start here: what's your problem?
Stuck? Find the line that sounds like you. Each one points to the principles, the framework and the book chapter that answer it.
1/ "I post, but nobody engages."
Principles: 5 (content is for the audience, not for you), 6 (relevance before relatability), 7 (posts travel when they're new or relatable)
Framework: 9, The reach map
Read: Chapter 5, Why posts travel
The fix: your posts are probably about your world but not about what your readers care about. Move from the narrow thing you sell to the bigger thing many people care about (not your niche service, but getting leads). Then borrow a format people already enjoy and use it on that topic.
2/ "I don't know what to post."
Principles: 3 (know what you want from LinkedIn), 1 (every founder is selling), 2 (your goal, filtered by your comfort)
Framework: 1, The founder strategy stack, then 2, The content funnel
Read: Chapter 1, Every founder is selling
The fix: decide what you want from LinkedIn and who you need to convince to get it: customers, investors, hires or partners. Then post stories that make them care, how-tos that make them trust you, and plain asks like "we're hiring". Keep only the kinds of post you're comfortable writing.
3/ "My posts sound like everyone else's."
Principles: 8 (people are tired of advice, they want people), 17 (momentum compounds)
Framework: 15, Writing as yourself
Read: Chapter 8, Writing as yourself
The fix: stop writing advice. Write the way you'd text a friend: what you're doing, what you're thinking, what just happened. Write by hand through the day, not 1 batch in an hour, and post often.
4/ "I've run out of stories."
Principles: 10 (every event has more in it than anyone talks about), 9 (audiences remember less than you think)
Framework: 3, The story timeline, and 7, Shaping an idea
Read: Chapter 2, Getting the stories out, and Chapter 4, The shape of an idea
The fix: list stories from the last 10 years, 3 years, year, quarter, month and week. Then take 1 thing that happened and look at it from 6 sides: the thing, how it happened, the number, what was easy or hard, the speed, and how someone else felt. It's fine to retell old stories, just not the same lines.
5/ "My hooks are weak."
Principles: 11 (the shape of the idea decides the post), 12 (the hook decides whether anything else gets read)
Framework: 8, Idea, then hook
Read: Chapter 4, The shape of an idea
The fix: a weak hook is usually a vague idea. Write the idea in 1 line, pick 1 angle, and only then write a hook that says clearly what the post is about. Check the first 2 lines as a stranger would before you publish (framework 10).
6/ "I don't sell anything on LinkedIn, so why post?"
Principles: 1 (every founder is selling)
Framework: 1, The founder strategy stack
Read: Chapter 1, Every founder is selling
The fix: you're still persuading someone: investors, senior hires, partners, a board, your next role. Name that person. Post the stories and opinions that would change their mind about you.
7/ "AI writes my posts and they all come out bland."
Principles: 19 (understand why it went wrong before you correct it), 20 (when all your drafts fail, the problem is upstream)
Framework: 11, The AI correction loop
Read: Chapter 7, Working with AI without losing your mind
The fix: "these are bad" isn't feedback. Say exactly what was wrong and why the AI went there, and put that in the prompt. If every draft fails, stop rewriting and fix the ideas first. If it's just dull, tell it "this is too boring".
8/ "I don't know why some posts flop."
Principles: 15 (skill is built in the edit), 14 (judgment only comes from reps), 16 (depth beats collection)
Framework: 12, The post-mortem edit
Read: Chapter 6, How you actually get better
The fix: reread the post slowly, write down your best guess at why it flopped, and edit it hard. Do this on your own posts again and again. There's no formula, only reps, so get good at 1 or 2 formats before chasing more.
principles
These are the beliefs underneath everything Niraj teaches about LinkedIn. Almost every piece of his advice is one of these principles applied to a specific situation. Understand these, and you can work out the right move in situations nobody has written about yet.
About strategy
1/ Every founder is selling something.
Even founders who say they only want thought leadership are trying to convince someone: investors, senior talent, partners, a board, a future employer. Say you run a product business. You don't sell your product on LinkedIn, but you are selling your company to investors and to great people you want to hire. So "no sales content" never means there's no one to persuade. It means the persuasion is aimed somewhere other than customers.
2/ Strategy is your goal, filtered by what you're comfortable with.
The goal decides who the content is for. The funnel decides what kinds of content get made. Your comfort decides how. Funny content only works if you're comfortable being funny. A plan that ignores your comfort won't survive you.
3/ Everything begins with knowing what you want from LinkedIn.
When you're stuck on what to post, the cause is usually that you don't clearly know what you want from LinkedIn right now. The fix is to decide. You should be able to say it in 1 line without thinking about it.
4/ A clearly defined space makes everything downstream easier.
When your company's industry is clear, research, ideas and relevance come easily. When it's fuzzy (marketing? media? tech?), you keep swinging between too broad and too narrow. You can talk about many topics, but they should all trace back to 1 clear space.
About the audience
5/ Content is for the audience, not for you.
An idea can sit perfectly inside your niche and still fail because your audience doesn't care. Every idea has to pass 2 tests: is it about your world, and would your audience actually enjoy it?
6/ Relevance comes before relatability.
If something isn't relevant to you, you will never relate to it. But relevance doesn't guarantee relatability: health is relevant to Niraj, yet he relates to only some health content. First make sure the topic matters to a lot of people. Then do the harder work of making them see themselves in it.
7/ Posts travel for 2 reasons: they're truly new, or a lot of people relate to them.
Novelty is something nobody has seen, like the Google CEO doing a podcast. Relatability is people seeing themselves in it. "Make your audience feel something" is just relatability described from the outside. Most good LinkedIn content lives in relatability.
8/ People are tired of advice. They want people.
The internet is full of slop and advice. What's scarce is a real person talking the way they'd talk to a friend. When readers feel like they know you, they get involved in your life.
9/ Audiences remember much less than you think.
If you can't recall the stories you've already posted, your audience can't either. Old stories can be told again. What doesn't work is dropping the same story or line over and over in an obvious way.
About ideas and posts
10/ Every event has more in it than anyone talks about.
One thing that happened can be seen from many sides: the people, the method, the numbers, the speed, how someone else felt about it. Most people only ever talk about the obvious side. That's why there are always more stories, and why material doesn't really run out.
11/ The shape of the idea decides the post.
A raw idea floats in midair. Shaping it means choosing which part of it you're showing. Most bad posts and bad hooks are really unshaped ideas. Good input doesn't automatically make good posts: good input has to become well-shaped ideas first. (Niraj used to say good input makes good posts. This is his correction.)
12/ The hook decides whether anything else gets read.
No post works without a really good hook. A good hook tells the reader clearly what the post is about, and that clarity comes from the shape of the idea. So the idea gets shaped first, and the hook still matters most.
13/ You can't be specific before you have context.
You can't ask specific questions about something you know nothing about. Broad comes first, specific comes once you know what exists. The same is true for research: specific ideas need a defined niche first.
About getting better
14/ Judgment only comes from reps.
There's no framework that tells you whether the idea, the hook or the structure is broken, and no formula for relatability. There are fundamentals, and past them only practice. Frameworks help you practise. They don't replace it.
15/ Skill is built in the edit.
Niraj can pinpoint what's wrong with a post and word the fix quickly because he edited his own work again and again. Most people who post don't edit enough. It's slow and painful because you have to admit mistakes in work you wrote, and that's exactly why it works. There's always room to improve, especially when you're posting every week.
16/ Depth beats collection.
Knowing many ways to go viral doesn't make you go viral. Master 1 or 2 until you can execute them and have seen results, then add more.
17/ Momentum compounds.
The more you write, the better you write, because the oven is already hot. Volume also creates outliers, the few posts that do more than everything else combined, and the only way to get them is to post more.
18/ Learn from what's already working.
Patterns show up when you watch creators who post consistently. A format that works in one niche can be moved into another. You rarely have to invent from nothing.
About writing with AI
19/ Understand why something went wrong before you correct it.
AI was trying to do a good job and misread something. Look at it from its side, find the core cause, and the right feedback becomes easy. "These are bad" isn't feedback. Saying exactly what was bad, and why, is.
20/ When all your drafts fail, the problem is upstream.
One weak post usually means one weak idea. When every draft is weak, it's almost never a writing problem. It's an input or ideas problem. Fix the source, not the drafts.
21/ Do things properly, in their own slot.
Don't squeeze a review into a call that's meant for something else. Every conversation has a purpose. Protect it, and book the other thing separately. If you're working with a writer, an editor or a coach, an input call is for input, and reviewing posts gets its own time.
If you write for someone else
If you're a ghostwriter or you run content for a founder, 2 principles change shape:
- Principle 3 becomes: know what your client wants. When you're stuck on a client, you usually don't clearly know what they want right now. Ask them. You should be able to say what each client wants without looking it up.
- Principle 9 applies to you too. If you, the person who works on the account, can't recall the stories already posted, the audience can't either.
- Principle 15 at an agency. There's always room to improve, especially at an agency that optimises for speed.
- Principle 21 on a team. If you're on an input call with your lead and want them to look at a client's posts, ask for a separate review call and bring everything else that needs checking.
The writer-specific frameworks are listed at the end of Frameworks.
frameworks
These are the repeatable structures Niraj uses. Each one says what it's for, how it works and why it works, because a framework used without its reasoning goes wrong. They're tools for practice, not a replacement for judgment (principle 14).
1/ The founder strategy stack
For: deciding what you should post.
How it works:
1/ Goal: what do you want from LinkedIn?
2/ Audience: who do you need to convince to get it? Customers, investors, talent, partners, a board.
3/ Funnel: what content moves that audience? (Framework 2.)
4/ Comfort: of that content, what are you comfortable with?
Why: every founder is selling something (principle 1), so your goal tells you who the content is really for. Comfort comes last because it's a filter, not a starting point.
2/ The content funnel
For: choosing the types of posts, once you know who you're persuading.
How it works:
- Top: stories and opinions that push your narrative. If you're hiring and raising, stories about the team speak to talent, and stories about where your market is heading speak to investors.
- Middle: educational posts ("how I do X", for the goal you have) and case studies ("how we went from A to B": how we onboarded 10 people, how I run investor meetings).
- Bottom: the fears, pain points and frustrations of the people you're selling to, wins that matter to that audience, and plain direct offers ("we're hiring", "we're raising", "we're taking on clients").
Why: the same audience is at different stages of readiness. Stories make them care, education and case studies make them trust, and the bottom of the funnel gives them a reason and a way to act. A lot of people following you don't know you're hiring or raising, so just tell them.
3/ The story timeline
For: getting your stories out when you don't know where to start.
How it works:
1/ List stories across time: the last 10 years, the last 3 years, the last year, the last quarter, this month, this week.
2/ Aim for a few from each: say 1 to 3 from your wider life and 1 to 2 from each recent period.
3/ If you struggle to remember, think of wins and losses. It's only a way to remember, not boxes to sort your stories into.
4/ Optionally, nudge yourself towards stories around the topics you want to be known for (your content pillars).
Why: you can't ask yourself specific questions before you've laid out what's there (principle 13). Moving through time gives a balance of old and new stories, and the wins and losses prompt helps memory.
4/ When the obvious stories are out
For: the second round, when you've already told the stories that came to mind first.
How it works:
1/ Look at what worked in your first month of posting.
2/ Go in knowing there are always more stories (principle 10).
3/ If you use AI to suggest questions from the stories you've already told, tell it not to ask obvious follow-ups. Ask for questions that serve your content strategy.
Why: given past stories, AI tends to ask questions that feel like the next line of a conversation, not part of a content strategy.
5/ The niche layer map
For: research and relevance, especially if you talk about many things.
How it works: write your company's space from narrow to broad. For LAAS: LinkedIn ghostwriting, then content agency, then marketing agency, then running a business. Content can come from any layer. Give this map to the AI before you ask it for news or research ideas.
Why: a clear space makes research easy (principle 4). The broader the layer, the harder it is to stay relevant: broad industry news on a founder with a narrow, specialised business would be technically relevant but feel odd.
6/ The research post recipe
For: writing from news without sounding like a news update.
How it works:
1/ Your goal (who this post is for) gives you the angle. Only pick news that matches it.
2/ The funnel stage gives you the format: story, opinion, educational, or fears, pain points and frustrations.
3/ Build it on what you already have: your goals, core opinions and stories.
Why: news on its own is generic. Your angle and beliefs are what make it yours. "Without fresh input" never means zero input.
7/ Shaping an idea
For: turning a raw idea into something worth writing.
How it works:
1/ Write the core idea in 1 line.
2/ List the angles: the thing itself, how it happened, the number or milestone, what was surprisingly easy or hard, the speed, and someone else's point of view and how they felt.
3/ Pick 1 angle and write the post around it.
Example: "We closed a new client yesterday" can become: the client is doing cool things; an outreach text won them; it's our 12th client this year; they had zero objections on the call; they paid within 12 hours; my accounts lead was thrilled, or stressed because we're already stretched.
Why: the shape decides the post (principle 11), and every event has more sides than people talk about (principle 10).
8/ Idea, then hook
For: hooks that work.
How it works:
1/ Shape the idea first (framework 7).
2/ Check it. When a draft is bad, ask the AI: is the idea strong, weak, well formed? Refine the angle if needed.
3/ Write a hook that says clearly what the post is about.
4/ Use a swipe file of hooks that work for others: match them by instinct, or give the list to AI and ask which pattern fits this post.
Why: the hook is about the idea, so a vague idea can't get a clear hook (principle 12). Not every hook has to be broad.
9/ The reach map
For: posts meant to reach a large audience.
How it works:
1/ Novelty or relatability? If it isn't genuinely new in a way that matters, go for relatability.
2/ Relevance gate: move from the narrow thing you sell to the broader thing many people care about. A niche B2B service isn't relevant to most people. Generating leads is.
3/ Format transplant: find a format people already relate to in another niche and adapt it. Say you sell a niche B2B service: take the A tier, B tier, C tier format and apply it to generating leads.
Why: relevance comes before relatability (principle 6), and borrowing what already works is the fastest route to relatability (principle 18).
10/ How Niraj reads a batch
For: knowing how a batch of posts gets judged. This is how Niraj approves posts from writers on his team. Use the same steps to check your own posts before you publish.
How it works:
1/ He reads only the hook of the first post, 2 lines max.
2/ Really bad: the whole batch is rejected.
3/ Kind of bad: is it a mistake that will repeat across the batch? Then the whole batch is rejected. Is it specific to this post, because of a weak idea or topic? Then he checks the next posts.
4/ If every hook works, he reads 3 to 5 lines of the body on most posts to check the quality holds.
5/ A good body never saves a bad hook.
On your own posts: read only the hook, as a stranger scrolling past would. Really bad, don't post it. A mistake you keep making, fix the habit across all your drafts. If the hook works, read 3 to 5 lines of the body.
Why: the hook decides whether anything else gets read (principle 12), for him and for readers. So every single hook has to hold, and reading the full posts is your job before they go out.
11/ The AI correction loop
For: when drafts keep missing and the back and forth gets exhausting.
How it works:
1/ Define exactly what was bad.
2/ Work out why the AI went there.
3/ Put that into your prompt (for example: relevant to me, but too niche for my audience).
4/ If the ideas are weak, ask the AI to refine them through research on what you already have, private notes and public material.
5/ Retry. If a few posts are still stuck, drop them.
6/ If the whole batch is stuck, drop it and sit with someone who can help you find the real problem (on Niraj's team, that's Niraj). It's usually your goal, your input or your ideas.
7/ If it's just bland, say "this is too boring".
Why: understand before you correct (principle 19), and when everything fails, the problem is upstream (principle 20).
12/ The post-mortem edit
For: a post that didn't do well.
How it works: reread it slowly, write down guesses ("this was probably the reason"), make lots of edits, and repeat this on your posts regularly.
Why: skill is built in the edit (principle 15). Your guesses are guesses: an edit can make a post better or worse.
13/ Tracking what works
For: knowing which formats are working right now.
How it works: keep a long list of creators who post consistently. Scroll their feeds. Ask AI for a daily summary of what each posted (for example, what Unnati Bagga posted today) and a weekly review of the types of posts.
Why: patterns show up when you watch consistent creators (principle 18).
14/ The hook swipe file
For: learning hooks you admire.
How it works:
1/ Copy hooks you like by hand into 1 place, like a doc or a Notion database.
2/ The ideal step: write real posts using those hooks or patterns, for your own account or a client's.
3/ Add them to your swipe file.
4/ Let AI use the collection when it writes hooks.
Why: copying and writing by hand puts the hook in your brain. Letting AI copy it doesn't teach you anything (principle 14).
15/ Writing as yourself
For: growing your own LinkedIn.
How it works: write by hand, all through the day, like you're texting friends in public. Not 1 batch in an hour. Post a lot.
Why: people want people, not advice (principle 8). Volume creates outliers and keeps the oven hot (principle 17).
If you write for someone else
If you're a ghostwriter or you run content for a founder, these are the pieces written for you:
- Know what your client wants (principle 3, writer version). If you're stuck on a client, ask them what they want to focus on right now and build your questions from that.
- Framework 3, run as a call. Ask the founder for stories across the timeline. Tell them the wins and losses prompt is only a memory aid, so they don't start sorting their life into boxes.
- Framework 4, the next ideation call. For the 2nd call with a founder: check what worked in month 1, and prompt AI for strategic questions, not obvious follow-ups.
- Framework 11, with no founder access. You usually can't go back to the founder for more input. Tell the AI that, and ask it to refine ideas from the private and public material you already have.
- Framework 13, watch other ghostwriters. Add about 5 LinkedIn ghostwriters who post consistently to your tracking list.
- Framework 14, practise on real accounts. Write real posts with your swipe-file patterns, for your own account or a client's.
- Framework 10, before you send for approval. Expect the person approving to read like Niraj does: first hook only, whole batch rejected if it fails. Read every full post yourself before you send.
- Framework 11, when the whole batch is stuck. Don't keep retrying. Drop it and sit with your lead to find the real problem.
the book
This is everything Niraj has taught about LinkedIn content so far, written as 1 long read in his own voice. It's built from his answers on calls and voice notes, kept true to how he explains things: the setup first, then the point. Read it like a book. Nothing here needs to be applied today. For things you can apply, see Principles and Frameworks.
Chapter 1: Every founder is selling
People ask me how strategy differs from founder to founder. Honestly, I wouldn't even call it a strategy. There are a few aspects to it, but the 1 thing underneath that never changes is this: whatever you post has to be built around your goal as a founder.
Now, you might tell me your goal is just thought leadership and reach. You don't want to sell through LinkedIn. Fine, then you don't do a lot of sales posts. But here's what's also true at the same time: every founder is selling. Say you run a product business. You're not selling your product on your LinkedIn. But you're at a place where you're selling to investors. You're selling to really good talent, because you want great people at your company. You want great investors reaching out to you. You might want suppliers, delivery partners, service partners. You are selling, just not your product.
It's the same even if you're a CEO who doesn't own the company. The CEO is selling to investors, to the board, to other companies. He's thinking about his own career too. Maybe he eventually wants a board seat somewhere. I remember someone who was high up at Tesla and ended up on the board of a competitor after leaving. That's a real possibility, and that's something people at that level are thinking about. So no matter who you are, you're selling something to someone.
So the first thing is: work out your goal. The second thing is: how do you get to that goal? And that depends on what kind of content you're comfortable with. If you're comfortable with funny content and it's working, great, do that. If you're not, what are you comfortable with? Figure that out, and then figure out which types of content get you to your goal.
This is where the whole funnel becomes really useful. We already know stories do really well. But if you're selling to investors, you tell the stories investors care about. If you're selling to talent, you tell stories about the team and what you do for your team, because that sells the narrative directly to the people you want. Say that founder wants to build a narrative around being the most trusted name in their space. Then you tell stories that push that narrative. Then there are opinions, again the ones that help you get where you want to go.
The middle of the funnel is educational content. How do I do X, but for the goals I have. How do I manage my team? How have I given my investors a good return? How do I make sure my finances are clear? And case studies: here's how we onboarded 10 people, here's how we keep our culture, here's how I take investor meetings. Going from A to B is basically a case study.
Then comes the actual sales content. First, the fears, pain points and frustrations of whoever you're selling to. If it's investors, what scares an investor about a company like mine? If it's senior talent, what frustrates them? Second, wins: the wins with my team, the wins that matter to investors. Third, plain direct offers. "We're actively looking for investors." "We're actively hiring." A lot of people follow a founder without knowing they're raising or hiring. So you just straight up tell them.
So you already know the fundamentals of the funnel. All you're doing is matching them to your goal, plus the kind of content you're okay with.
And underneath all of that is something simpler that people forget. You should always know what you want from LinkedIn. It should be crystal clear in your head. If you're ever stuck, not knowing where to start, it's usually because you haven't decided that. So decide it. When I work with founders, this is the conversation I have first. When someone on my team is stuck on an account, I tell them to ask the person they write for what they want to focus on on LinkedIn right now, and build the questions from that. You can ask yourself the same thing: what do I actually want from this right now? Unless you understand that deeply and keep remembering it, nothing you post will add up. That is where everything begins.
Chapter 2: Getting the stories out
When you sit down to find things to post with nothing planned, go back to the fundamentals of what you want from LinkedIn. You need clarity on the strategy first. Then, when it comes to stories, the broad approach is time.
Take a few stories from your broad life, the last 10 years or so. Then a few from the last 3 years. Then from the last year. Then the last quarter, this month, this week. When you do that, you get a really good balance: maybe 1 to 3 stories from your life that you'd like to talk about, 1 or 2 from the past couple of years, 1 or 2 from the past year, and so on.
You'll find it hard to recall things. That's normal. Every founder I sit with does. What helps is thinking of wins and losses, or wins and failures. Not every story is a win or a failure, and every failure is technically a lesson that adds up to something good. That's not the point. It just helps you recall. So don't get confused and start trying to sort your life into wins and losses. It's only a tool to remember.
You can also try to recall stories around the things you want to be known for, your content pillars. For a founder, that might be the core of what the company does, how the business started, where they are in their career, or a milestone the business has hit. But don't overthink this. Pillars are optional. The necessity is going through time in this order. Pillars are something you can add while you're thinking: "Let me also think of stories around this."
People sometimes ask me how to get more specific so their posts don't sound vague. But how will you ask specific questions if you haven't laid out what's there? You can't. The first time, your job is to broadly think of these types of stories, in this order. Specific questions come later, once you know what stories exist.
Another worry I hear: I've already told these stories on LinkedIn, won't it be repetition? So I ask, what do you remember about those stories? How many of them do you actually remember? If you, the person who posted them, don't remember, how can you expect anyone else to? If they don't remember, you can always talk about them again. What you shouldn't do is keep dropping the same story or the same line in an in-your-face way. Use your core ideas and personality again and again. Don't repeat the same sentence.
And for the second round, when you've already covered the last few years and the last few months: first, look at what worked in the first month. Second, there are always more stories. Always. One thing AI does badly here: if you give it the stories you've already told and ask for new questions, it thinks too closely around what's already there. It asks very obvious follow-up questions, the kind that feel like the next line of a conversation, not part of a broad content strategy. So be careful when you prompt it. If you just tell it not to make that mistake, it'll give you a decent set of questions.
Chapter 3: Knowing your space
Research-based ideas work really well for founders whose niche is clearly defined. Imagine a founder whose company does 1 specific thing. Their niche is extremely clear. What do they do? That 1 thing. That's 1 clear layer. Now imagine a founder whose business sits across a few fields, and whose niche isn't clear. What are they in? Marketing? Media? Tech? Where are they? When the space isn't defined, you struggle to find ideas, because you're either going too broad or too narrow.
Even with the founder whose niche is clear, the layer above it is usually far too broad. Say their whole industry is the layer above. There's so much under it. You can't pick up any industry news and post it on their account. It's technically relevant, but it'll feel odd, and it'll be hard to find ideas there.
So the fix is to get clear on your niche. Not "marketing", that's too broad. You can want to talk about A, B, C and D. That's fine. But what's the industry? What's your company's niche? What does it all go back to? For the founder with the clear niche, it all goes back to their company. When you're selling, it always goes back to the company.
Once the space is defined, everything else gets easier. And there are layers. For us, the narrowest layer is LinkedIn ghostwriting. A little broader, we're a content agency. Broader, a marketing agency. Broader still, we're a business, so we're in the niche of running a business. I can make content at every 1 of those layers. I can talk about business because the agency is a business. I can talk about marketing because what we do is marketing. I can talk about content because that's what we make. And what we really do is LinkedIn. So you need a rough map of your layers, and once you have it, give it to the AI. Then it understands who you are and how to find news that's actually relevant to you.
When you're writing posts from research, the way to not sound like a generic news update is the angle. Most research posts are educational or opinions, and some can address fears, pain points and frustrations. So you pick a part of the funnel you're matching it to, and you focus on the goal: who is this particular post for? Only pick stories that match that goal, and give them an angle. The goal is the angle. The part of the funnel is the format.
And obviously, don't do this with zero input from yourself. You already have your goals, your core opinions, your core stories and past experiences. Write on top of that. The post should be built around your core ideas and personality, not around the headline.
Chapter 4: The shape of an idea
Let's say there's 1 core idea: we closed a new client yesterday. That's the idea. Now look at how many ways you can approach it.
We closed a really cool client who's doing really cool things. We closed this client through outreach, a really good outreach text that helped us win them. We closed another client, it's our 12th this year. We closed this client and they had zero objections on the sales call. I closed this client in less than 12 hours: we had the conversation, and within hours he paid and we got started.
Do you see what's happening? It's 1 core idea, seen from lots of different perspectives. With anything that happens, there's more to it. There are parts of it most people never talk about, and the combinations get to infinity very quickly. I could talk about it from my accounts lead's point of view. Were they happy? Were they sad? Maybe they were extremely happy because it's a client they've wanted to work with for a long time. Maybe they were really sad because we're already barely keeping up with our existing clients.
Same idea, lots of different angles. That is what it means to give an idea a shape. The core idea can float in midair. How you choose to look at it, which part of it you choose to show, is what matters.
This is also the real answer to hooks. People ask me how to write hooks that are broad and relatable, to get more impressions. It is important to write hooks that are broad and relatable. But the hook is about the idea. If the idea itself isn't broad and relatable, the hook will fail. Take "I want to talk about black jeans". That's an idea. You can shape it into whatever you want, into something relevant and relatable to a lot of people. Once you have that refined idea with a specific shape, then you write the post and the hook. First comes a refined idea, then comes the hook.
That said, not every hook needs to be broad. The hook that goes viral is the hook that clearly communicates what the post is about. If the shape of the idea is bad, you won't get a good hook, so the post won't work. If the shape is well defined, the right hook gets much easier. I've seen this again and again writing with AI. When I get a bad post, I ask: is the idea strong? Is it weak? Is it well formed? Almost every time, the idea is the problem. So I refine it, find a good angle, and write again, and the hook comes out well.
So does the hook matter the most? Yes, 100%. No matter what the post is, unless the hook is really good, it won't work. Both things are true together: the hook matters most, and the hook comes from the idea.
People also want to know how to tell whether the problem is the idea, the hook or the structure. They're hoping there's a framework. There isn't. There are fundamentals of writing a hook and thinking of an idea, and you already know them. Past that, it's practice. You keep doing more and more, and at 1 point you can finally see it. There's no shortcut.
Chapter 5: Why posts travel
There are a bunch of ways to go viral, but they fall into 2 big categories.
The first is novelty. There's a whole category of people trying to be MrBeast. What they're really doing is creating content that's novel, never seen on the internet before. Another example: imagine Sundar Pichai doing a podcast. The CEO of Google isn't seen much, so anything he says comes out as novel, because you can't find it anywhere else. Anything truly novel, in ways that matter, goes viral.
The second is relatability. Within that, there are all kinds of emotions: content that makes people angry, content that makes people laugh. When people say "make your audience feel something", what they mean is make your audience relate to it. While they're relating, they feel something. But the thing underneath is that they relate.
To go viral, a piece of content has to relate to a very large audience. So the idea can't be aimed at a topic that only a small group of people care about.
And here's the distinction that matters. Relevance is different from relatability, but you need relevance to get relatability. If something isn't relevant to me, I won't even look at it, so I'll never relate to it. Advice on getting healthy is relevant to me, I need it. But I don't relate to all health advice, only to certain pieces. So relevance doesn't create relatability, but without relevance there's no way to ever get relatability. If going to the gym isn't relevant to me at all, there's no way in hell I'll relate to a gym post. To go viral, you need something relevant to a very large audience, and relatable to that audience too.
So how do you make something relatable? There are a hundred ways and endless experiments. But the quickest way is to look at other creators and what people are already relating to, then figure out the equivalent for yourself. Say people are really vibing with the A tier, B tier, C tier rankings on Instagram. And say you sell a niche B2B service. That service isn't relevant to most people. Generating leads is. Far more people care about leads than about your specific service. So you talk about leads, and then you work out how to fit the tier format to leads for your business.
How do you find those formats? By scrolling. Make a long list of creators who post regularly and watch their feeds. Patterns will emerge. It's that easy. You can ask AI to tell you what Unnati Bagga posted every day. Pick 5 LinkedIn ghostwriters who post consistently and ask for a daily review of what they posted and a weekly review of the types of posts.
How do you balance relevance and relatability? Sadly, that only comes from experience. You're really asking how to shape an idea, and that only comes from doing it. Talk about it with me, with yourself, with AI, with anyone. All of that is experience. I haven't found a formula that works wonderfully. AI might be 60 to 70% of the way there, so consult it, but this is something you build.
And don't try to collect all the ways to go viral. You could write a book on it, because the more specific you get, the more methods appear. Knowing ways to go viral doesn't mean you'll go viral. You still have to build the muscle. Pick the 1 or 2 ways you understand, stick with them until you can execute them really well and you've seen results, and then add more to your stack.
The same goes for hooks you admire. Collect them in 1 place, a doc or a Notion database. You can let AI read that list when it writes hooks for you. But the cleaner way is to copy the hooks you like by hand and write real posts with those patterns, for your own account or for clients. That's what puts the hook in your brain. Right now you don't understand those hooks deeply. Writing with them is how you start to.
Chapter 6: How you actually get better
When a writer on my team sends me posts for approval, here's what I do. I look at the hook of the first post. Just the first line or 2. I'm not reading the rest.
If it's really bad, I don't even read the second post. I reject the whole thing. If it's kind of bad, I ask myself whether it's a one-off. If it's a common mistake that's going to be in every post, I reject it all. If it's specific to that post, because the idea or the topic is weak and not the writing, I give it another chance and look at the second and third posts. If the post is good but the hook is really bad, it has failed. If the hook makes sense, I read the next hook, and the next, until I've read all of them.
Once all the hooks work, once these are posts a normal person would actually start reading, I read a little of the body. Never the whole post, never every post. 3, 4, 5 lines across most of them, just to check the rest keeps the quality the hook promised. If it does, it's good to go.
If you write your own posts, read them the same way before you publish. Your readers give you the same 2 lines I do.
Now, how do you find out what's wrong with a post that didn't do well? Something I learned early: most people starting out are lazy. They don't edit their own work enough. Today I can look at a post and pinpoint exactly what's wrong and word the edit quickly. That took time. It only came from editing my own work again and again. I can pick any post on the internet, by anyone, and find edits that would make it better.
Those are subjective improvements. An edit might make a post do worse. But there's always room to improve, and most of the time it's objective, because nobody spends enough time on a piece of content, especially at an agency, where we optimise for speed, or when you're posting every week. The way to find it is to sit, read the post and think. It's hard, slow and painful. Painful because you have to find and admit mistakes in something you wrote or approved. But the more you do it, the better you get at turning a decent post into a good one and a good one into a really good one. So if a post didn't work, sit with it, guess the reasons, make the edits. Practise that.
And 1 small thing about how we work. When there's a review to be done, do it properly. If we're on an input call and you want me to look at a client's posts, that's a separate review call. Ask our account lead to set it up, and bring everything else that needs checking too. It's another 30 minutes, but it's better done properly than squeezed in. The same goes for you with anyone who helps you with content: an input call is for input.
Chapter 7: Working with AI without losing your mind
If I'm writing 1 post, I might spend a lot of time going back and forth. But for a batch, I couldn't figure this out for a long time. Now I have, and it's worked every time I've tried it.
Part of it is getting better at saying the right things to the AI. This is where empathy helps. You try to understand how the AI was thinking. It got it wrong, but why? It was trying to do a good job. It wanted to. It just didn't know where it went wrong. Once you find the core cause, how it misread something, giving feedback that actually gets you good posts becomes easy.
That's also why "these are bad" isn't feedback. Define what was bad. Say the research ideas are too specific, so specific the audience won't enjoy them. Why did it go there? Because it searched for ideas about your niche and forgot that the content is for the audience. So you put exactly that in the prompt: these are relevant to me but too niche for my audience, find ideas the audience will also care about. And when output is just bland, say "this is too boring". That's all. I once told an AI tool an email was too boring, and it came back with an email that was still formal but that people would actually want to read.
Now the exhausting part. If there's a lot of back and forth, I ask the AI whether the ideas are bad or unrefined, and then I refine them through research we can do right now. We can't go back to the founder and ask for more input. So I tell it: we can't go back to the founder, research what we already have about them, private and public sources, refine the ideas, then let's try the posts again. If it's your own account, tell it to research what you already have about yourself.
If it's still too much back and forth, drop those posts. And if every draft is like that, drop the whole thing and sit with me to figure out the big problem. Because a whole batch is almost never bad because of the writing. The writing bot is good. When everything fails, the problem is the input.
I used to say good input makes good posts. That's not quite right, and I didn't see it clearly back then. Good input has to be turned into good, well-shaped ideas, and those make good posts. Sometimes a good idea already has a shape and you don't need to refine it. But a lot of the time, even with a good idea, AI still sucks, because the idea hasn't been given a shape. You can tell it's a good idea because of all the connections in your head. AI sometimes can't. So refine every idea before you ask for the post.
Chapter 8: Writing as yourself
This one was first said to my team, about their own profiles, not clients. It works the same for any founder.
If you want to get 100k impressions in a week, and actually beat me, write a lot of posts by hand. Share your raw thoughts, what you're doing, what you're thinking, throughout the day. Not in 1 sitting where you write a batch in an hour. Through the day, like you're dropping texts to your friends. Imagine you're talking to your friends through LinkedIn, publicly. Nothing beats that.
The moment you write like that, the reader starts to feel like your friend. They feel involved in your life. They want to talk to you more, because you're a good, interesting person. Most people are good, interesting people. If they talked on the internet the way they talk to a friend, they'd have a huge audience. Because the internet is full of slop. Nobody wants slop. It's full of advice, and people are exhausted by advice. They want relatable things. Talking to them like a friend is relatable, it's relevant, and it makes them curious.
When you post a lot, some posts become outliers. They do more reach than all your other posts combined. The only way to get those is to post more.
And the more you post, the more you're in the momentum of writing, and when you're in that momentum, you naturally write well. I don't really know how ovens work, but think of bread. Put it in an oven that isn't on and it takes ages to get golden and crisp. Put it in an oven that's been on for an hour and it's golden in 30 seconds, because the oven is already hot. Writing is the same. If you've been writing, the oven is hot, and whatever you write comes out well. If you haven't, it takes time to heat up and get into the flow again.
If you write for someone else
Most of this book works the same if you write for a founder instead of yourself. A few parts were first said to writers on my team, and here's what they look like in that form:
- Know what your client wants (Chapter 1). If you're stuck on a client, you usually haven't had that conversation. Have it. "I don't know what you want. How am I supposed to help you if I don't know what you want?"
- Running the story call (Chapter 2). Walk the founder through the timeline. Explain that wins and losses are only a memory aid, so they don't get confused.
- Repeating stories (Chapter 2). If you, the person who works on the account, don't remember the old stories, the audience doesn't either.
- Separate review calls (Chapter 6). Don't squeeze a review into an input call. Book it separately and bring everything that needs checking.
- No founder access (Chapter 7). You usually can't go back to the founder for more input. Tell the AI that, and refine from what you already have.
- Practise on real accounts (Chapter 5). Write real posts with the hooks you admire, for your own account or a client's. Watch about 5 LinkedIn ghostwriters who post consistently.
- Before you send for approval (Chapter 6). Whoever approves your batch will likely read only the first hook. If it fails, the whole batch fails. Read the full posts yourself before you send. If a whole batch keeps failing, drop it and sit with your lead to find the real problem.