What Changed and Why It Matters
Sources:
- YouTube: "What Changed on LinkedIn in 2026" (Compound, founder content agency managing 30+ founder-led accounts)
- YouTube: "How I Fully Automated My LinkedIn Posts Using Claude Code" (Claude Code + GPT Image 2 + Cloudinary + Buffer carousel system)
- LinkedIn Hook and Post Template Library (50+ formulas: hooks, post bodies, lead magnet posts)
Something changed on LinkedIn this year and most founders can feel it without being able to name it. Your posts used to do fine. Same effort, same topics. Now the reach is not there. So you blame the algorithm and post less.
Here is what actually happened: the timeline got significantly more crowded. Two forces collided. First, AI tools lowered the barrier to create content to near zero so the volume of average posts went up dramatically. Second, more executives and founders than ever are now posting, because founder-led content is now a mainstream B2B marketing strategy rather than a fringe one. LinkedIn is also suppressing organic reach in favour of paid media, and the product itself is making questionable decisions (same posts re-appearing on refresh, suggested posts with three likes that have no relevance to you).
Despite all of that, LinkedIn remains the default B2B content marketing channel. Until a true competitor arrives, it stays that way. The question is how to adapt.
This playbook covers: what changed, what the new strategy looks like, how to automate the production side with Claude Code, and the 50+ post templates you can use today.
The Three Content Moats Against AI Slop
Source: Compound (managing 30+ founder LinkedIn accounts)
The founders who win on LinkedIn right now are publishing content that is impossible for AI to copy. Three moats make your content uncopyable.
Moat 1: Narrative mode
The core shift: it is no longer enough to post generic advice or how-to content. Five tips for growing on LinkedIn is not enough on its own. You have to pair every take with a story or a specific anecdote that only you could have written.
Two ways to use narrative:
- The entire post is one singular story. The most powerful version of this for founders is the origin story post: how and why you started the company. Run this once. It consistently outperforms.
- Supporting anecdotes embedded in advice content. The whole post does not need to be a story. Even this strategy piece uses specific anecdotes from day-to-day experience to make it non-copyable. Look at a post that performs and ask: what specific detail in here could only come from the person who wrote it?
Example from Compound: a simple hiring post that got nearly 60,000 impressions and almost 600 likes. Why? It included specific anecdotes: naming the two hires they just made, a line about the founders being the de-facto head of client services for the first three years, and context about DMs asking about freelance writers. None of that is something you would get from asking Claude to write a hiring post. That specificity is the moat.
Common mistake: founders hear narrative mode and think every post needs to be a long-form origin story. Those take almost a full day to write well. Do not rely on those alone. Use supporting anecdotes in shorter posts. One specific detail that only you could know is enough.
Critical filter: the anecdotes and stories you include must be relevant to your ICP. Personal life posts (I got engaged, here is what it taught me about B2B sales) can get engagement, but the quality of that engagement is wrong. The impressions are from people who will never buy from you. Use stories and anecdotes that come from sales calls, conference conversations, and experiences with past or current customers.
Moat 2: Data mode
Share unique proprietary data that only you have access to. AI cannot copy data it does not have. On top of that, LinkedIn audiences respond strongly to data visualisations. If you can put unique data into a chart or graph, it outperforms almost any other format.
Example: Peter Walker at Carta (Head of Insights) consistently posts graphs and charts using Carta data about the startup and VC ecosystem. Seed round trends, valuation data, round volume. This data only exists inside Carta. Every post lands well. Strong like and comment counts, almost certainly driving dark social sharing in Slack channels and group chats.
What you likely already have:
- Response rate data from your outreach sequences
- Conversion rates at specific funnel stages
- Industry benchmarks you have collected from working with clients
- Aggregate anonymised client performance data
- Internal hiring or team growth data
- Revenue milestones or growth trajectories
You do not need a data science team. A screenshot of a chart from your analytics platform with your own commentary is enough to qualify as a data mode post.
Moat 3: Physical mode
Taking things from real life and putting them on LinkedIn as content. IRL photos. Real faces. Real locations. Proof that a human made this.
Why it works: LinkedIn is actively working to filter out AI-generated content. Real photos are the clearest possible signal that your content is not AI-generated. They also drive higher engagement because people engage with people, not with graphics.
Practical applications:
- New hire posts: take a photo with the hire and include it. Outperforms text alone.
- Conference or event posts: always get photos and B-roll. The social footage from an event is often more valuable than the event itself.
- Office behind-the-scenes: turn your office into a content set. Have someone on your team snap photos of you going through your day. It feels odd at first. The library of images pays dividends over months.
- Milestone moments: NASDAQ billboard photo, office move, team offsites. The post about the billboard gets more impressions than the billboard.
Where to start: if you have an in-person office, assign someone to follow you around with a phone for one day per week. Capture the real moments. That library becomes the physical mode content you draw from for months.
Bending the Algorithm: The Daily System
Source: Compound
You cannot just rely on LinkedIn's organic algorithm to serve your content to the right people. Comments can get more impressions than posts themselves right now. Here is how to bend the algorithm in your favour.
Daily connection strategy (20 per day)
LinkedIn gives you more control over your follower base than almost any other social platform. When you send a connection request and it is accepted, it creates a mutual follow. They are likely to see your next one to two posts. If they engage, they enter the flywheel and keep seeing your content.
The tactic: send 20 outbound connection requests every weekday. Monday through Friday. That is 100 per week, which is near the upper limit for most LinkedIn accounts. Every week you are manually seeding your ICP into your audience.
Who to send to: your ICP, not random people. This requires having a clear ICP definition. Filter by title, industry, company size, geography, and send only to people who match. Quantity means nothing if the connections are not your buyers.
Note on messages: research shows connection requests without a message have higher acceptance rates than those with a message. A message looks salesy. A clean connection request looks like a genuine peer who found you interesting. Do not attach a pitch to your connection requests.
Daily commenting strategy (5-10 per day)
In the same 10-15 minute window as your connection requests, leave 5-10 comments on other people's content. Specifically: other people in your ICP, or influencers your ICP also follows.
Why it works two ways:
- Reciprocity. If they see you consistently commenting on their posts, they will check out your content. If it is relevant to them, they engage. That is usually what happens on social platforms.
- Comment visibility. People in your ICP who follow the accounts you commented on are seeing your comments in the comment section. Comments are getting more impressions than posts right now on LinkedIn. Your comment on a high-reach post can reach more of your ICP than your own post.
Do not spend hours doing this. 15 minutes per day. 5-10 comments. Done consistently, this is one of the highest-leverage activities available on the platform for a B2B founder.
What to comment: add something the post did not say. Agree or disagree with specifics. Ask a question that extends the conversation. Do not just say great point or I agree.
Thought leader ads: leaning into the suppression
LinkedIn is suppressing organic reach and pushing paid. The mistake is to fight this. The better move is to lean into thought leader ads.
Why thought leader ads work differently: they do not look like ads. The only indication is a tiny promoted text under the author's name. If you take your best-performing organic posts (specifically those that attracted ICP engagement) and put budget behind them, you get controlled reach to the right people without the creative looking like a paid ad.
How to target precisely:
- Use Apollo to build a list of companies matching your ICP.
- Upload that company list directly into LinkedIn Ads.
- The match rate is imperfect but often better than LinkedIn's native filters.
- The right people from the right companies see your content.
Budget to start: $30 per day ($900-$1,000 per month). In B2B ad budgets, that is a small test. Scale up if a specific post is performing well.
The link trick: run the post organically without a link for about a week (links do not hurt performance anymore but you do not want to add one to every post organically). After a week, edit the post and add a URL as a PS line. That URL shows up even when the post is boosted as a thought leader ad and serves as your call to action, since thought leader ads can only use brand awareness or engagement objectives. You cannot add a direct CTA button, but the PS line URL is the workaround.
Claude Code: Automate Your Carousel and Post Production
Source: YouTube transcript (Claude Code LinkedIn auto-poster system)
Full pipeline: topic in, carousel PDF out, scheduled to LinkedIn via Buffer. Built in one Claude Code session. Open-sourced to GitHub.
What the system produces
A multi-slide LinkedIn carousel post, fully generated by Claude Code:
- 4-6 on-brand image slides using GPT Image 2
- Consistent visual style across all slides (same colours, font, layout)
- Slide numbering (1 of 5, 2 of 5) to encourage progression through the carousel
- A comment CTA on the final slide (Comment CRM and I'll DM it to you)
- The full LinkedIn post copy written alongside the carousel
- Everything stitched into a PDF (required format for LinkedIn native document posts)
- Uploaded to Cloudinary for public URL hosting
- Scheduled to LinkedIn via Buffer API
All from one message to Claude Code. You give it a topic or a video title. It does the rest.
The technical stack
| Tool | Role |
|---|---|
| Claude Code | Orchestrates the entire pipeline. Writes and runs all code. |
| GPT Image 2 (OpenAI) | Generates the carousel slide images. Best image model available as of 2026. |
| Cloudinary | Stores images and PDF with public URLs. Required for the Buffer API to retrieve the PDF. Also auto-tags all assets with AI for easy retrieval. |
| Buffer | LinkedIn post scheduling tool. Free plan works. Claude Code connects to the Buffer API to schedule the post automatically. |
| LinkedIn MCP or Linked API CLI | Optional: for posting directly to LinkedIn without Buffer. |
Why Cloudinary: Buffer's API requires a public URL for the PDF carousel. You cannot just host the file locally. Cloudinary gives you a public URL per asset, auto-tags everything with AI, and makes retrieval easy if you need to find or modify an asset later.
How to build it
Step 1: Set up Cloudinary
"Set up Cloudinary media storage in this folder.
Whenever images or videos are generated in this project, also upload them to Cloudinary automatically.
Ask me for my Cloudinary account details."
Claude finds the Cloudinary API documentation, asks for your cloud name and credentials, and sets up the integration. Add your API key via environment variables, not directly in chat (security).
Step 2: Set up the carousel generator
"I need you to become a LinkedIn carousel post creator.
I've uploaded style reference images from my existing content for you to take the design style from.
Use the GPT Image 2 model for image generation (it is now available as of 2026).
Store all generated images in Cloudinary.
Create 4-5 slides that flow together as a cohesive carousel.
Each slide should include a slide number in the corner (1 of 5, etc).
The last slide should include a comment CTA.
Topic: [PASTE YOUR TOPIC OR VIDEO TITLE]"
Step 3: Add a LinkedIn post writer skill
The default post copy Claude generates will be too generic. Download the LinkedIn post writer skill from MCP Market (search: linkedin post writer) and upload it to your Claude Code instance via Settings > Skills > Upload. Then ask:
"Using the LinkedIn post writer skill, write the post copy to accompany this carousel."
Step 4: Connect to Buffer and schedule
"I've set up a Buffer account and connected it to my LinkedIn.
Schedule this carousel as a LinkedIn post for tomorrow morning at 8am.
Connect to the Buffer API to do this automatically."
Claude finds the Buffer API docs, sets up the connection, combines the images into a PDF, enables PDF delivery in Cloudinary settings (you'll need to enable this once in your Cloudinary account), uploads the PDF, gets the public URL, and schedules the post via the Buffer API.
For future posts: give Claude a new topic and it generates everything in one message.
Step 5: Edit carousel slides if needed
If any slide needs changes, make modifications in Cloudinary using their AI editing tools (generative fill, background removal, colour adjustments), tell Claude which file was updated, and it will automatically re-stitch the PDF and re-send to Buffer.
API key security
Never paste API keys directly into the Claude Code chat. They can be found in the code easily. Instead:
"I need to add my OpenAI API key securely. Can you give me instructions on the right way to store it?"
Claude will instruct you to add it as an environment variable. This is the correct approach for every API key in this system (OpenAI, Cloudinary, Buffer).
The 50+ Template Module: Hooks, Posts, and Lead Magnet Formulas
Source: LinkedIn Hook and Post Template Library (full lead mag available separately)
Every formula below is fill-in-the-blank. Replace brackets with real, specific data.
No em dashes anywhere. One idea per line. Three lines maximum per paragraph.
The 5 hook mechanisms
Every winning hook works through one of five mechanisms. Know which one you are using before you write.
Curiosity gap: the reader does not know the answer and has to click to get it.
Contrarian claim: the first line directly contradicts something the reader believes.
Specific number: a precise figure implies real, detailed information follows.
Pain recognition: the first line names something the reader is actively frustrated by.
Stakes escalation: the first line implies something significant will be won or lost.
10 proven hooks for B2B founders
# 1. Mistake with specific loss (curiosity + stakes)
I made X huge mistakes during my first Y months of [endeavour].
(these cost me [£155k+ and 417 hours])
# 2. Hard truth reveal (pain recognition)
Here's the no.1 mistake [specific group] make on LinkedIn
(this may be hard to hear):
# 3. External blame reframe (relief + curiosity)
Your [skill] is lacking because [external factor] let you down
(But fix this 1 thing and it'll be 10x better)
# 4. Not your fault (contrarian + stakes)
Stop [common misguided behaviour], [related actions], then [common follow-up]
It's [list negative outcomes].
# 5. The specific progression (credibility + story)
At [age], I began [endeavour].
At [later age], I [achieved significant milestone].
Here are X lessons that helped make this happen:
# 6. ICP callout (direct relevance)
If you're a [specific audience], you're likely making this mistake:
[common mistake].
# 7. The reframe (contrarian)
Most people know exactly [common knowledge].
The problem isn't [common limiting belief].
# 8. Direct address (authority + stakes)
[Target audience]: You need [critical element].
(And not just because [superficial reason]).
# 9. Proven strategy (specific number + story)
I [achieved desirable outcome] in just [short time frame].
I also [additional related outcome].
But there's a strategy behind this.
# 10. Trend callout (data + pain)
There's a rising trend in [industry/platform] to [behaviour],
[get sub-par outcome] then [misguided conclusion].
5 post structures for B2B founders
# 1. Contrarian opinion (thing everyone does is dumb)
[thing everyone does] is dumb.
People do [thing]. [They only get X results]
I [did not do that thing]
Never even thought about [doing it]
[I'm getting significantly better outcome]
Skip [the dumb thing]
[your name], out
# 2. Secret revealed (the triple repetition)
People ask me how I [achieved desirable outcome].
My secret? I [daily action].
People ask me how I [achieved desirable outcome].
My secret? I [daily action].
People ask me how I [achieved desirable outcome].
My secret? I [daily action].
But the truth is: there is no secret. Just [core principle].
Here are X steps to get you started:
1. [Step]
2. [Step]
3. [Step]
# 3. 3 reasons you don't get what you want
X reasons [audience] don't [take desired action].
1. [Reason]
2. [Reason]
3. [Reason]
Here's the solution:
1. [Step]
2. [Step]
3. [Step]
[Summarise the benefit]
PS. [Personal insight or content tease]
# 4. I achieved X without Y
I [achieved goal Y] without [common pre-requisite 1].
I [achieved goal Y] without [common pre-requisite 2].
I [achieved goal Y] without [common pre-requisite 3].
To [achieve goal Y], all you really need to do is:
1. [Core action]
2. [Core action]
3. Rinse and repeat.
Everything else is just a distraction.
# 5. Client story
How my client [achieved desirable outcome] in [timeframe]
(with [specific method])
[Client's starting point]
But they [encountered problem]
They said: "[Direct quote from client]"
I suggested they [solution].
- Reason why
- Reason why
Here's what we did:
- Strategic step 1
- Strategic step 2
- Strategic step 3
The result? [Key outcome].
PS. If you want to achieve the same in [timeframe], [CTA]
5 lead magnet post formulas
CTA formats that work: Comment "WORD" and I'll DM you / Like and comment "WORD" / Connect + comment "WORD"
# 1. After X years of testing
After X years of testing, I've finally created the most effective [channel/mechanism] playbook ever.
([insane numeric outcome]. Not a joke.)
Most [target audience] [do unstructured thing] and hope for the best.
I've built a repeatable system. It focuses on two key things:
[Key thing 1] and [Key thing 2].
So I put together a full document with our whole process:
1) [What is covered]
2) [What is covered]
3) [What is covered]
Just connect with me and comment "[thing]" and I'll send it your way.
# 2. I charge clients X for this
I charge $X to implement this system for my clients.
But here it is for free.
[why the thing matters]
Because by having this in place, you can:
- [achieve desirable outcome]
- [achieve desirable outcome]
- [achieve desirable outcome]
Yet, most [target audience] don't have [specific thing].
So I created a free mini-course showing you how to [do specific thing] in less than [timeframe].
It includes:
- [what it includes]
- [what it includes]
Want access?
- Like this post
- Comment "[WORD]"
- Send me a connection request so I can DM you
# 3. These frameworks got X outcome
These [mechanism] frameworks generated [tangible outcome].
We didn't reinvent the wheel.
We collected the best frameworks from top [industry] experts.
And we tested them.
We ditched the ones that didn't perform and kept the ones that did.
Just comment "[thing]" and connect with me and I'll DM you.
# 4. You will never get X if you don't do Y
You will NEVER get [desirable outcome].
If you don't [do specific thing].
I see this every day.
People trying to [achieve outcome] without [specific thing]
[more context]
I've helped [X target audience] get [desirable outcome], and I've documented our exact process.
Inside my [helpful resource]:
- [value chapter]
- [value chapter]
- [value chapter]
Comment "[thing]" and connect with me for access.
# 5. This is the biggest arbitrage on LinkedIn right now
This is the biggest arbitrage on LinkedIn right now.
In the past [timeframe], I have:
- [growth metric]
- [growth metric]
- [pipeline metric]
And [X%] of my growth has come from ONE main strategy:
[Strategy name].
So I just created a [word count]-word guide breaking down my exact process, including:
- [specific element]
- [specific element]
- [specific element]
Want access? Like this post. Comment "[word]".
I'll send you a copy.
Quick Reference: The Full System in One Place
Everything from this playbook condensed into one daily and weekly system.
What changed and what the new strategy is
What changed: more people posting, AI tools flooding the timeline with average content, LinkedIn suppressing organic reach in favour of paid.
What to do about it:
- Create uncopyable content using one of three moats: narrative (stories and anecdotes only you could write), data (proprietary numbers and visualisations), or physical (real photos and IRL moments).
- Bend the algorithm manually: 20 outbound connection requests per day to your ICP, 5-10 comments per day on ICP-relevant posts.
- Lean into paid: use thought leader ads on your best-performing organic posts, target via an Apollo-built company list uploaded to LinkedIn Ads, start at $30 per day.
Daily LinkedIn routine (15 minutes)
# Morning (15 minutes)
"Open LinkedIn. Send 20 connection requests to [ICP TITLES] at [COMPANY TYPE] companies.
No message attached. Log each to outreach/connections-[DATE].csv.
Stop at 20. Wait 3-5 seconds between each request."
"Leave 5-10 comments on posts from [ICP-ADJACENT ACCOUNTS / INFLUENCERS YOUR ICP FOLLOWS].
Each comment must add something the post did not say.
No 'great post' or 'I agree'. Extend the conversation or add a specific data point."
Weekly content production (Claude Code)
# Generate a carousel and schedule it
"I need a LinkedIn carousel post.
Topic: [PASTE TOPIC OR VIDEO TITLE]
Use the GPT Image 2 model.
Reference the style guide images in /style-guide/ for consistent design.
4-5 slides, numbered in the corner, comment CTA on the last slide.
Store all images in Cloudinary.
Combine into a PDF.
Write the LinkedIn post copy using the LinkedIn post writer skill.
Schedule via Buffer API for [DAY] at [TIME]."
# Generate post copy only (no carousel)
"Using the LinkedIn post writer skill, write a LinkedIn post about [TOPIC].
Read my voice from CLAUDE.md.
Hook formula: [CHOOSE FROM SECTION 4 OF THIS PLAYBOOK]
No em dashes. Under 900 characters for top-of-funnel posts.
CTA: [DESCRIBE]"
The content filter to run on every post
Before publishing any post, run it through this filter:
- Is there a narrative, data, or physical moat in this content?
- Could this exact post have been written by someone who entered my topic into Claude?
- Is there at least one specific detail (a name, a number, a quote, a date, a place) that only I could have included?
If the answer to question 2 is yes or the answer to question 3 is no, the post needs more work before it goes out.
If none of those three checks pass, you still have work to do.
Template quick selector
| Goal | Hook to use | Post structure to use |
|---|---|---|
| Share a lesson from experience | Mistakes that cost dearly | Secret revealed |
| Position against common advice | Hard truth reveal | I achieved X without Y |
| Build authority with a client story | ICP callout | Client story |
| Drive comments for a lead magnet | Proven strategy | After X years of testing |
| Reframe a common belief | The reframe | Contrarian opinion |
| Share proprietary data | Specific number | 3 reasons you don't get what you want |
| Drive DMs with a free resource | Progression milestones | I charge clients X for this |
| Announce a new lead magnet | Stakes escalation | This is the biggest arbitrage right now |
