Eric Siu ✓
Founder @ Growth Marketing Agency - Single Grain, Podcaster @ Marketing School + Leveling Up
Eric Siu writes about AI adoption, marketing operations, and founder leadership for a 55K+ audience, posting 7.7x per week with surgical consistency. His most distinctive move is the pattern-interrupt hook that inverts conventional wisdom (e.g., 'Paid media used to mean buying ads. Now it can mean paying 1,500 editors') followed by a real-world case study that proves the counterintuitive claim.
7.7/wk
100 posts in 30d
12
avg per post
20.2%
comments / total
3.7%
reposts / total
mixed
Fri · 11 AM CET
Voice & tone
Direct, operator-focused, and slightly provocative without being clickbait. Siu writes like someone who just ran an experiment and is sharing the raw findings. He uses phrases like 'The money is interesting, but the operating model is the real story' and 'That is the shift marketers should be watching' to signal that he's teaching pattern recognition, not just reporting news. The tone assumes the reader is building something and wants to move faster.
Signature phrases
- “The money is interesting, but the operating model is the real story.”
- “That is the shift marketers should be watching.”
- “Here's what we missed.”
- “The result: A-players went crazy building things I never asked for.”
- “Mandate the measurement, not the tool.”
- “The job of the founder is to push the pace. If they lag, the entire org lags.”
Top 3 posts, dissected
Their highest-engagement posts, broken down line by line. Steal the structure.
Want to drive real AI fluency? Save this POV by Eric Siu ✓ 👇 "Not establishing AI fluency within any org is a death wish. Here's how we're thinking about it: 1. Founders need to trumpet AI from the top Founders' jobs are to be the 'chief reminder officers' -- to continually repeat the message until it starts to feel annoying. Share the new hotness, experiments you're running, things your friends are doing, etc. The job of the founder is to push the pace. If they lag, the entire org lags. But that's not enough. 2. Champions throughout the org need to push the message People working at junior, mid, or senior levels who *LOVE* AI need to champion it so it starts to feel like a team effort. It's not enough to just come from the top down. It has to be bottom up. We have a weekly 'AI corner' and an 'AI working group' channel where people are expected to share their learnings. 3. Establish an AI fluency table. An example for an engineer: L0 - doing everything manually L1 - using ChatGPT to help you with some basic coding tasks L2 - using Cursor to help with your code generation L3 - Creating 10 agents of yourself doing rote work that you'd rather not do Do this for every role and give everyone a timetable. For example, you might expect everyone in the org get to at least L2 for their role within the next 9 months. 4. Offer to help, but only help those who are willing At our company, we're more than happy to help those who are willing to learn. These people are GENUINELY excited about the opportunity and it's easy to see it on their faces. But if someone resists or makes excuses, it's tough to make a case to help. We can take a thirsty person to fresh water, but we can't make them drink. 5. Help people get to the 'aha' moment Once an engineer sees how IDEs like Cursor work, their 'aha' is realizing that they're 5-10x faster. Then they see products like Devin or Codex, and it starts to click that they can now create duplicate versions of themselves and better yet, work on high-leverage work that only they can do. How are you spreading AI fluency throughout your company?" -- 👋 P.S. What else would you add to the list? Make sure you give Eric a follow btw!
'Want to drive real AI fluency?' is a question hook that opens with reader benefit (fluency) and uses the word 'real' to signal that the post will debunk fake approaches. The follow-up 'Save this POV by Eric Siu' immediately establishes that this is a framework worth keeping, reducing friction to engagement.
The post uses a numbered list (1, 2, 3) to break down a three-layer organizational model: founder mandate, champion network, and fluency leveling. Each section includes a problem statement, a solution, and a concrete example (e.g., 'L0 - doing everything manually' through 'L3 - Creating 10 agents of yourself'). The structure mirrors a playbook, making it scannable and actionable.
- Founder as 'chief reminder officer' — reframes a leadership role using unexpected language that sticks
- Contrasts top-down with bottom-up to show that mandates alone fail — teaches systems thinking
- Levels framework (L0-L3) gives readers a template they can copy into their own org immediately
The post cuts off mid-sentence ('Do this for every...'), which creates curiosity and forces readers to comment asking for the rest. This is a deliberate engagement hack that drives the conversation index up.
Use numbered frameworks with concrete examples (L0, L1, L2) to make abstract concepts (AI fluency) feel implementable, and leave the post slightly incomplete to trigger comments.
Paid media used to mean buying ads. Now it can mean paying 1,500 editors to flood the internet with clips. Clavicular reportedly made $1.1M in January, with over 90% coming from the Kick Creator Incentive Program. During peak streams, he was pulling in around $20K per day. …
'Paid media used to mean buying ads. Now it can mean paying 1,500 editors to flood the internet with clips.' This inversion hook works because it redefines a familiar term (paid media) in a way that contradicts the reader's existing mental model, forcing them to read on to understand the new definition.
The post opens with the hook, then provides a case study (Clavicular's $1.1M revenue breakdown). It then pivots to the operating model analysis: 'The money is interesting, but the operating model is the real story.' The final section generalizes the pattern to marketing broadly, ending with a forward-looking prediction about 'the next generation of media buyers.'
- Specific revenue numbers ($1.1M, $20K per day, $50/month) create credibility and make the case study feel real
- Repeats the phrase 'The audience sees organic reach, but the operator sees spend' to hammer home the mental model shift
- Ends with a prediction ('They might be buying enough output that the algorithm eventually gives in') that feels like insider knowledge
No explicit CTA. The post ends with a forward-looking statement that positions the reader as someone who should be watching this shift. The implicit CTA is 'think about how this applies to your business.'
Use a real case study with specific numbers to prove an abstract claim, then zoom out to show the broader pattern. This structure makes the post feel both grounded and predictive.
Paid media used to mean buying ads. Now it can mean paying 1,500 editors to flood the internet with clips. Clavicular reportedly made $1.1M in January, with over 90% coming from the Kick Creator Incentive Program. During peak streams, he was pulling in around $20K per day. …
'Meta is firing engineers who don't use AI. We tried the same thing at Single Grain. It almost backfired.' This contrarian experiment hook works because it signals that the post will contain a lesson learned from failure. The reader expects to learn what went wrong and how to avoid it.
The post opens with the hook, then explains the failed mandate ('everyone becomes Claude Code proficient within 6 months'). It then diagnoses why it failed ('Code is verifiable... Marketing output takes 90 days to prove'). The final section offers three solutions: 'Beat Claude' hiring challenge, Friday AI agent, and mandate measurement instead of tools. Each solution is explained in 1-2 sentences.
- Diagnosis section ('Here's what we missed') shows self-awareness and builds trust
- Contrasts code (verifiable in minutes) with marketing (takes 90 days) to explain why the same mandate fails in different contexts
- Uses action-oriented names for solutions ('Beat Claude hiring challenge', 'Friday AI agent') that are memorable and copyable
The post ends with a result statement: 'A-players went crazy building things I never asked for.' This is a social proof close that shows the solution worked, and it's cut off mid-sentence to drive comments.
When sharing a failed experiment, diagnose why it failed before offering solutions. This builds credibility and makes the solutions feel earned rather than prescriptive.
What drives their engagement
Siu deliberately cuts posts off mid-sentence or ends with incomplete thoughts ('Do this for every...', 'Our SEO team built a competitive...'), which forces readers to comment asking for the rest. He also uses contrarian claims that invite debate (e.g., 'Mandate the measurement, not the tool'), which triggers discussion.
Posts with specific revenue numbers and case studies (Clavicular's $1.1M, Kick Creator Incentive Program) get reposted because they feel like insider knowledge. Predictions about future trends ('The next generation of media buyers might be buying enough output that the algorithm eventually gives in') also drive shares because readers want to signal they're ahead of the curve.
High. Posts with frameworks (L0-L3 leveling, three-layer AI adoption model) and actionable principles ('Mandate the measurement, not the tool') are saved because readers want to reference them later. The 20.2% conversation index suggests comments are driving engagement, but the 3.7% repost rate indicates that saves (not shares) are the secondary action.
Content pillars
AI Adoption & Fluency
35%How to build organizational AI competency from founder mandate through team incentives, with specific leveling frameworks and measurement systems.
Marketing Operating Models
30%Shifts in how paid media, distribution, and creator economics work, using real revenue numbers and case studies to show emerging patterns.
Founder Leadership & Execution
25%Tactical lessons from running Single Grain, including hiring, team incentives, and how to enforce behavior change without top-down mandates.
Industry Trends & Signals
10%Pattern recognition posts that flag what's changing in tech, platforms, and business models before it becomes obvious.
Hook patterns
Inversion Hook
[Old definition]. Now it can mean [counterintuitive new definition].
“Paid media used to mean buying ads. Now it can mean paying 1,500 editors to flood the internet with clips.”
→ Immediately signals that the reader's mental model is outdated, creating cognitive tension that demands resolution.
Question + Credential
Want to [outcome]? Save this POV by Eric Siu.
“Want to drive real AI fluency? Save this POV by Eric Siu.”
→ Opens with reader benefit, then anchors credibility before diving into the idea, reducing skepticism.
Contrarian Experiment
[Big company] did [thing]. We tried the same thing at [company]. It almost backfired.
“Meta is firing engineers who don't use AI. We tried the same thing at Single Grain. It almost backfired.”
→ Signals that the post will contain a lesson learned from failure, which drives comments because readers want to know what went wrong.
Specific Number + Shift Signal
[Creator] made [revenue] in [timeframe], with [percentage] from [source]. The money is interesting, but the operating model is the real story.
“Clavicular reportedly made $1.1M in January, with over 90% coming from the Kick Creator Incentive Program... The money is interesting, but the operating model is the real story.”
→ Concrete numbers establish authority, then the pivot to 'operating model' signals that the post teaches systems thinking, not just gossip.
Mandate Reframe
Mandate the [outcome], not the [tactic].
“Mandate the measurement, not the tool.”
→ Offers a memorable, actionable principle that readers can apply immediately, driving saves and shares.
Top-performing hooks
“Want to drive real AI fluency?”
“Paid media used to mean buying ads.”
“Meta is firing engineers who don't use AI. We tried the same thing at Single Grain. It almost backfired.”
“AI can make tiny teams look huge, but trust is the part that decides whether the leverage lasts.”
“7 marketing products Neil Patel and I use to get ahead:”
“We are hiring:”
“I built a 5-agent AI marketing team that's better than most marketers I've hired.”
“Anthropic has ONE growth marketer.”
“I think GitHub is one of the most underpriced marketing channels right now. I open sourced all my AI skills yesterday and already have 250 stars. You can see what people are using, what they want more of, and in the future you can capture emails too.”
“I got on a discovery call with an AI and honestly forgot it wasn't a real person. It handled a full 20-25 minute conversation end to end, connected me with recruits, pulled their LinkedIn profiles — the whole thing. If voice AI is already this good, ”
What works
- Inversion hooks redefine familiar concepts in unexpected ways. 'Paid media used to mean buying ads. Now it can mean paying 1,500 editors' got 32 engagement by making readers reconsider their mental model of paid media.
- Incomplete sentences at post end drive comments. Posts that cut off mid-thought ('Do this for every...') force readers to ask for the rest, boosting conversation index to 20.2%.
- Specific revenue numbers and case studies build credibility. Clavicular's $1.1M in January and $20K per day figures made the operating model shift feel real and provable.
- Frameworks with levels (L0-L3) make abstract concepts implementable. The AI fluency leveling system gave readers a template they could copy into their own org immediately.
- Diagnosis before solutions builds trust. Explaining why the Meta mandate failed at Single Grain ('Code is verifiable... Marketing output takes 90 days') made the three solutions feel earned rather than prescriptive.
- Predictions about future trends drive shares. Ending with 'The next generation of media buyers might be buying enough output that the algorithm eventually gives in' signals insider knowledge and makes readers want to share.
Steal this for your own LinkedIn
- Start your next post with an inversion hook: '[Old definition]. Now it can mean [counterintuitive new definition].' This pattern-interrupts the scroll and creates cognitive tension that demands resolution.
- End your post with an incomplete sentence or thought that forces readers to comment asking for clarification. This is a deliberate engagement hack that boosts conversation rate.
- Include a real case study with specific numbers (revenue, timeframe, percentage breakdown) before generalizing to a broader pattern. This makes abstract claims feel grounded and credible.
- When sharing a lesson from failure, diagnose why it failed before offering solutions. Use contrasts ('Code is verifiable in minutes... Marketing output takes 90 days') to explain why the same approach fails in different contexts.
- Create a framework with numbered levels (L0, L1, L2, L3) that readers can copy into their own org. This makes your idea immediately actionable and shareable.
- Post 7+ times per week on a consistent schedule, especially on Friday mornings. This keeps you visible and signals that you're actively building and learning.
Compare Eric with…
vs Christ Coolen
Christ Coolen vs Eric Siu: Psychology teacher or operator coach?
vs Vin Matano 🐝
Eric Siu vs Vin Matano: Pattern recognition beats vivid storytelling on LinkedIn
vs Mark Roberge
Eric Siu vs Mark Roberge: Reach vs Resonance on LinkedIn
vs Daniel Priestley
Daniel Priestley vs Eric Siu: Authority through threat vs. authority through pattern
vs Nathan Barry
Eric Siu vs Nathan Barry: Pattern recognition beats storytelling on LinkedIn
vs Arvid Kahl
Arvid Kahl vs Eric Siu: Vulnerability vs Pattern Recognition
Get the full Eric Siu ✓ playbook
Hook templates, content calendar, posting cadence, and the exact frameworks they use - delivered to your inbox.
Posting patterns
Siu posts 7.7 times per week with 100 posts in the last 30 days, showing extreme consistency. This cadence keeps him visible in followers' feeds and signals that he's actively building and learning.
Friday at 10 AM UTC performs best. This timing likely catches readers during end-of-week planning or Friday morning coffee scrolls when they're thinking about the week ahead.
Short posts (30-80 words) make up 47% of his output and average 8 engagement, while long posts (180-350 words) are 28% of output and average 11 engagement. The data suggests that medium-to-long posts (180-350 words) perform best, but Siu leans heavily on short posts for volume and consistency. His very long posts (350+ words) average 117 engagement, suggesting that when he goes deep, it's worth it.
