Chapter 5
Finding His Voice
The Post He Dashed Off
He posted it the next morning.
Not the version he wrote standing up. He spent another forty minutes cutting that one, because at seven in the morning the line he had typed under it, I'm afraid three of my five customers can read this and will know it's about them, looked like a lawsuit waiting to happen. He took out the part about the lender that sends the same note to everyone on the fifteenth of the month. He took out the word worse. What went up was true and careful and much weaker.
It got thirty-one likes and one comment worth having, from a servicing manager in Boise who said this matches what we see and nobody will say it out loud. Marcus kept rereading the comment.
It did well enough that he took the wrong lesson from it: be more careful.
Ten days of daily posting later, the numbers say it's working. He has a little over five hundred followers, up from two hundred and sixty. An average post gets twenty-five or thirty likes. People he has never met comment, and two of them have become real conversations in his DMs.
Rachel, his investor, gave him ninety days to build a pipeline, and most of them are gone.
It is 11:40 that morning, and Marcus is two hours into three paragraphs about the hidden cost of client churn. They come to a hundred and fifty words. He writes a sentence, reads it as a stranger who has decided he's a vendor would, and deletes it. He opens two posts from people with bigger audiences to compare their shape with his. He wonders whether the point is original enough and whether he can defend it.
Yesterday's post took three hours and got twenty-eight likes. Last Thursday he typed up something a customer said on a support call, with his coat still on, in twenty minutes. It got forty-seven likes and six comments.
This has happened twice now: the post he works hardest on does worse than the one he dashes off. He hasn't acted on it.
David, his co-founder, stops in the doorway with a coffee he isn't drinking.
"You started that this morning."
"It needs to be right."
"Does it have to be perfect? Write it the way you'd say it to a customer."
Marcus doesn't answer.
David looks at the screen a second longer, then leaves.
Marcus rereads his three paragraphs and sees what David meant. They are polished and correct, and they read like a competent stranger's summary of something Marcus knows from the inside. Anyone could have written them.
He selects the whole draft and holds delete.
Then he opens a blank one and types what he would say out loud.
Had a call yesterday with a loan officer who's been doing this twenty-two years. He said something I can't stop thinking about.
"When I started I knew my clients' kids by name. Now I've got a lot more clients and half of them I couldn't pick out of a crowd. The software was supposed to help with that."
I don't have a solution in this post. I just think he described the whole problem better than any deck I've seen this year.
He doesn't sand it. He posts it and goes to get coffee.
Three hours later it has sixty-seven likes and twelve comments. A woman at a lender he's been trying to reach since January has sent him a message.
Cutting It Four Ways
That evening he asks the AI model for help with a post, for the second time this month.
He gives it the loan officer's quote and asks for four versions: one that names the problem and stops, one that turns it into a framework, one that tells it as a story, and one that picks a fight. The model returns four drafts in forty seconds, and all four are competent.
Three of them are worse than the original, and he can see it right away. The framework version turns a man's regret into a maturity curve. The contrarian version picks a fight with a lender who doesn't exist. The story version is fine, but it explains the quote, so it's weaker than the quote.
The problem-naming one is close, except for a last sentence that reassures the reader. He keeps what he posted and deletes all four. He can tell which are worse because he was on the call and knows what the man meant.
Real Beats Right
He starts using his analytics to find out what works, not to keep score.
Ten days of posts, and there is a clean split he's been refusing to see, because the category that loses is the one he works hardest on.
The posts that do well have four things in common. They are specific. Instead of client retention matters, they give a number and the part of it nobody says out loud. They are unfinished: they name a problem and leave it open, which invites people to respond. They draw on something only he has, such as a customer call, his five beta lenders' usage data, or the question a woman from an Ohio credit union asked a consultant who couldn't answer it. And they read as if he is writing to one person.
So he stops rewriting.
Posting daily without rewriting costs him something else. Around eight every evening the day's post is live, he thinks it's mediocre, and he won't let himself edit it. He used to spend hours on a post.
Some land flat, single digits and no comments. One of those he still thinks was right. He posts the next one at eight anyway.
Content That Converts
Writing the posts isn't the slow part anymore, for Marcus or for you.
Marcus could publish forty posts a week, and so could anyone. Feeds are full of grammatically correct posts that say nothing. What limits your content now is whether you have something you would say out loud to a customer. The model can turn what you know into more posts. So start with what you know, and use the model second.
The mix. A reasonable mix is about 60 percent educational posts, 25 percent insight posts, and 15 percent personal posts. Educational posts teach the reader something about a problem they already have. Insight posts make them see it differently, and some of those will cost you a follower or a customer. Personal posts show there is a person behind the account.
This is a rule of thumb, not a finding. I have never seen it measured, and it won't fit every market. It is useful because it guards against two common mistakes: posting mostly personal material, which turns your page into a diary, and posting mostly pitches, which turns it into an ad.
All three kinds of post come from what you have seen yourself in your own work. The model can help you keep the mix across a month. It can't supply the content of any of them.
The four shapes. Most posts worth writing take one of four forms, and each one has a structure you can copy.
Problem analysis. Open with something specific, like a number, a quote, or a moment. Say what it means for the reader, in their own words. Then stop. Don't offer the fix. If you name the problem, the reader trusts you. If you solve it in the post, they don't need to call you. A reader who finishes thinking that is exactly my situation will contact you for the rest.
Framework. Give the idea a name and three or four parts. Each part should be a decision the reader can make, not a topic. "Pricing" is a topic. "Which customers get a discount" is a decision. The framework has to make sense without you explaining it. If it needs you to explain it, rewrite it.
Story. Tell one moment with one person in it. Give the concrete details first and the point after. End the story a sentence earlier than feels safe and let the reader reach the conclusion. Explaining the point ruins a story post.
Contrarian. State what everyone says. State what you have seen instead. Then say what it cost you to find out. That is what makes it more than an opinion.
A model writes frameworks easily, and it writes a plausible problem analysis. It can write something shaped like a story, but not a true one, because it wasn't there, and an untrue story falls apart when a stranger retells it. Ask it for a hot take and you get the average opinion.
Why less pitching works. Once someone has read you for a while, you don't have to persuade them. They describe their situation and you already understand it.
Why posts keep working. A cold email reaches one person once. A post reaches some people now and a few more later: someone pulls it up in a comment thread, someone forwards it into a channel you will never see, someone finds it after being handed the problem. Outreach stops when you stop. A post can keep getting read for months.
Use the model to repurpose. One real observation from your work can become four posts, a talk, or the outline of a guide, and cutting one idea several ways is mechanical work. Do the thinking yourself and let the model do the cutting. This week, take one thing you saw in your own work and ask for the four versions.
The Call She Started
Marcus writes down where the month stands.
He has 612 followers, up from 260 when he started. A good post gets forty-five to sixty likes and six to eight comments, more when it came from a real conversation and fewer when he sanded it. His email list has 52 people on it, all from a plain PDF Tom had him write, Five Questions to Assess Your Client Retention Strategy. It has no design and no funnel behind it.
Besides the one from the loan-officer post, five conversations opened in his DMs. Three went nowhere: wrong size, wrong quarter, wrong problem. Two are alive, and neither is close.
Rachel met him the Friday after he wrote to her, read the spark file for twenty minutes, and asked when it would turn into pipeline. By her count the pipeline is still thin. He writes that down too, because she will ask.
Karen Walsh is the woman who wrote after the loan-officer post. Her message came in three hours after the post went up. She is VP of Sales at Regional Home Lending, which has two hundred loan officers. She has never answered one of his cold emails.
"I've been reading your posts for a week or so. The one about the LO who loses track of his clients. That's my whole problem. Would like to hear what else you're seeing."
They set a call for tomorrow. Today he screenshots her message and sends it to Tom.
"First real inbound from content. VP, right company, right title. What do I do."
"Nothing you'd call a plan. Don't pitch, don't demo, don't bring slides. Ask her what her people actually do three months after a closing. Then be quiet for longer than is comfortable."
Then, a moment later:
"The post that got her wasn't the one you sweated over."
Marcus scrolls back. Tom is right. It annoys him a little that Tom could tell without seeing any numbers.
Since that first post, he has written about thirty posts. Three of them are really his. One brought Karen Walsh and the other two brought the conversations that are still alive. Three of the rest brought conversations that went nowhere in the polite way.
Tomorrow at two he has thirty minutes with somebody who came looking for him.
He opens a blank document and writes down every question he wants answered, including the ones with no obvious link to a sale. He stops at nine. He doesn't rank them.
Endnotes
This chapter has no external citations.
The engagement figures throughout are illustrative.