Recently one of my notes went viral (1.6K likes and 276 comments). And many people who congratulated me on my Substack anniversary shared their pain:
“11 months with 41 posts but only 73 subscribers. I publish notes almost every day. They just don’t work.”
Or: “I publish notes every day. But nobody sees them.”
The frustration was real. I went and looked at their accounts — genuinely good writing, the kind that deserves to be read. Just sitting there in the silence.
I wanted to help. And since I spent years in academia doing social research, I did what comes naturally to me — I treated it like a problem worth investigating properly.
With AI, anyone can do this — no special training needed.
In this article I’ll walk you through the analysis of notes from one of my subscribers — let’s call him Dan.
Or skip straight to the blueprint and use the ready-made prompts on your own notes.
It takes 20–30 minutes. You won’t even need to copy your notes manually — Claude for Chrome can collect everything on its own. (!)
You’ll understand why your notes aren’t getting likes, how to fix it, and Claude will build you a growth strategy based on the results. More likes. New subscribers.
Who This Article Is For
Substack writers:
who publish notes regularly but get 0–2-5 likes
who want a concrete, repeatable way to use AI to improve their Notes performance
Dan’s Case: Why Notes Aren’t Working
Dan writes about India-Pakistan Kashmir geopolitics — a narrow niche, but one with real global interest. Academics, policy professionals, diaspora members, journalists. Thousands of people follow these developments daily.
After 11 months of consistent posting, he had 73 subscribers. Notes averaged 0–1 likes. No comments. No restacks. No conversations.
I’d watched other writers in equally niche topics build audiences of thousands.
The topic wasn’t the problem. The approach was. We often can’t see it ourselves — too close to the content. That’s where outside perspective and Claude’s analysis help.
3-Step Notes Analysis with Claude
Step 1: Gathering the Data
If you’re on Claude’s free plan, collect your notes and metrics manually into one document. Or try the Comet Perplexity web scanner.
If you’re on any paid plan, use Claude for Chrome (you’ll need to connect it) to pull your notes and metrics (text, number of likes, comments) into a table.
Use this prompt:
“Can you extract my last 10 notes from my Substack page with their metrics (text, likes, and number of comments) and put them in a table. Don’t include restacks.”
Pro tip:
If you have more than 10 notes, split the requests. Ask for 10 at a time , Claude handles it easily in 3–5 minutes. Then merge everything into one table.
When I asked for 30 notes at once, it worked for an hour and quit several times.
Pro tip:
I’m often asked how many notes are needed to run an analysis. 10 notes is enough to identify the main problem.
Save this table in a dedicated Substack folder. You’ll use it again.
Step 2: The First Prompt That Changed Everything
Once your table is ready, ask Claude to analyze it and explain why your notes aren’t getting traction.
Pro tip:
You can do this in the same chat. Or — better — move the table into a Claude Project. Let’s call it “Substack Growth Strategy.” You’ll find it easily later, and Claude will build up your full Substack data over time, making future analysis sharper.
For Dan’s notes I used:
“Analyze the file ‘Data on Dan’s Substack notes’ and explain why he gained only 73 subscribers in 11 months. What are his main mistakes in promoting his account?”
Claude identified four patterns:
No engagement or community building — He informed but didn’t connect. No questions, no CTAs, no conversation starters.
Weak content format — Mostly links with minimal analysis. Reads like a news aggregator, not a person.
Narrow niche without depth — Not going deep enough to become the go-to voice. Complex topics with no entry point for new readers.
No promotion strategy — No collaborations, cross-promotion, or networking.
What I love about Claude is that it doesn’t just name the problems — it gives practical suggestions.
What Claude recommended:
Ask questions, respond to comments, build community
Add original analysis — don’t just share links
Share your opinion, stories, behind-the-scenes perspective
Create pillar content that establishes authority
Network with writers in foreign policy and South Asia niches
Offer exclusive content to incentivize subscriptions
Write stronger headlines and opening hooks
He was treating Substack like a bookmark service instead of a relationship-building platform.
Step 3: Comparing With Viral Notes
The first analysis was useful, but I wanted more specific guidance. So I asked Claude to compare Dan’s Notes with high-performing ones:
“Analyze the types of notes he posted and compare them with the highest-performing Substack notes from other writers that typically attract lots of likes and new subscribers.”
I didn’t upload anyone else’s notes. Claude already knows which types perform best on Substack.
The Brutal Truth: What Was Really Missing
Problem #1: Broadcasting, Not Connecting
Dan’s typical Note: “Shocking! An estimated 90,000 people gathered at a stadium in X’s provincial centre to watch a public execution. [link]”
Claude concludes: Engagement: 1 like. Leads with ‘Shocking!’ then goes clinical. Missing: human story, personal reaction, emotional processing.
Claude’s rewrite:
“90,000 people cheered at a public execution yesterday. I’ve reported from that province. I know people who were there. What haunts me isn’t the execution — it’s the normalization. Three years ago this would have sparked international outrage. Today? Barely a headline. When did we stop being shocked?”
One is a news alert. The other is a human grappling with something painful. People connect with humans, not headlines.
My take:
Although I would certainly change Claude’s ‘language,’ which clearly reveals an AI style. We need to keep our own voice. But the insight is right: people don’t need more raw information. There’s already too much. They want your story, your emotion, your take, and a sense of connection.
Problem #2: Never Asking Readers Anything
Claude noted: “Notes are one-way broadcasts. No questions, no conversation starters, no calls to action.”
Meanwhile, successful writers post things like:
“Can we normalize starting a business at 45?” “Anyone else feel like something’s shifting?”
Or they invite other writers to drop their Substack links. Or they offer encouragement: “We can’t protect ourselves from failure. Write the post, even if it flops.”
Why does it work? They create space for others, start real conversations, build community. That’s what draws people in.
My take:
I completely agree with Claude.
Problem #3: No Personal Voice
This was the hardest feedback for Dan.
He shared what was happening. He never said what he thought about it. No opinions, no feelings, no sense of why it mattered — to him or to his readers.
Claude explained that successful Notes make readers feel:
“That’s exactly how I feel!”
“I thought I was the only one.”
“This person gets me.”
Dan’s Notes were informative. They weren’t relatable.
Dan’s original: “X seems to have now escalated the rhetoric against Y...”
Claude’s rewrite: “I’ve covered X-Y relations for years. Yesterday’s rhetoric crossed a line I’ve never seen. X’s foreign ministry just called them ‘HHHHH’ — the first time they’ve dropped ‘government’ from their language. This isn’t semantics. This is preparation for something bigger. What do you think happens next? 👇“
Same information. Completely different impact.
My take:
I would have said the same thing.
One Thing Claude Missed
Overall, the analysis Claude did looks solid to me. But it still missed one important point.
Any successful Substack writer uses different types of notes — educational, motivational, community-building — not only the ones that are directly tied to their niche.
Remember: Notes on Substack are a conversation, not a broadcast. The goal of Notes is not just to inform people, but to build relationships with them.
Want a Deeper Analysis?
The analysis will be deeper and more useful if, for comparison, you include text of notes from other Substack writers whose posts get lots of likes and comments.
You can collect these notes manually into a file.
Or, if you want to save time, you can use my swipe file with 27 high-performing notes. Some of them received more than 1.6K likes and over 200 comments. If you want this file, send me a message or leave a comment.
Blueprint: Substack Notes Audit
What This Blueprint Does
Takes you from “0–2 likes per note” to understanding exactly what’s broken and how to fix it — using Claude as your analyst. Takes 20–30 minutes.
Step 1: Collect Your Data
If you’re on Claude’s free plan: Manually copy your last 10 notes + their metrics (likes, comments) into one document. Alternative: try Comet Perplexity web scanner.
If you’re on any Claude paid plan: Use Claude for Chrome to collect automatically. Use this prompt:
“Can you extract my last 10 notes from my Substack page with their metrics — text, likes, and number of comments — and put them in a table. Don’t include restacks.”
Important:
Do 10 notes per request — not all at once. Claude struggles with 30+ in one go.
If you have 30 notes: run 3 separate requests of 10, then ask Claude to merge them.
10 notes is enough to spot the main problem.
Save the table in a dedicated Substack folder for future use.
Step 2: Run the First Analysis
Move your table into a Claude Project (name it something like “Substack Growth Strategy”) — or use the same chat.
Use this prompt:
“Analyze this table of my Substack notes and explain why they aren’t getting likes or new subscribers. What are my main mistakes?”
What Claude will look for:
Are you engaging readers or just informing them?
Are you adding original analysis or just sharing links?
Do you have a personal voice or does it read like a news feed?
Is there any promotion or networking strategy?
Save Claude’s output. This is your baseline diagnosis.
Step 3: Compare With High-Performing Notes
Use this prompt in the same chat:
“Analyze the types of notes I posted and compare them with the highest-performing Substack notes from other writers that typically attract lots of likes and new subscribers.”
You don’t need to upload other writers’ notes — Claude already knows which formats perform best.
Optional (for deeper analysis): Manually collect 10–15 notes from writers in your niche who consistently get strong engagement and add them to your file before running this prompt. The comparison will be sharper.
Step 4: Request Your Growth Strategy
Once the diagnosis is clear, use this prompt:
“Based on this analysis, create a practical growth strategy for my Substack notes. What types of notes should I be posting? How often? What should I change starting this week?”
Now Your Turn — Try It Right Now
You don’t need to read the full blueprint to get started. You can run a mini version of this experiment in the next 20 minutes with what you already have.
Take your last 5-10 notes. Copy them into a document with their likes and comments. Then paste this prompt into Claude:
“Analyze this table of my Substack notes and explain why they aren’t getting likes or new subscribers. What are my main mistakes?”
That’s it. One prompt. You’ll already see something useful come back.
Then come back here and share with us in the comments:
Did the feedback feel accurate or did it miss something important? And the bigger question: do you trust AI advice about your writing? Does it feel helpful or does something about it bother you?
I’m genuinely curious. Because I think we’re all figuring out together where AI is actually useful for writers — and where it falls short. There’s no right answer yet. But your experience is part of building one.
If this post was useful and you feel like saying thanks, a small coffee donation goes a long way. It truly fuels my drive to keep exploring and writing. Every bit of support means the world! ☕️
Thank you Alesia for this generous share. Your giving is certainly part of your success. 🙏😊❤️
this is great - thank you for sharing the step by step audit!