
The NEW LinkedIn Content Strategy Framework To Get Leads, Clients, Or Both
May 13, 2024Look, I'm going to show you exactly how to use AI to come up with content ideas. Not the theoretical stuff. The actual process I just used to create 25 pieces of content.
Unlike other tuts prompting AI to "write me a post about marketing." That's amateur hour and everyone can tell it's AI garbage.
This is about using AI to synthesize what you know with how your audience actually talks. Then turning that into content that doesn't sound like a robot wrote it.
Here's how it works.
Step One: Map Out Your Content Foundation
Before you even open ChatGPT or Claude or whatever, you need to do some thinking. Grab paper or a tablet or whatever you use.
Draw two columns.
Left column is where your prospects are right now. Write down everything. Their current situation. Their problems. What they're struggling with. The misconceptions they believe. All of it.
Right column is where they want to be. Their dream outcome. What changes when they work with you. What gets better.
Be specific. Don't write vague stuff like "they want success." Write what success actually looks like for them.
I'll give you my example. I'm selling an account-based marketing course.
Left column says my prospects have a $5,000 offer. They're paying for ads that don't work. They have a good product but lead gen marketing isn't cutting it. They don't know how to use LinkedIn properly. They believe marketing on social should be free. They can't focus on any one strategy long enough to make it work.
Right column says they want to triple their income. They want better clients. Dream clients who show up ready to buy. They want ads to work without spending a fortune.
Write as much as you can in both columns. The more specific, the better this works.
Now comes the important part.
Write down the three big steps that get someone from the left column to the right column. The transformation. The journey. Whatever you want to call it.
For me it's pick 500 perfect companies, find decision makers and influencers, create personalized experiences.
Under each of those three steps, break it down further. What are the smaller actions or activities that fall under each big step?
Under "pick 500 perfect companies" I wrote things like research lists of companies, look at your CRM for evidence, check LinkedIn engagement, attract and filter using workshops.
Do this for all three steps. You should end up with at least five sub-points under each main step.
Count them up. You just created at least 30 different content angles. That's a month of posts right there.
And you didn't use AI yet.
Step Two: Steal Your Audience's Language
Here's where people mess up with AI. They feed it generic information and get generic output.
You need to give AI something specific to work with. Specifically, the exact words your audience uses when they talk about their problems.
Go to Reddit. Search for your topic. Find a thread where people are actually discussing it. Copy the whole conversation.
I searched "account based marketing" on Reddit. Found a thread. Copied everything.
You can also do this with YouTube comments, Facebook groups, Twitter threads, whatever. Just find where your people hang out and what they're saying.
The key is you're not looking for ideas here. You're looking for language patterns. How do they describe their frustrations? What words do they use? What phrases come up over and over?
You're going to feed this to AI in a minute. But first you need one more thing.
Step Three: Prep Your Sub-Steps for AI
Remember those sub-steps you wrote down? The ones under your three main steps?
Type those into a spreadsheet or a document. Just list them out clearly.
From my example, I typed researching companies from Google search, exploring your CRM for examples of dream clients, looking at people who engage with you on LinkedIn, searching LinkedIn for companies, seeing who grabs your lead magnets, segment the top 20%.
That's it. Just a simple list of the activities.
Now you have two things ready. Your sub-steps and your Reddit scrape full of language patterns.
Step Four: Let AI Do the Heavy Lifting
Open Claude or ChatGPT. I use Claude but either works.
Here's the prompt I used and you can steal it word for word.
"The very first step in [YOUR TOPIC] is [DESCRIBE FIRST BIG STEP]. Below I have listed out some of the steps that people can take to [ACCOMPLISH THAT STEP]. Attached to this prompt is a scrape from Reddit from people talking about [YOUR TOPIC]. What I'd like you to do is analyze that data and then link it to each of the steps below using the language patterns that you find in the Reddit scrape. Come up with topic ideas for social content based on each of the steps using the language patterns you find in the Reddit scrape."
Then paste in your list of sub-steps and the Reddit scrape.
Hit enter.
What happened when I did this? Claude gave me 25 blog post titles. All of them good. All of them using language patterns from real people talking about real problems.
Things like "Stop Googling Companies That Need Your Service. Here's What to Search Instead" and "Your CRM Already Knows Your Perfect Clients. Here's How to Find Them."
I didn't have to think. I didn't have to brainstorm. I just combined what I know with how my audience talks and AI connected the dots.
That's the whole trick. You provide the expertise and the language. AI provides the synthesis.
Step Five: Turn Ideas Into Actual Posts
Okay, you have 25 topic ideas now. Time to write an actual post.
Pick one of the titles. Now you need to give AI some source material to work with. Go to YouTube and search your topic. Find a video that covers something related to your chosen title.
Open the transcript. You know that little button under the description? Click "show transcript." Copy the whole thing.
Back to Claude or ChatGPT. Here's the prompt.
"I'd like to write a short LinkedIn post for this topic. [PASTE YOUR CHOSEN TITLE]. You can use the included video transcript below. However, the video transcript is not me speaking. So if you're going to use any language patterns from the video transcript, you've got to change it so that we don't have any copyright issues. This post should be short and to the point and should have the following format. State the problem. Explain why it's a problem. Give a single data point about the problem. Then some practical advice about solving the problem."
Then paste in the video transcript.
Also paste in your natural writing prompt. The one about avoiding AI giveaway phrases.
Hit enter.
Claude will spit out a post. It will be okay. But it won't be great yet.
Step Six: Edit Out the AI Tells
This is the part most people skip. They take whatever AI gives them and post it. And everyone can tell it's AI.
Copy what AI wrote into a document. Now you're going to edit it.
Look for phrases like "here's why it doesn't work" or "let me share some data with you" or "better approach." These scream AI.
Change them to sound conversational. Instead of "here's why it doesn't work," write "I was talking to somebody the other day and we were tossing around ideas why it doesn't work."
Instead of "better approach," write "I wonder if you'd be open to considering a different approach."
See the difference? The second version sounds like a human talking to another human.
Add some filler language. The kind of stuff people actually say. Things like "you know what I mean?" or "here's the thing though" or "look."
Cut anything that sounds like marketing copy. If you read it out loud and cringe, change it.
This editing process takes maybe five minutes. But it's the difference between content that sounds AI generated and content that sounds like you.
Step Seven: Create the Visual
You need an image for your post. You could use a stock photo but those are boring. You could hire a designer but that's expensive and slow.
I use Nano Banana. There are other AI image tools but this one works well for me. You need to get good at prompting though.
My prompt was something like "create an image of me sitting at a coffee shop working on a laptop, casual setting, natural lighting, medium shot, sitting at a normal sized table."
I had to specify the table size because the first version gave me this tiny table that looked ridiculous. I also had to specify to include my face because I forgot that part initially and it generated some random person.
The output was good enough. Two coffee cups on the table. Me looking focused. Believable.
Download it and you're done.
Step Eight: Post It Everywhere
Write your call to action. Keep it simple. "If this sounds interesting, I'm doing a free class on it. DM me for the link."
Post it on LinkedIn. Post it on Facebook. Post it on your YouTube community tab. Wherever your people hang out.
The whole process from picking the topic to posting took me about 20 minutes. And I was narrating the whole thing on a live call.
If you're just doing it without explaining every step, you can probably do it in ten minutes.
Why This Works When Other AI Content Doesn't
Most people use AI wrong. They ask it to create something from nothing. "Write me a post about marketing." That's garbage in, garbage out.
What I showed you is different. You're giving AI three specific inputs.
Your expertise broken down into steps. That's the what.
Your audience's language from real conversations. That's the how they talk.
Source material from someone else covering the topic. That's the proof it's a real issue.
AI just connects those three things. It's synthesizing, not creating.
Then you edit it to sound like you. Remove the corporate speak. Add your personality. Make it conversational.
That's why the output doesn't sound like AI wrote it. Because AI didn't write it. AI just helped you organize what you already know into content that resonates with how your audience actually communicates.
The Part Everyone Forgets
Here's what matters more than any of this. You need to actually know your stuff.
AI can't fix it if you don't understand your topic. It can't help you if you haven't thought through the transformation you provide. It can't write in your audience's language if you don't know where they hang out or what they struggle with.
Do the mapping exercise first. Figure out where they are, where they want to be, and the steps to get there. Break those steps down into specifics.
That foundation is everything. The AI part is just speeding up the execution.
Once you have that foundation, you can create content faster than you ever thought possible. And it won't sound like garbage because you're starting with real expertise and real language patterns.
I just showed you how to create 25 pieces of content in two hours. You can do the same thing for whatever you're selling.
The process is simple. Map your transformation. Steal your audience's language. Feed both to AI. Edit the output to sound human. Post it.
Do that once and you have a month of content. Do it for each of your main steps and you have three months of content.
Stop overthinking it. Just follow the process.

