How many cold emails die because nobody ever follows up? For most sellers the number is high, because the first send is usually the only send. Backlinko's study of 12 million outreach emails found only 8.5 percent ever get a reply, and one follow-up lifts replies by 65.8 percent. The replies are there. Most arrive after senders stop.

The answer is a cold email follow up sequence that runs itself: five touches over three weeks, each with one job, scheduled in advance and switched off the moment someone answers. This walkthrough builds that sequence for a founder or solo seller on a normal Gmail inbox. ChatGPT writes the drafts, Zapier sends them, and a Google Sheet decides who gets which email on which day. The work before the first send lives in our companion guide to cold email personalisation with Claude, Hunter and n8n. Here we cover the cadence: timing, the five touches, reply detection and stopping the machine.

Why most cold email dies after the first send

Send one email and you only reach the people interested on the day it lands. Everyone else decides the moment is wrong and never hears from you again. The platform datasets show how much reply volume sits after the first touch. Woodpecker's analysis of 20 million cold emails found campaigns with 4 to 7 emails replied at 27 percent, against 9 percent for 1 to 3. Saleshandy's analysis of 53 million cold emails found 44 percent of positive replies arrive on follow-ups, and the first follow-up alone produces 26 percent. Belkins' dataset of 7.5 million emails found over 53 percent of email-sourced meetings came from step 3 or later.

Most buyers do not ignore you. They open the email, decide the timing is wrong, and move on. A touch every few days raises the chance you are in front of them during the week the problem becomes urgent, which is why five touches beat a single send even though each later message earns a smaller reply share.

What you'll need to run this workflow

Everything here runs on accounts you most likely already hold. A Zap is what Zapier calls one automation, and a task is counted each time an action in that automation runs.

If you already run Claude with self-hosted n8n, the same logic works with no task caps. At low volume, run the cadence manually with calendar reminders.

Step 1: Map the five-touch cold email follow up sequence

A follow-up sequence works when every touch has one job and the gaps give the buyer room. Here is the cadence this post builds.

Touch Day after first send The job of that touch Attachment or link
1. Opening email Day 1 One clear reason to reply, tied to their world One link max, to the page you want them to see
2. First follow-up Day 3 A light nudge in the same thread, a new angle if you have one None
3. Second follow-up Day 7 Proof: a similar customer, a concrete result or an example One link to the proof
4. Third follow-up Day 14 A useful resource or observation, and no ask at all One link to the resource
5. Breakup Day 21 A gracious exit that leaves the door open None

Why those days? WarmySender's study of 8.3 million emails compared spacing patterns and found the balanced Day 1, 3, 7, 14, 28 rhythm produced the most replies at 7.1 percent, while daily sending produced fewer at 4.1 percent plus roughly nine times the spam complaints. The pattern is the point: close early while the memory is fresh, then patient.

Give the second touch room

The most common mistake is the second touch arriving too soon. LeadHaste's campaign data found moving it from day 2 to day 3 or 4 lifted that touch's reply rate by roughly 15 to 25 percent. Two days reads as anxious. Three or four reads as organised.

Write the cadence into a sheet before you build anything

Create the tracking sheet now, because the automation reads it later. Columns: first name, company, email, current touch, status, last sent, reply date, notes. Keep status values to a fixed list: waiting, replied, not interested, bounced, out of office. A stray spelling silently stops the machine.

Step 2: Write each touch with ChatGPT

The drafting job is five small emails with different jobs, and ChatGPT is a fast first drafter. Write them in one sitting so the thread has one voice, then edit cold the next day. This prompt produces the whole sequence; swap the square brackets for your details.

Write a five-email cold outreach thread for [name], [role] at [company].
Context: [what the company does], and the specific reason I am emailing: [trigger].
My offer in one sentence: [offer].

Email 1, the opening, 90 to 120 words: one clear reason to reply and one light question.
Email 2, day 3 follow-up, 60 to 90 words: a friendly nudge, no new ask.
Email 3, day 7 follow-up, 60 to 90 words: one piece of proof, like [example customer or result].
Email 4, day 14 follow-up, 60 to 90 words: one useful resource, ask for nothing.
Email 5, day 21 breakup, 40 to 60 words: gracious, leaves the door open.

Same subject line for all five. Plain professional language.
No hype words, no "just checking in", no "I hope this finds you well".

Feed ChatGPT one prospect at a time

Generic sequences get generic replies, which is to say none. Give the model the company name, what the company does and the trigger that makes this person worth emailing now. SalesHandy's ChatGPT prompt library fixes the same inputs: role, ideal customer, pain point, value, tone, word count.

Keep every message under 120 words

A follow-up's job is to be read in ten seconds. A message that only says bumping this adds nothing a busy buyer cares about; each email earns its place with a new fact or reason to reply. Woodpecker's follow-up playbook keeps early messages light and conversational.

Draft all five in one sitting, then edit cold

Writing the whole thread at once keeps the voice consistent, and editing the next day catches the hype words that crept in overnight. Read each draft aloud and cut anything that makes you wince. The breakup email is the one people skip and the one that quietly converts.

Step 3: Automate the sends with Zapier

Now the sequence sends itself: one scheduled automation per touch, and every automation checks the sheet before it sends. Zapier's email automation guide walks through the full spreadsheet-to-email pattern.

Connect the sheet and pick your trigger

In Zapier, create a Zap with Schedule by Zapier as the trigger, set to business days at a time you actually check email. The schedule fires daily; the sheet decides who gets mail. Add a Google Sheets action that finds rows where the status equals the touch you are sending, then a Gmail action sends the mapped email and a final action updates the row to the next waiting touch. Roughly one task runs per email sent, which is why the free tier covers a small list. Zapier's staged follow-up template shows a delay-based alternative for low volumes, no sheet required.

Keep every follow-up in the same thread

A cold sequence should read as one conversation, so touches 2 to 5 reply inside the thread the opening email started. Zapier's template for follow-up replies using the captured thread identifier shows the mapping.

Keep the volume honest

One honest line before you scale: a warm personal inbox that suddenly blasts hundreds of cold emails a day trips spam filters, because volume is the loudest signal a filter sees. Stay at a few dozen sends a day from one address; platform-scale sending is a different game, covered in our guide on marketing emails going to spam.

Step 4: Stop the sequence when someone replies

The moment someone replies, the machine must stop; an automated follow-up to someone who already answered reads as broken. Prospi's sequence guide is blunt: stop all outreach on reply, and honour opt-outs instantly.

Label replies as they arrive

In Gmail, create a label called Sequence Reply and build one Zapier automation that watches for it. Zapier's Gmail integration reference explains that New Labeled Conversation fires the first time a label is added, including when a reply arrives inside a thread, while the plain New Email trigger does not fire on replies within a thread, a gotcha documented in this thread on Zapier's community forum.

Flip the status when the label appears

When the label appears, the automation finds the matching row by email address and sets the status to replied. Zapier ships a ready-made version, an inbound reply automation that updates contact status, and it filters auto-replies such as out-of-office messages, so a holiday responder does not end the sequence.

Let the status gate every later send

Every scheduled Zap from Step 3 starts by looking for rows whose status still says waiting, so a flipped status means nothing left to send. Test the loop: reply from a second address, watch the label fire, confirm the sheet flips, and check the next run skips the row.

Step 5: Track the numbers that matter

Three numbers tell you whether the sequence works: reply rate, positive reply rate and meetings booked. Log them the day you build it, because the baseline starts with the first send.

The reply rate, read per touch and in total

Reply rate is replies divided by sends; watch it per touch to find dead messages and in total to judge the sequence. Benchmarks vary because denominators differ: Instantly's 2026 benchmark report puts the overall average at 3.43 percent with top performers above 10, while Belkins' strict-cold study measured 0.45 percent per total send. Judge your sequence against itself first, then against a range.

The positive rate and the meeting rate

Positive replies want the next step, and the meeting rate is booked meetings divided by positive replies. If replies are healthy but meetings are not, the problem sits downstream in the offer or qualification.

When to rewrite a touch

Give each touch about a hundred sends before judging it. LeadHaste's data puts 89 percent of replies inside the first four touches, so if touch 4 or 5 earns nothing across a hundred sends, rewrite once, then cut. A touch that only collects not interested replies has the wrong angle.

What to do with the replies

Reply fast, within a day or two, while the thread is warm. A positive reply gets a meeting proposal and a calendar link within a few days; information requests go out the same day. A not interested reply gets one short, gracious message and, only if it costs nothing, a single question about timing or fit. Mark every outcome in the status column; the log is what makes the next campaign's second touch smarter.

People who go quiet again are re-engagement territory, and cold emailing them a second time burns the trust you rebuilt; our re-engagement sequence for quiet lists handles that group.

This five-touch machine is the kind of workflow we build at Supernodes: audit the outreach, draft the touches, connect the automation, set the measurement, live in two weeks. If that sounds like your Monday morning, speak with us.

Frequently asked questions

How many cold email follow ups should I send?

Four follow-ups after the opening email, five touches in total, is where the data lands. Woodpecker's 20-million-email analysis found sequences of 4 to 7 emails replied at 27 percent against 9 percent for 1 to 3, and Instantly's 2026 benchmark names 4 to 7 touchpoints as ideal. Treat the gains as cumulative: per-message reply rates fall from 2 to 4 percent at message two to 0.3 to 1 percent by message seven, per RevenueFlow's benchmark review, so later touches catch replies the earlier ones missed.

How long should I wait between cold email follow ups?

Three days after the opening, then roughly a week, then ten days to two weeks, with the breakup around day 21. WarmySender's 8.3-million-email study found the balanced Day 1, 3, 7, 14, 28 spacing produced the most replies at 7.1 percent, while daily sending produced 4.1 percent and roughly nine times the spam complaints.

Should each follow up use a new subject line?

No, keep every touch in the same thread with the same subject line. A new subject reads like a fresh cold email and resets the trust the earlier messages built. Prospi's sequence guide treats same-thread replies as the default, and tools like Mailmeteor send scheduled follow-ups as replies for the same reason.

What do I do when someone replies not interested?

Stop the sequence and send one short, gracious reply. If it costs them nothing, one question about timing or fit gives useful signal, but do not push after a clear no. Mark the contact so they never re-enter the sequence, honour opt-outs quickly, and remember cold email rules differ by region.

Do the free plans cover this cold email follow up sequence?

Yes, at small scale. ChatGPT's free plan drafts all five touches, with limits that change over time. Zapier's Free plan includes 100 tasks a month (help page, September 2026), and five touches to ten prospects costs roughly one task per send. Past that, a paid plan or self-hosted n8n runs the same sequence without task caps.