How to Write Outreach That Gets Replies

Everyone's inbox is now full of your competitors

Let us start with the uncomfortable part. The reason cold outreach works worse than it did three years ago is that AI made it free to send. When one person could write forty emails a day, forty was the ceiling. Now one person can send four thousand, and roughly four thousand people worked that out at the same time.

Your buyer's inbox is the crime scene. They receive twenty near-identical messages a week, all opening with a compliment about their recent growth, all pivoting to a product they did not ask about. They have learned to delete on sight, and they now delete your genuinely good message along with the rest, because from the outside it looks the same.

This is not a moral lecture, though there is a moral point in here. It is a practical one. Volume outreach has stopped working, and AI is what stopped it.

What generic actually looks like

You can spot AI outreach in under a second, and so can your buyer. Here is the same prospect, approached two ways.

Notice what the good version does not do. It does not compliment. It does not ask for fifteen minutes. It does not describe the product. It says one true, checkable thing and asks a question a real operations lead would have an opinion about.

Use AI to research and sharpen, not to blast

Here is the split that keeps you honest and gets better numbers at the same time.

AI is excellent at helping you understand a company, generate angles, tighten your writing, cut your word count and tell you when your message sounds like everyone else's. AI is bad at knowing anything true about a specific stranger, and terrible at deciding who deserves a message.

So the humans do targeting and the facts. The AI does the drafting and the cutting.

Here is a verified fact about a company: [the thing you actually checked].
We sell [product] and we help with [the problem it solves].
Give me three possible angles connecting that fact to a problem this
company might now have. For each, tell me the assumption I would be
making, so I know what I am risking if I am wrong.
Do not write an email yet.

Write a cold email under 90 words.
Recipient: [name], [job title] at [company].
The one real, verified fact I am referencing: [fact].
My angle: [the assumption you chose].
Rules:

Open with the fact, not with a compliment
Do not use: hope this finds you well, impressed by, reach out,
circle back, synergies, quick chat, touch base, game-changer
Do not describe our product features
End with one specific question, not a meeting request
Write like one working person to another

The generic detector

Before sending, run one more pass. This is the single highest value prompt in the lesson because it is the one your buyer is unconsciously running.

Here is a cold email I am about to send: [paste it].
Answer three things bluntly:

Could this exact email be sent to 500 other companies with only the
name changed? Quote the lines that give it away.
What would a busy person feel in the first four seconds of reading it?
Cut it to the shortest version that still says something specific.
Do not be encouraging. I want the honest read.

If the answer to question one is yes, do not send it. Either find something real to say or accept that this prospect is not worth the research, which is a legitimate answer and a much better use of your week than sending three hundred messages nobody opens.

The line, stated plainly

Using AI to prepare better and write more clearly is good. Using it to fake personal knowledge you do not have, or to send hundreds of near-identical messages, damages your reputation and does not work anyway.

Both halves of that matter. The ethical case and the commercial case point in exactly the same direction here, which is a rare and convenient thing in sales.