I get the request roughly every other week. A founder I'd otherwise want to work with sends me a sample of what she'd like me to write for her, and the sample is something I cannot do. Not because I don't have the skill. Because doing it would damage her in a way she hasn't yet considered, and damage me in a way I have.
I've started keeping a list. Not a list of refusals — that sounds more dramatic than it is. A list of the specific kinds of work I've turned down, and the specific reason each one fails. I want to walk through it because I think the refusals are the most honest map of what this practice actually is.
Every "yes" is a kind of marketing claim. The "no"s are where you find out what someone really believes.
The takedown post.
A founder comes to me wanting to publicly criticize a competitor, a category, a methodology, or a specific person's approach to outbound. The request is almost always framed as "I want to take a strong position." The position is almost always a takedown.
I won't write it. Not because takedowns don't perform — they perform extremely well, and I know exactly which mechanisms make them perform. I won't write it because the founder will pay for it for two years after it goes live. The takedown post creates an enemy. The enemy reads the post. The enemy's friends read the post. Six months later the founder is in a sales conversation with someone who knows the enemy, and the conversation is shaped, invisibly, by the post she wrote in 45 minutes a year ago.
The version of a "strong position" I will write is the one about the founder's own working rules. "I will not write for founders whose TAM is over 5,000 companies" is a strong position. It's about my behavior. It creates no enemy. It also makes a clearer claim about what I believe than any takedown of someone else's approach could make.
The distinction is: a strong position is what you do. A takedown is what they do wrong. The first compounds. The second corrodes.
The confession that isn't.
The "I was wrong about X" post where the founder wasn't actually wrong about X.
The Founder Confession is a powerful post type. The pattern works because vulnerability buys the right to authority later. The mechanism is real and I use it.
But I won't write the confession when there's no confession to make. A founder comes to me saying "I want to do one of those vulnerable posts, can you find something I was wrong about?" The request itself tells me the answer is no. If she can't immediately name the thing, the thing doesn't exist, and what she's asking me to do is manufacture vulnerability for performance purposes.
Manufactured vulnerability has a specific tell. It reads slightly off, the way a forced laugh reads slightly off. Most readers can't articulate what's wrong with it. They just feel a small distance from the writer, where they're meant to feel a small closeness. The post performs worse than a real confession would have, and it costs the founder a slice of her credibility that she won't get back, because manufactured vulnerability — once spotted — colors every future post she writes.
If a founder wants a confession post and can't tell me what she was wrong about, I tell her to wait until she finds out. There's usually one available within a quarter. It's worth waiting for the real one.
The stake-claim she'll regret.
The stake-claim on a topic the founder hasn't actually thought about for six months.
Stake-claim posts are the rarest and most powerful post type I write. They work because they make the founder uncopyable. A founder who's publicly committed to a specific way of working becomes someone with standards rather than someone with opinions. The mechanism only works if the commitment is durable.
I will not write a stake on a topic the founder is taking up this week. I'll write it on a topic she's been thinking about for at least a quarter, ideally longer. Because the stake-claim post will follow her for years. Prospects will quote it back to her. Peers will hold her to it. If the stake was casual — if it was the thing she felt strongly about for the eleven days between when she read an article and when she asked me to write the post — she'll regret it within a month. She'll either have to publicly walk it back, which damages her, or quietly contradict it in private conversations, which damages her differently and worse.
I ask one question before writing any stake-claim. "Will you still believe this in a year?" If the founder pauses, the answer is no. If the answer is no, I won't write it.
The sequence without customer language.
The sequence for a founder who can't tell me her last conversation with a current customer.
This one sounds operational and isn't. The reason matters more than the rule.
The thirty-minute research process per prospect is built on a specific kind of knowledge — the knowledge of what the founder's customers actually say, in their own words, about why they bought. That knowledge is what makes the layer-four observation hit a nerve. It is also the knowledge that is most often missing.
When a founder cannot tell me, with specificity, what her last customer conversation contained — what the customer was struggling with, what made the customer choose her, what the customer said in week three after the work started — I know the work I write will be downstream of a hole I cannot fill. The messages will be technically competent. They will not resonate, because the resonance lives in the customer language the founder hasn't internalized. The prospect will read the messages and they'll sound like a vendor's pitch, because that's what they'll be — a pitch built on assumed customer pain rather than observed customer pain.
I tell founders in this position to go have three customer conversations and come back. I've had founders take this badly. The ones who don't take it badly are the ones I end up doing the best work for, because the conversations they come back with change the entire shape of the campaign.
Content that names competitors.
Even when the competitor named is a category — "templated outbound agencies," "AI-personalization tools," "high-volume SDR shops" — I tighten the language away from anything that lands as a specific accusation. The brand position is clearer when stated as a constraint on my own behavior than as a critique of someone else's.
I'll say "I write fewer than ten messages a day." I won't say "agencies sending two thousand messages a day are scamming founders." Both are pointing at the same thing. The first one compounds for me. The second one creates a fight I haven't decided whether I want.
This isn't moral squeamishness. It's a practical observation about how brands get built and how they get burned. The founder who builds her brand on naming what others are doing wrong has to keep naming, or the brand collapses. The founder who builds her brand on what she does and doesn't do has somewhere stable to stand for a decade.
The post written to a buyer I haven't met.
This one's the most uncomfortable to admit, because it cuts against how this practice is supposed to scale.
The posts I write for the founders I work with are good in proportion to how well I understand the buyer the founder is writing to. If a founder is selling to VPs of Clinical Operations and I've never had a real conversation with a VP of Clinical Operations, the content I write for her will be one step removed from real. It'll be informed by research. It won't be informed by the specific cadence and texture of how that buyer actually talks, what she actually worries about at 7pm on a Sunday, what she'd never say to her CEO. The posts will be good. They won't be the posts the buyer would have written herself if she could.
When a founder approaches me with a buyer category I don't yet understand, I do one of two things. Either I take three conversations with that buyer before starting the work, or I tell the founder the work won't be as good as it will be for the buyer categories I already know, and let her decide. Most founders, when given the choice, ask me to do the three conversations. It's the right answer.
The pattern across all six.
I notice the refusals don't follow any single principle. They aren't all ethics. They aren't all craft. They aren't all economics. They're a mix.
What unifies them is this: every refusal is a thing I would write if I were optimizing for the next sale. Every one of them is the version of the work that a templated agency would happily produce. I don't refuse them because I'm precious about the craft. I refuse them because each one, if I said yes, would produce a result the founder pays for years after the invoice clears.
The honest version of what I do is this: I write the work that I would still want to have written a year after it goes live.
The work I would not want to have written, I don't write. The list above is the most reliable way I've found to keep that line drawn.
A founder who can't get what she wants from this list isn't a founder I can't help. She's a founder I can't help with what she's currently asking for. There's almost always a different version of the work I can do — the constraint becomes the case, the stake becomes a slow-built belief, the takedown becomes a public commitment. The conversation about what I won't write is almost always the conversation that surfaces the work I'd be best at doing instead.
Most of the engagements I'm proudest of started with a "no" on the first request.