After this lesson you should be able to
- Set a follow-up policy and hold to it across a campaign
- Write a second email that adds something rather than nagging
- Recognize the signals that mean stop now
- Sequence a campaign in batches instead of one large send
Why follow-up works at all
A first email arrives at an arbitrary moment. The recipient is in a meeting, on deadline, on holiday, or looking at a phone with wet hands. Non-reply is usually not rejection; it is timing. A single well-judged follow-up converts a meaningful share of the people who intended to reply and did not, and in most campaigns it produces a substantial fraction of the total response.
That is the honest case for following up, and it is a good one. The dishonest extension of it — if one follow-up helps, five will help five times as much — is where the practice turned into the thing everyone hates. Each additional message converts fewer people and annoys more, and the curve crosses over quickly.
There is also an asymmetry worth understanding. The cost of a follow-up is not borne evenly. To the recipient who was going to reply, it is a helpful nudge. To the recipient who already decided no, it is a demand to spend attention declining something twice. You cannot tell the two apart in advance, which is exactly why the number has to be capped by policy rather than judged case by case.
How many, and at what spacing
My policy, which I would defend to anyone: one follow-up as standard, a second only when there is something new to say, and never a third.
On spacing:
- First follow-up: four to seven days after the original. Long enough that the first message has left the visible inbox, short enough that the context is still recoverable. Under three days reads as impatience. Over two weeks and you are effectively sending a new cold email to someone who now recognizes your name from a message they ignored.
- Second follow-up, if it happens at all: ten to fourteen days later, and only if you have new information — the study published, the tool launched, a related page of theirs also broke. A second follow-up with nothing new is a nag.
- Then stop. Mark it closed and move on.
Two things to avoid regardless of count. Do not follow up on a Friday afternoon or over a weekend; it lands in the Monday flood. And do not send the follow-up at exactly the same hour on exactly the same weekday interval, because a perfectly regular cadence is a machine signature and recipients notice.
One exception to the cap. When someone has replied and the thread is live, normal professional correspondence rules apply and the cap is irrelevant. The limit is on unanswered cold messages, not on conversations.
What a follow-up should actually say
The default follow-up is some variant of Just bumping this up your inbox or Did you get a chance to look at this? Both are pure cost to the recipient: they add no information and ask for the same thing again.
A follow-up should do one of three things.
- Add something. A second broken link on the same page. A related page of theirs with the same issue. A fact that strengthens the case. Anything that makes the second email independently worth opening.
- Reduce the ask. If the first email asked them to consider adding a resource, the second can ask only whether the page is still maintained. A smaller question is easier to answer, and an answer of any kind restarts the conversation.
- Close cleanly. Tell them you will not follow up again, and mean it. This is more effective than it sounds, partly because it is rare and partly because it converts a decision people were deferring into one they make now.
A workable closing follow-up runs about three sentences: a one-line reminder of what it was about, the new information or the reduced ask, and an explicit last-message note. Reply to your own original message so the thread carries the context, rather than starting fresh and forcing them to reconstruct it.
Never imply a prior relationship that does not exist. As per my last email, following up as discussed and I wanted to circle back on our conversation are all false when there was no conversation, and the recipient knows it.
When stopping is the correct call
Stopping is a professional decision, not a failure. These are the signals, and each is sufficient on its own.
- They said no. Any form of no ends the sequence, including a soft no. Not right now means diary it for later, not follow up next week.
- Your follow-up policy is exhausted. Two messages, no reply, closed. The policy exists precisely so that this decision is not made in the moment by someone under pressure to hit a number.
- The reply was irritated. Stop, apologize briefly if a reply is warranted, and remove them permanently. Do not defend the outreach.
- They asked to be removed. Remove immediately and completely, across every list and every sending address. This is a legal obligation in most jurisdictions and a basic courtesy in all of them.
- The prospect turned out to be unqualified. If the second look reveals the page is abandoned or the site sells links, stop regardless of where you are in the sequence.
- You are following up because the numbers look thin. Be honest about this one. It is the most common reason a fourth email gets sent, and it is never a good reason.
Record the reason you stopped, not just the fact. A prospect closed because the contact left is a prospect worth revisiting when you find the new one. A prospect closed because they told you to go away is not, and the record is what stops a colleague re-adding them in a year.
Sequencing across the campaign
Sequencing is the campaign-level version of the same question: not what you send one person, but the order and rate at which you touch the whole list.
Send in batches, never all at once. Twenty to fifty per address per day is a reasonable working ceiling for genuine outreach, and staying well under whatever your infrastructure permits is deliberate rather than timid — Lesson 5.4 explains what happens to a domain that ramps too fast.
Start with a pilot. The first batch is a test of the pitch, not a campaign. Send twenty to your Tier B prospects, wait a week, and read what comes back. Refusals with reasons tell you the pitch works and the fit does not. Silence with no bounces tells you to check delivery. Only scale a pitch that has produced replies.
Hold the best prospects back. Tier A — the ten prospects that would genuinely matter — should not be tested on. Pitch them individually once the message has been validated on the middle of the list.
Stagger the follow-ups too. If you send a hundred first emails on Monday, a hundred follow-ups fire on the following Monday. That is a machine pattern and it lands as one. Spread the original sends and the follow-ups spread themselves.
Keep the batch small enough to answer. The failure mode nobody plans for is a batch that works: forty replies arriving in two days, several of which want a conversation, and nobody available to have it. A reply left for a week is a relationship discarded after you paid for it.
Recording outcomes
Every message in a sequence produces an outcome, and most systems record only two of them: link or no link. That throws away most of the value.
Record at minimum: the date sent, the date of any reply, the outcome category, and the reason in the prospect's own words where they gave one. Useful outcome categories are narrower than people expect — linked, agreed but not yet live, refused with reason, refused without reason, wrong contact, asked to be contacted later, asked never to be contacted, no response, bounced.
Two of those deserve special handling. Agreed but not yet live needs a diary date, because a meaningful share of agreed links never actually get published and nobody notices. Asked to be contacted later is the most valuable row in the sheet and the one most reliably lost when a campaign ends and the spreadsheet is archived.
Bounces need their own column and their own attention, because they are a deliverability signal rather than an outreach one. A rising bounce rate mid-campaign means your address verification failed and you should pause sending, not push through. That is the bridge to the next lesson.
Questions
How many follow-up emails should I send?
One as standard. A second only when you have genuinely new information — the study published, another broken link on the same page. Never a third. The conversion curve falls steeply after the first follow-up while the annoyance curve rises, and the crossover happens fast. Capping it by written policy stops the decision being made by whoever is behind on numbers.
How long should I wait before following up?
Four to seven days after the original. Under three days reads as impatience; beyond two weeks the context is gone and you are effectively cold-emailing someone who already ignored you once. If a second follow-up is warranted, leave ten to fourteen days, and avoid Friday afternoons and weekends entirely.
Is it worth telling someone it is my last email?
Yes, when it is true. An explicit closing note is rare enough to stand out, it removes any sense of an open loop, and it converts a decision people were deferring into one they make now. It only works if you honor it — sending again after saying you would not is worse than not saying it.
What should I do when someone replies angrily?
Stop immediately, remove them from every list and sending address, and if a reply is warranted keep it to one line of apology with no defense of the outreach. Do not explain your process or argue the relevance. Then record the reason so nobody on your team re-adds them next year from an old export.