I run email strategy for a B2B software company, and over the last year I've reviewed cold sequences from our own SDRs, two agencies, and a steady stream of founder friends who slide into my DMs with 'can you just look at this real quick.' Sixty-something sequences later, the same five mistakes appear in almost every one. They aren't secret. Every guide names them. They survive anyway, because each one feels correct while you're the person writing it — and feeling correct is a very different thing from being correct, especially in cold email, where your gut is calibrated to impress and the reader is calibrated to delete.
I want to be clear that I make these mistakes too. I'm not standing on a hill. I caught myself committing three of the five in a sequence I wrote last month for a webinar, which is roughly like a dentist with a cavity. The mistakes are gravitational. You have to build something that resists them, because willpower at 4:45pm on a Friday is not a building material.
The five that show up almost every time
- Email 1 is too long and talks about the sender — anything over 100 words in a first cold email quietly dies in the preview pane
- 'Quick question' or 'I hope this finds you well' openers — experienced buyers now read these as noise and skim straight past the body
- No angle progression — all five emails make the same argument in five slightly different costumes, so opening one means you've read them all
- No breakup email — the sequence just stops, leaving the 15–25% of replies that the honest goodbye reliably pulls from people who ignored the first four
- Subject lines that summarise instead of intrigue — a subject's job is one click, not a TL;DR of the email underneath it
How one prompt kills four of them on the first pass
The Cold Outreach Sequence — 5-Email B2B Campaign That Earns the Reply prompt fixes these structurally, which is the only kind of fix that survives a tired human. It literally can't produce the same pitch five times, because each email has a declared job — first contact, new angle, proof, pain question, breakup. The hard word counts strangle the 'too long' problem before it starts. The banned-phrases block in the rules nukes the opener clichés. And the breakup is built in as Email 5, not an afterthought you forget at 5pm on a Friday when you're already mentally at the pub.
I tested this the lazy way, because lazy is how the prompt will actually be used. I gave Claude Opus 4.8 a deliberately mushy brief — the kind of thing a rushed founder types with autocorrect fighting them — with a vague offer and no proof point. It still returned five distinct emails with cleanly progressing angles and left the [NEEDS PROOF] flag sitting in Email 3 where I couldn't miss it. The structure isn't a suggestion in this prompt; it's the scaffolding the whole thing is bolted to. The output read like a real sequence written by someone who'd had coffee, slept, and re-read their own work, which describes almost none of the cold email actually being sent on this planet.
That's the honest pitch for a structural prompt over a clever one. A clever prompt makes a good email when you give it a good brief and a good mood. A structural prompt makes a structurally sound email when you give it a mediocre brief and a bad mood, which is the realistic operating condition. I'll take the floor-raiser over the ceiling-raiser every time for a task I do under deadline.
The mistake the prompt can't fix
Mistake six is personalisation, and that's on you. The prompt writes a strong skeleton for a segment, but Email 3's proof point and the relevance line in Email 1 are where a human still has to do real work — find the real number, name the real trigger, know the real audience. The prompt even leaves a [NEEDS PROOF] flag so you can't fake it without consciously deciding to. I still occasionally ship a sequence with a generic proof point because I'm in a hurry, and those sequences always, predictably, embarrassingly underperform the ones where I bothered to dig up a real result. The tool removed four of my excuses. It cannot remove the fifth, which is laziness, because that one is structural in me and no prompt has access to my character.
Run the Cold Outreach Sequence — 5-Email B2B Campaign That Earns the Reply prompt on your worst-performing sequence on Prompt Dock, then read the five emails it returns and ask one question of your old version: does each email make a different argument? If two of them make the same one in different words, congratulations, you've found exactly why nobody is replying — and the fix is already sitting in your other tab.