Why your "personalised" cold email still gets deleted

Five years ago, mentioning a prospect's recent funding round in line one felt like a personal touch. It told them you'd done your homework. Today, every AI tool on the market does exactly that, automatically, for thousands of prospects at once. So the move that used to signal effort now signals the opposite. It reads as a machine.

That's the shift almost nobody has adjusted to, and it's why your open rates hold up while your replies quietly die.

Here's the uncomfortable part. Personalisation, as most sellers mean it, is dead. A first name, a job title, a nod to their last post: buyers read straight through all of it now, because they know a template with a name dropped in is still a template. The word to hold onto instead is relevance. Not "I know a fact about you," but "I understand something about your world that could only apply to you."

The test is brutally simple.

If an AI could find it in fifteen seconds, it no longer sets you apart.

Their funding, their headcount, their last post: all fifteen-second facts, all invisible now. The research that lands is the stuff a bot skimming their profile would never hand you. For a public company, that's their own words: the risk factors section of their 10-K is the business describing its problems in the language its executives actually use, and echoing that language back is close to the highest-leverage move in enterprise outbound. For everyone else, it's their open job ads, which quietly show where a team is stretched, something a leader said on a podcast, or a quirk in how they run a process that you only catch by actually looking. Read their careers page properly and it often tells you what they're building toward and where the strain already is.

Then comes the part no tool can do for you, and it's the part the whole email lives or dies on: the bridge.

AI will happily summarise a profile and list your features. What it consistently fails at is connecting this signal about this company to the specific problem you solve. That one sentence, the one that says "I noticed X, which usually means Y is getting harder for teams like yours," is the sentence a human has to write. Everything around it is scaffolding a machine can build. That line is not.

You can feel the difference in a single swap. Weak, and instantly binned:

"Congrats on the Series B! We help companies like yours scale outbound."

Relevant, and worth a reply:

"Saw you've opened three SDR roles in two weeks. When a team scales headcount that fast, ramp time is usually the thing that quietly breaks first."

Same product. Same prospect. One reads like a mail merge. The other reads like a person who thought for thirty seconds about their actual situation.

So use AI for what it's good at and stop trusting it with what it isn't. Feed it four things: the role stated specifically, the exact signal you found, the company's context, and the real problem you solve. Let it draft the structure. Then do the pass that matters: cut every sentence that could have gone to a thousand other people, and put the words back the way you'd actually say them out loud. The machine gives you a frame. You supply the specifics and the voice, because those are the only two things left a buyer can't get from a bot.

There's a full cold-email teardown in the library if you want to see this run end to end on a real prospect.

This week, take one email you'd normally send. Before you write a word, find one thing about the prospect an AI couldn't surface in fifteen seconds, and build the whole email around it. Then watch your reply rate, not your opens.