the reader in the first second. The click was never the goal; the reply was.

New thread per touch. Covered above, worth repeating: it destroys context and multiplies your entries in their inbox, which is exactly how spam complaints happen.

Following up too fast. A next-day follow up reads as pressure. Two to three business days is the floor.

Attachments and link stacks. Every link and attachment in a cold thread is a spam-filter risk and a cognitive tax. Plain text, one idea, one question.

When to stop, and what stopping means

Stop after the permission-to-close message. In the sequences we see, touches beyond the fifth collect low single-digit percentages of replies while carrying most of the complaint risk. Persistence past that point is not grit, it is list denial.

Stop immediately, regardless of step, on any of these: a clear no, an unsubscribe request, a hard bounce, or an out-of-office that names a better contact. Two of those four are legal matters, not judgment calls.

And be precise about what stopping means. It ends the sequence, not the account. A contact who ignored five emails in March is a fresh contact in September, provided you return with a genuinely new angle rather than the same sequence re-armed. Park them for a quarter, watch for triggers, and restart when something real changes.

When email is exhausted, change the channel

A finished sequence tells you the channel failed, not the account. The same person who filtered five emails will often accept a LinkedIn connection request the same week, because the context is different: your face, your title and your mutual connections are doing work that a sender name never could.

The move is simple. After the day-30 breakup, wait a few days, then send a connection request. Do not reference the ignored emails; nobody enjoys “I emailed you five times.” Open with the same problem hypothesis in the new context and let the thread start clean. How to run that motion well, from positioning to first message, is covered in our LinkedIn outreach strategy guide, and the wider system around it in the complete guide to LinkedIn outbound in 2026.

One warning from the data: replies on the new channel arrive at conversational speed, and they expect answers at conversational speed. A prospect who answers a LinkedIn message and waits two days for your response is colder than one who never replied at all. The numbers behind that are in why response time decides who books the meeting. This is the gap an AI reply layer exists to close: a tool like Replaiy drafts or sends the response in your own voice while the prospect is still online, instead of when a rep next opens the tab. If you are weighing that category, start with what an AI SDR actually does.

The short version

Write the follow ups before you write the opener, because they will earn most of your replies. Run five total touches on days 0, 3, 7, 14 and 30. Give every message one new reason to reply and make each one shorter than the last. Kill the bumps, stop at five, and when the sequence ends, take the thread to LinkedIn and answer fast when it comes back to life.