Cold email follow up: 5 templates and when to stop
Cold email follow up templates, how many to send and how long to wait, plus how to keep nudges in one thread and stop the moment someone replies.

A good cold email follow up is short, goes out in the same thread, adds one new thing (a piece of value or one clear question) and stops the moment the person answers. Space the nudges a few days apart, set a fixed number before you start, and end with a polite break-up message.
Two things go wrong most often: nudging someone who already replied, and nudging forever. This post gives you the wording first (five templates you can copy), then the mechanics that stop the sequence on its own. It's for founders, indie hackers, marketers and developers who send first-touch email, and for people building agents that do it for them.
How long to wait and how many follow-ups to send
There is no magic number, and we won't quote reply rates. Here is a practical cadence, offered as guidance and nothing more:
| Message | When | Job |
|---|---|---|
| First email | Day 0 | One idea, one question |
| Follow-up 1 | About 3 days later | Quick bump or new value |
| Follow-up 2 | 4 to 5 days after that | A single yes/no question, or a redirect |
| Break-up | About a week after that | Close the loop |
That is three follow-ups at most. Pick your own cap, but write it down before the sequence starts. A cap you decide on while staring at an empty inbox tends to creep upward.
Make the cap countable. The number of outbound messages in the thread is the number of nudges already sent. If your tooling can read the thread transcript, it never has to remember anything. It counts.
5 cold email follow up templates
Each template assumes you reply in the original thread, so the subject keeps its "Re:" and the recipient sees the history. Don't retype the subject line. Don't type a fake "Re:" onto a brand-new email either.
The "just circling back" graveyard is crowded. These try to stay out of it.
Template 1: the quick bump
When to use it: first nudge, about three days after the original, when you have nothing new yet.
Hi {{first_name}},
Quick bump on the note below. Is this worth a short call this month?
{{sender_name}}Subject: reply in the thread. Keep the original subject.
Template 2: the new-value follow-up
When to use it: when you have something relevant to add, such as a link, a short result or a number you can source.
Hi {{first_name}},
Since I wrote, {{new_thing}}. Here's the short version: {{link_or_one_line}}.
If {{problem}} is on your list this quarter, I can send over how {{similar_company}} handled it. Want it?
{{sender_name}}Subject: same thread. Only cite a figure you can point to a source for. If you can't, leave it out.
Template 3: the no response email reminder with one yes/no question
When to use it: second nudge. It lowers the cost of answering to a single word.
Hi {{first_name}},
One question, so you don't have to write an essay: is {{problem}} something you're working on right now? Yes or no is plenty.
{{sender_name}}Subject: same thread. One question only. Stacked questions get ignored.
Template 4: the right-person redirect
When to use it: when silence might mean you wrote to the wrong person.
Hi {{first_name}},
I may have this one wrong. If {{topic}} isn't yours, could you point me to whoever owns it? A name is enough.
{{sender_name}}Subject: same thread. This gives the recipient an easy way out that isn't "no".
Template 5: the break-up email
When to use it: last message in the sequence. It closes the loop, and then you actually stop.
Hi {{first_name}},
I haven't heard back, so I'll assume the timing is off and stop here. If that changes, reply to this thread and I'll pick it back up.
All the best,
{{sender_name}}Subject: same thread. Don't add a sixth email after the break-up. That defeats the purpose.
Write a no response email reminder that does not annoy
A follow up email after no response has one job: make the next step easy. Edit with these rules:
- One ask. One question or one link, not both.
- Short. If it needs scrolling, cut it.
- No guilt. Skip "I know you're busy" and "I'm sure you saw my last email."
- No fake threads. Don't put "Re:" on a brand-new email to pretend there was a conversation.
- An easy way out. "Not now" should be a valid answer.
What to cut: apology openers, "bumping this to the top of your inbox", and stacked questions. They add length and subtract goodwill.
Make opting out easy, and honor it right away. If someone says stop, the sequence ends, including every nudge still scheduled. Rules about cold email differ by country, so check the ones that apply to you. The stopping section below covers how to make that automatic.
Keep follow-ups in the same thread
Two reasons.
- The recipient sees the whole conversation. Your first message is right there under the nudge. Nobody has to hunt for context.
- Your tooling can tell which message a reply answers. A reply that lands in a separate thread is a reply nobody connects to anything.
On SendRaven, a reply is joined to the original message on Message-ID (the unique ID every email carries), never on subject. Subject lines get edited, translated and prefixed. Message-ID doesn't change.
To send a follow-up into the same thread, set reply_to_message_id on POST /v1/emails to your last message. The text field of an inbound message arrives with quoted history and signature stripped, with raw_text as a fallback.
For the details, read about threads joined on Message-ID, not subject.
Stop the sequence when someone replies
Here is the part most template lists skip. "Follow up if they don't answer" sounds like one instruction. It is three questions a plain send API cannot answer:
- Did they reply?
- How many nudges have already gone out?
- Was that reply an out-of-office?
Two approaches work. Both use the same primitives.
Approach A: schedule up front, cancel on reply
Send the first email and keep its id and thread_id. Right away, schedule the follow-up on the same thread. scheduled_at accepts plain words like "in 3 days". If the reply arrives first, cancel the scheduled send.
The three calls, in order:
POST /v1/emails
-> 202 { "status": "sent", "thread_id": "1276…" }
POST /v1/emails
{ "reply_to_message_id": "0d87…", "scheduled_at": "in 3 days", … }
-> 202 { "status": "scheduled", … }
DELETE /v1/emails/047c…The first call sends the email. The second schedules the nudge in the same thread. The third withdraws the nudge if the answer arrived first.
Because the nudge is scheduled up front, it goes out on time even if your agent isn't running on day three. The full walkthrough is in the recipe to schedule the follow-up and cancel it when the reply arrives.
One edge case from that recipe: if the reply arrives while the follow-up is being sent, cancel_scheduled_email answers 409 invalid_state and names the status. Read the status rather than assuming. If the nudge already went, so be it. If the thread shows a reply, handle that first.
Approach B: check the thread at send time
Keep the sequence in your own scheduler and look at the thread before each send. GET /v1/threads/:id returns the thread with awaiting_reply, handled_at and the messages.
awaiting_reply turns true when a person answers, and it stays false for an out-of-office. GET /v1/threads?awaiting_reply=true lists the conversations waiting on you, and POST /v1/threads/:id/handled clears the flag once you've dealt with one.
The use-case page describes the send-time check like this: if awaiting_reply is still false and the delay since your last outbound message has passed, send the follow-up with reply_to_message_id set to your last message, so it lands in the same thread. Add your cap to that:
- No person has answered, so
awaiting_replyhas not flipped. - The delay since your last outbound message has passed.
- You are under your cap, counted from the thread's transcript on every pass, never from memory.
Multi-step sequences
If you run sequences as automations, they start from an API call, a new contact or an event. They have five automatic exits: unsubscribe, reply, suppression, topic opt-out and tag rule. Those end a multi-step sequence without you writing the stop logic.
Let an agent run the follow-ups, with limits
An agent can run a sequence, but split the labor. Plain code decides when to act and how many nudges remain. The model only writes the words.
The follow-up agent post makes the same split. It calls the mailbox half "everything the model should never be trusted with": whether the person has already answered, how many follow-ups have gone out, who the email goes to, and whether a "reply" was an out-of-office.
Some guardrails to add, per API key:
- Idempotency key. Derive each send's key from the task and the follow-up number. A retry with the same key delivers once.
- Daily send limit. Counted in recipients per UTC day. Hit it and you get
429 daily_limit. - Recipient allowlist. Addresses or domains. Anything else returns
403 recipient_not_allowed. requires_approval. Sends returnpending_approvalwith anapproval_id, and a person releases them from a queue. Held drafts expire after 72 hours, and a key cannot approve its own drafts.
Autonomous does not have to mean unsupervised.
To see it working, read about a follow-up agent that stops when someone replies, or the use case page on AI agent email follow-ups until someone replies. Runnable examples live in the examples repo. You can also connect your agent to the remote MCP server at https://mcp.sendraven.ai/mcp, or call the REST API directly.
Mistakes that make follow-ups backfire
- Following up after a reply or an unsubscribe. Unsubscribe should cancel what is queued, not just suppress the address.
- Changing the subject line and splitting the thread. Now your history is gone, and so is your tooling's ability to match a reply to a message.
- Sending the same text three times. A model asked to write the same email twice writes two different emails. A human pasting one template three times writes the same email three times. Both are a problem.
- Treating an out-of-office as an answer. It isn't one. Don't tell your team a prospect replied.
- Treating an out-of-office as silence you can ignore. It also isn't a refusal. If the auto-reply says when the person is back, consider moving the nudge until after that date.
On deliverability: we won't promise anything here, and no honest template can. What you can control is plain. Send relevant mail, keep it short, honor opt-outs quickly, and stop when told to.
If you also send no-reply notices, see the no-reply email template. A reply-friendly address and a follow-up sequence work better together.
Copy the templates, then let the sequence stop itself
Wording gets you the first reply. Mechanics keep you from the embarrassing ones. Copy the templates above, then start free and follow the recipe for following up when nobody replies. Or connect the MCP server to your agent.
Free is $0 a month with 3,000 emails, unlimited contacts and inbound included. Replies are free and don't count toward usage. You pay for emails sent, never for contacts stored. A verified payment method is required before any sending, including on Free.
FAQ
How long should you wait before following up on a cold email? About three days is a reasonable first gap, then four or five days between later nudges. Treat that as guidance, not a statistic. Pick a rhythm and keep it consistent.
How many follow-ups should you send on a cold email? A small fixed number, such as three, ending with a break-up message. Decide the cap before the sequence starts, and count it from the number of outbound messages in the thread.
What do you write in a follow-up email after no response? Add one new thing: a relevant link, a sourced number or a single yes/no question. Keep it short and in the same thread. The five templates above cover the usual cases.
Should a follow-up be in the same thread or a new email? Same thread. The recipient sees the whole conversation, and your tooling can match replies to messages. On SendRaven, set reply_to_message_id to your last message.
How do you write a no response email reminder without sounding pushy? Make one ask, skip the guilt and give an easy way out. Cut apology openers, "bumping this to the top of your inbox" and stacked questions. "Not now" should be an acceptable answer.
When should you stop following up? When the person replies, unsubscribes or you hit your cap. After the break-up message, send nothing more.
How do you stop a follow-up from sending after the person replies? Either schedule the nudge up front and cancel it with DELETE /v1/emails/:id when a reply arrives, or check the thread right before sending. If you use automations, reply is one of the five automatic exits.
Does an out-of-office count as a reply? No. On SendRaven, awaiting_reply turns true when a person answers and stays false for an out-of-office, so the sequence doesn't mistake an auto-reply for an answer. You may still want to move the nudge until after the person is back.
Can an AI agent send cold email follow-ups safely? Yes, with limits. Let plain code decide when to act and how many nudges remain, and let the model write only the words. Add per-key guardrails: a daily send limit, a recipient allowlist and a requires_approval hold. Use an idempotency key so a retry delivers once.



