The B2B cold email follow-up sequence: how many touches, and what to send in each

Most outbound programmes send one email, wait a week, then send two versions of the same reminder and declare the list dead. The first email almost never does the work. The sequence does, and only when each message carries a separate job instead of repeating the first one louder.

Here is the follow-up structure I use for B2B outbound at Sail Force Marketing: the scripts, the numbers that tell you which part is broken, and the technical rules that decide whether any of it reaches an inbox in 2026.

Why the follow-up carries the sequence, not the first email

A buyer replies when your message meets a problem they already have, in a week when they have budget and attention for it. You cannot time that from outside. What you can do is stay present across a window wide enough to intersect it, without becoming the sender they mute.

That is the whole design problem. Persistence is not the issue, repetition is. Seven messages that each say something new read as diligence. Seven messages that each say "just checking in" read as a robot with your name attached. The difference is not tone, it is whether the message contains information the reader did not have before opening it.

One thing has to be true before any of this matters: the list has to be worth the sequence. If the account cannot buy, does not have the problem, or has no route to a decision, a better follow-up will only produce a more polite rejection. I wrote the exact five-part standard I use in the qualification bar that decides whether your pipeline is real. Run that first. A sequence is an amplifier, and an amplifier applied to a weak list gives you louder silence.

The seven-touch sequence, day by day

Four working weeks, seven touches, decreasing length. Each one has a single job:

1. Day 1, the hypothesis. One specific observation about their operation, one hypothesis about what it costs them, one small ask. Under 90 words.
2. Day 3, the evidence. Same thread. One artefact that stands alone: a short teardown, a structural comparison, a two-line breakdown of how a comparable operation handles the same problem. The ask is a yes or no, nothing larger.
3. Day 6, the angle change. You may have aimed at the wrong pain. Same company, different problem, stated in one sentence. If touch 1 was pipeline volume, touch 3 is conversion or cost per meeting.
4. Day 9, the sideways move. New thread, new subject, second contact in the organisation. Reference the business, never the silence. Do not copy the first contact in.
5. Day 14, the short one. Three lines. Ask whether this sits with them at all, or with someone else. Routing questions get answered when pitches do not.
6. Day 18, the give. Something they keep whether or not they reply: a checklist, a benchmark, the questions to ask their current supplier. No ask at all.
7. Day 23, closing the file. A plain statement that you are stopping, and one line describing the condition under which it would be worth reopening.

Fourteen touches do not beat seven, except on a spreadsheet that counts sends rather than reputation. After the third message with nothing new in it, you are training a buyer to recognise your address and skip it, and that cost follows you into every future campaign.

A follow-up that adds nothing new is not persistence. It is noise with your name on it.

The rules each message has to pass

• One ask, and the same ask throughout the sequence. Two asks in one email halves the reply rate on both.
• New information in every message. If you cannot name what is new, the message is not ready to send.
• Under 90 words for touches 1 to 3, under 50 after that.
• No "circling back", no "bumping this up your inbox", no "did you see my last email". All three tell the reader the message is about you.
• No false familiarity. Do not open with a compliment you found on a website.

Three scripts you can lift

Day 3, the evidence touch. "Following the note on Monday: most teams in your position run outbound off one shared mailbox, which caps you at roughly 40 sends a day before deliverability starts to slip. The usual fix is three mailboxes on a separate sending domain. Worth a look at your setup, or is this already handled?"

Day 14, the short one. "Quick one: is outbound pipeline yours, or does it sit with someone else there? Happy to send this to the right person and leave you out of it."

Day 23, closing the file. "I will stop here. If the picture changes and you end up needing to add meetings without adding headcount, reply to this and I will pick it up from where we left off. Either way, good luck with the quarter."

Notice what the last one does not do. It does not say "I will assume you are not interested", and it does not offer a final discount. Manufactured guilt converts once, then poisons the relationship.

The numbers that tell you which part is broken

Reply rate on its own diagnoses nothing. Read the sequence as a chain and find the first link that fails:

• Bounces above 2 per cent. A data problem, not a copy problem. Stop the campaign, reverify the list, restart. Sending into bad data damages the domain that carries every later campaign.
• Almost no replies of any kind. Suspect delivery before you suspect the writing. Send a test to seeded mailboxes across Gmail, Outlook and a corporate Microsoft 365 tenant and find out where you actually land.
• Replies arriving, all negative. The targeting is wrong, or the opening claim is generic. Change the account list before you change a word of copy.
• Positive replies but no meetings booked. The ask is too big for the message. Drop from "30 minutes" to a single question they can answer in one line.
• Meetings booked but no second meetings. You are qualifying too loosely at the top. That is a bar problem, and no sequence tuning will fix it.

Change one variable at a time and hold a real sample before judging it. I do not read a result under about 150 contacts per variant, and I do not change two things in the same fortnight. Most outbound programmes are not underperforming, they are being rewritten faster than they can produce evidence.

Deliverability decides the sequence before the copy does

The best written sequence in your industry earns nothing from a spam folder, and the rules hardened considerably over the last two years.

Microsoft now requires SPF, DKIM and DMARC on domains sending roughly 5,000 or more messages a day to Outlook.com, Hotmail.com and Live.com addresses, enforced from 5 May 2025, with non-compliant mail rejected outright rather than filtered (dmarcian's summary of the policy). Google and Yahoo set a similar bar for bulk senders: authentication, one-click unsubscribe under RFC 8058 honoured within two days, and a reported spam rate kept under 0.3 per cent, with under 0.1 per cent the practical target (Mailgun's breakdown).

What that means operationally: send outbound from a separate domain, never your primary one. Warm each mailbox before it carries volume. Cap sends per mailbox and add mailboxes rather than raising the cap. Put a one-click unsubscribe in cold email even where B2B rules in your market would let you skip it, and honour it the same day. Across Europe, Africa and the US the local consent regimes differ, and the safe operating floor is the strictest of the markets you touch.

Outbound asks for time that belongs to someone else. Ask precisely, ask a small number of times, and stop when you said you would.

I treat a stranger's attention as borrowed rather than owed. That conviction is not decoration on the method, it is the reason the sequence ends at seven and the last message actually means it.

The failure modes I see most often

Automating the sideways move. Touch 4 sent to five people at one company in the same week does not look like coverage, it looks like a mail merge, and it usually ends with an internal forward that closes the account.

Treating the breakup email as a tactic. If you announce you are closing the file and message again three weeks later with a new subject line, you have taught the buyer that your word is a device.

Rewriting the first email forever. Teams polish touch 1 for months while touches 2 to 7 stay as filler. The compounding is in the back half of the sequence, because that is where nearly everyone else has already stopped.

Reusing the same angle next quarter. If you re-enter an account, re-enter it with a different hypothesis and a different contact, or you are simply restarting a conversation that already failed.

What to do this week

Take 40 accounts that clear your qualification bar. Write touches 1, 3 and 7 today, because those three carry most of the outcome. Set up the sending domain and authentication before the first send, not after the first campaign disappoints you. Run the full 23-day cycle without editing mid-flight, then read the chain above and fix the first link that fails.

Forty accounts done properly will teach you more than four hundred done at speed, and the sequence you end up with is an asset you keep. More on building revenue systems that hold under pressure sits in the Sales Growth library.