How ChatGPT Revealed the Secret Formula Behind My Top Substack Posts

ChatGPT Helped Me Audit My Substack—Here’s What I Learned

This is a guest post by

Landon Poburan

---

I’ve always been curious about what makes a winning Substack post.

Some posts hit. Others… flop.

2024 was a year of testing and discovery with my content.

I wrote 100+ posts on Substack and experimented with various topics.

I’ve written 3,000-word tutorials and heartfelt descriptions of failed businesses. I’ve written about Facebook ads, content strategy, Substack growth, and running webinars. I’ve shared about my business, myself, and my relationships.

Some got 0 likes while others got 400+ and generated 100s new subscribers.

But why?

Was it the topic? The timing? The way I wrote it?

So, I decided to get serious.

It was time to dive deep and figure out what works—and what doesn’t.

With the help of AI, I audited every single post I’ve written. And let me tell you—it changed everything. We broke down the winners, the losers, and everything in between.

The result?

Surprising insights about what my audience loves (and what they skip).

In this post, I’m sharing it all:

• How I audited my Substack with ChatGPT.

• What I learned from my top and worst posts.

• How I’m using this info to write better, smarter, and more engaging content.

This isn’t about guessing anymore.

It’s about writing with purpose—and growing with confidence.

Let’s get started.

Step 1: Collecting the Data

To start, I needed to get all my posts in one place.

Every title, topic, and metric—organized and ready to analyze.

Here’s What I Did:

Reviewed My Substack Posts.

I added all of my posts into a spreadsheet (CSV file). This included the basics: titles, URL, publish dates, and full content.

Added Key Metrics.

I included important data points:

• Views

• Open rates

• New subscriber counts

• Recipients

• Likes

• Comments

• Restacks

These numbers prepared the file so that it could be analyzed effectively by AI to see what was working and what wasn’t.

It took me 1-2 hours to pull all my posts.

There’s currently no way to automate this (without using potentially TOS-violating methods) but a couple of hours to build a content strategy is well worth it.

To speed the process up, you can do this more frequently, track each post as it is published, or review a smaller batch of content if you publish at a higher frequency.

Once the data was ready, I handed it off to ChatGPT.

And here’s what I asked it to do:

I’ve exported all my Substack posts into a spreadsheet. Analyze the data to identify trends, such as my most engaging topics, formats, and post types. Provide a summary of my top 20 performing posts with actionable insights.

Or you could ask:

Here’s a CSV file of all my Substack posts with their titles, topics, and performance metrics.  
Analyze this data to identify trends.  
Which posts performed the best and why?  
Which posts performed the worst?  
What patterns can you find in the topics, formats, or other details?  

ChatGPT didn’t just give me a summary.

It broke things down in ways I hadn’t thought of. It showed me patterns, gaps, and most importantly, opportunities to grow.

Step 2: Analyzing Top and Worst Performers

With my data and ChatGPT’s help, the next step was figuring out why some posts worked better than others—to inform and improve my content strategy.

First, I looked at my top 20 posts.

Here’s what I found:

• High Relevance to Audience Goals: My audience is deeply interested in building their Substack and online business. These posts align directly with their desire to grow, monetize, or strategize effectively.

• Educational Value: Posts combined entertainment with education, delivering practical advice alongside engaging storytelling performed well.

• Broad Appeal with Specific Focus: Topics like monetization appeal broadly, while specific angles (e.g., why not to paywall content) create a unique value proposition.

• Trust-Building Through Transparency: Sharing personal successes, challenges, and lessons learned fosters trust and builds authority, making readers more likely to engage.

• Community Engagement: Topics address questions or challenges common within your audience, making them highly engaging and discussion-worthy.

Try this prompt:

Based on the top 20 posts, please identify the following:

- Trends based on content and topic.
- Why were these the best posts.
- What about these posts made them the winners.
- What questions or topics were left unanswered?

Then, I flipped the script.

I asked ChatGPT to analyze my 5 worst-performing posts.

Here’s what didn’t work:

• Lack of Audience Alignment: These topics did not align strongly with my audience’s core interests.

• Repetition or Predictability: For example Facebook ad tips are helpful, but they may lack relevance, novelty, or specifics that stand out in a saturated topic area.

• Generalized Topics: Many of these posts address broad themes like Facebook ad performance or general tips for client success. Audiences might be looking for more niche insights or personalized takes that feel directly applicable to their challenges.

Lastly, I asked ChatGPT to compare them directly.

Compare my top 20 and worst 5 posts.What patterns or differences stand out in tone, format, or topics?

The insights were simple but powerful:

• Infuse personal stories.

• Align topics with audience aspirations.

• Add engagement triggers.

• Write great headlines that evoke curiosity.

The posts that worked were personal, aspirational, and action-oriented.

The ones that didn’t? Impersonal, misaligned, and poor structure.

I wasn’t just seeing what worked—I was learning what my readers wanted most.

Next, I used these lessons to build a new content strategy.

Step 3: Generating New Content Ideas

With all the insights from my audit, it was time to brainstorm. But instead of starting from scratch, I let ChatGPT take the lead.

How I Did It.

I asked ChatGPT to generate ideas based on my top-performing posts.

Try this:

Using the themes and formats from my top-performing posts, generate 10 Substack post ideas.Focus on actionable, relatable, and engaging topics.

Or this:

Based on the top 20 posts, please provide me a list of 15 article ideas. These should be based on your assessment provided. Following the same trends as to what has worked.

I would like suggestions as to how to make these stand out, captivate the audience.

With each topic, provide me a headline -- using my top post headlines as a template or framework to write new ones.

What Came Back.

The suggestions are the foundation of my upcoming posts.

Here are a few examples:

1. 5 Substack Creators Making $10K+ Per Month — And How You Can Too

2. My 5-Step Workflow to Never Run Out of Substack Content Ideas

3. I Went Viral (Again) on Substack — And Here’s Why It Won’t Guarantee Success

4. Don’t Rely on Paywalls: Creative Monetization Hacks for Substack Creators

5. Why Giving Away Free Content on Substack is Your Best Monetization Strategy

ChatGPT even suggested new formats, like:

• Turning a popular topic into a multi-part series.

• Sharing personal lessons in Q&A format.

• Repackaging older posts into compilation lists.

• Featuring the personal stories of others.

This isn’t random brainstorming.

It’s targeted, and rooted in data from my best content.

The ideas felt fresh but also aligned with what my audience already loves.

Results and Reflection

After a year of writing over 100 posts, finding my voice, and testing dozens of ideas, I finally have the data to see what works—and what doesn’t.

It’s not just guesswork anymore.

I know the topics my audience loves, the formats they engage with, and the tone that keeps them coming back.

This exercise gave me more than just ideas—it gave me confidence and clarity on what to do this year.

I’m no longer wasting time on posts that miss the mark, hopefully ;-) Instead, I can focus on creating content that connects, resonates, and drives growth.

But, this isn’t a one-and-done process.
Content evolves. Audiences change.

That’s why I’ll be doing this audit at least once a year—maybe even quarterly.

It’s a simple way to make incremental improvements and stay aligned with what my readers want.

Here’s the thing…

You don’t need 100 posts to start.

If you have even 20 or 30 posts, this exercise can give you real insights into what’s working.

It’s an hour or two of work that can completely shift how you approach your content.

And with tools like ChatGPT, it’s easier than ever.

From analyzing patterns to brainstorming new ideas, AI makes this process faster and more actionable.

Your content isn’t just a collection of posts.
It’s a blueprint for what your audience values.

When you take the time to step back, look at the data, and refine your strategy, you’re setting yourself up for consistent growth—without burning out.

Whether you’re just starting or deep into your Substack journey, this kind of audit is worth it.

And when you’re armed with insights and a plan, that progress becomes unstoppable.

---

This was a guest post by

Landon Poburan

.

For more insights like these, make sure to subscribe to his Substack.