Coldpit.ch

Handbook

Why Coldpitch only writes what it found

A cold message that states one wrong fact is worse than no message. So the writer is not allowed to know anything it was not given.

Written down 4 October 2026

Coldpitch's writer is handed a short list of facts about one company, each with the place it came from, and it may use nothing else. Not general knowledge about the industry, not what companies like this usually need, not a plausible guess. If the facts do not add up to something worth saying, the right output is no message, and that is what it produces.

The failure we built against

Every scaled personalisation system has the same weakness: when there is nothing interesting to say about a company, it produces something interesting anyway. A model asked to be specific about a stranger will be specific, and some of what it says will be untrue: a product that was never launched, a post that was never written, a number from a different company.

A generic message is ignored. A confidently wrong one is read, and it tells the recipient exactly how it was made. For an agency, whose whole pitch is that it understands the client's marketing, that is the worst possible first impression.

What the writer gets

That is the whole input. The writer cannot look anything up, and the rules it writes under forbid filling a gap with what is usually true. Personalising without inventing comes down to that separation: collect the facts first, then write from them and nothing else.

When there is nothing to say

Some companies arrive with healthy numbers and no event worth mentioning. A system that must produce a message for every row will invent one. Ours is allowed to stop: a company with nothing observable to open with does not become a lead.

That makes the list shorter, on purpose. A short list where every message is true is worth more to an agency than a long one where some are not, because the cost of a wrong message is not one lost reply. It is the name of the agency that sent it.

What it means for you

Every claim in a draft can be checked by the person who receives it, and by you before you send it, because the facts it came from are on the lead. A second step has already read the draft against those facts; why that step is separate is its own page.

Try Coldpitch and skip the manual part

It finds the accounts worth writing to, spots the weak point, digs out who to talk to on Telegram or X, and writes the first message from what it actually found. Less work than doing this by hand, and better leads at the end of it.

Get my first messages

More from the handbook

The whole handbook