Why your cold email gets ignored (and what actually gets a reply)
Most cold email fails for one reason: it could have gone to anybody. Here is the test that separates a message that gets a reply from one that gets deleted.
Open your sent folder and read your last cold email. Now ask one question: could this exact message have gone to two hundred other companies without changing a word?
If the answer is yes, you already know why nobody replied.
The mistake is not the template, it is the trigger
Most advice about cold email is about the writing. Shorter subject lines. Fewer links. A softer call to action. All of that is real, and none of it matters if the message has no reason to exist.
A reply happens when the reader thinks "how did they know that." Not "what a well-written email." The difference is whether something actually happened at that company that made your message make sense this week.
Things that qualify as a trigger:
- They posted a job for the role that owns the problem you solve.
- They shipped something that creates the problem you solve.
- Somebody there asked a public question you can genuinely answer.
- They raised money, which means budgets are being set right now.
- A competitor of theirs did something they will be asked about on Monday.
Things that do not qualify:
- They exist and are in your target industry.
- They match a filter you set in a lead database.
- Their headcount grew, according to a number nobody at the company recognises.
The first list gives you a first sentence you could not have written about anyone else. The second gives you a mail merge.
Write the first sentence last
Here is a practical ordering trick. Do not start with the greeting. Start by writing the one sentence that proves you looked. Then build the rest of the message around it.
That sentence has to name a specific, checkable fact. Not "I saw you're growing fast," which is a guess dressed up as research. Something closer to: "You've had a structural engineer role open since March and three of the four projects on your site are mid-permit."
If you cannot write that sentence, you do not have an email. You have a list entry. Skip it and spend the time on someone you can write it for.
Say what you want, once
The second failure mode is the ask. Cold emails routinely ask for a thirty-minute call, then hedge it with "or whatever works for you," then add "no pressure either way," then attach a calendar link, a deck and a case study.
That reads as someone who does not believe the reader will say yes. Pick one ask, make it small, and make it easy to answer in a sentence. A question the reader can answer from their phone in the queue for coffee beats a meeting request every time, because answering it does not require a decision about their calendar.
Good asks are questions only that person can answer. "Are you handling the DOB filings in-house right now, or is that going out?" invites a one-line reply. "Would you be open to a quick call to explore synergies?" invites nothing.
The length rule that actually holds
Sixty to a hundred and twenty words. Not because short is magic, but because the constraint forces you to cut everything that is not the trigger, the relevance and the ask. When people write four hundred words, three hundred of them are about the sender.
Delete every sentence that describes your company. The reader can click your signature. What they cannot do is work out why you emailed them specifically, and that is the only job the message has.
What follow-ups are for
A follow-up that says "just bumping this to the top of your inbox" adds nothing and costs you the reply you might have got later. Here are four follow-ups that carry something new, with timing. Each follow-up should carry a new angle: a different problem, a different piece of evidence, a shorter question. Three of those, spaced across two weeks, will outperform seven bumps.
Then stop. The people who were never going to reply are not going to reply on the eighth attempt, and you will have taught your domain reputation something you cannot unteach quickly.
The uncomfortable maths
Personalized outreach does not scale the way spreadsheet outreach does, and that is the point. It is also exactly the trade-off an AI sales agent is sold as fixing. Fifty messages that each pass the "could this have gone to anyone" test will beat a thousand that do not, and they will not burn your sending domain doing it.
The work is not the writing. The work is finding the fifty.