Write Viral LinkedIn Articles with Gemini

If you look at my LinkedIn article publishing history, you might notice a significant gap – roughly ninety days where activity went silent. This wasn’t intentional neglect; it was a consequence of success in other areas.

My client workload became so demanding that dedicating time to content creation became a luxury I couldn’t afford.

Writing a high-quality article, particularly the detailed step-by-step guides I often produce, is a considerable time investment.

A single piece can easily consume over four hours, factoring in topic research, outlining, creating header images, capturing and annotating screenshots, and linking to relevant previous content.

While I genuinely enjoy the process of sharing knowledge, the time commitment became unsustainable.

It was clear I needed a more efficient method to maintain consistency without sacrificing quality or my client work.

This case study breaks down the exact process I developed, leveraging AI, specifically Gemini Advanced, to go from a topic idea to a published LinkedIn article in approximately 30 minutes.

I will also outline every tool involved in this streamlined workflow.

Does This Method Actually Work?

Theory and potential are interesting, but results are what matter in business. Before I go into detail, let me share some results. Before you listen to my advice, you need to ask an important question.

Does using Gemini to write LinkedIn articles work?

In the past years, I’ve written over one hundred LinkedIn articles manually. Those articles received 62,223 views. While this is respectable, I’m going to triple those numbers with less work over the next 365 days using Gemini.

I put in a lot of time and energy into every article and when I ran out of time, the results collapsed. You can clearly see the occasional spike when I posted an article between October and March. That’s a lot of lost opportunity.

Other than those spikes my results basically plummeted. You’re only as good as your most recent post and I suffered for my inconsistency.

Let’s zoom in on my views over the past four weeks. I think you can tell the exact day I started using Gemini.

Article views have surged to over 7,000, with the daily average consistently exceeding 500 views. This is not a spike as the daily numbers are now very solid. When you compare to my numbers over the past year, I’m still beating the best articles that I ever wrote myself.

I we annualize my daily 500 views by multiplying by 365 days in a year, I have an annual viewership potential exceeding 180,000 views. This assumes that I don’t increase from my consistency and that I just maintain the same.

I’m getting the best results I’ve ever received and I’ll easily beat last years numbers.

The effectiveness of this method stems from three key factors:

1. Consistency: AI overcomes the primary bottleneck of human time constraints, enabling daily publishing. LinkedIn’s algorithm rewards regular content.
2. Quality at Scale: This process utilizes AI as an advanced assistant to generate well-structured, relevant drafts rapidly. I’m able to crank out high-quality articles faster.
3. Leverage and Efficiency: I’m still very busy. But now I’m able to stay a week ahead with my LinkedIn articles so I never have to worry about finding time to publish and article today. I have breathing room.

The data shows that my new approach works.

The Step-by-Step Process

Let me break my process down into a series of steps that you can replicate and modify to create content for whatever platform is most important for your business.

I’ll try to keep this as brief and easy to replicate as possible. IF you get stuck on a step, please just leave a comment below and I’ll update this article.

Step One: Managing Your Project

I use a very simple, but extremely powerful project manager called Blue to organize everything. You can use any project management options that you prefer, even a Google sheet can work.

I prefer to use a Kanban-style board where each article a single card. As the article progresses, teh card moves along the board from left to right.

The columns on the board mirror the stages of my content creation process, so I always know how many articles I have at each stage. The screenshot above shows that I have multiple articles at every stage in my pipeline.

The columns are structured from left to right:

• Ideas: Articles that I think might be interesting but need more research.
• Research: Articles that I am currently researching.
• Drafting: Articles that are currently being written by me or Gemini.
• Scheduled: Completed articles scheduled for publication on LinkedIn.
• Posted: Articles that have been published.

I could add additional columns for steps like “Header Image Creation,” I want to keep my process as simple as possible. It’s very easy to overcomplicate the columns and then you can’t fit them into a single screenshot.

That leads to diminishing returns.

I want to be able to see and understand all my articles in a single second.

The most important part of this process is keeping a lot of cards in every single column. This ensures that I don’t need to come up with new ideas when I don’t have any ideas. Or that I need to do research when I don’t feel like doing research. I do all of my work in batches and this is the most effective method for my creation process.

I have quite a few sources for coming up with ideas. I never want to rely on just one method as sometimes there is nothing interesting in the news or everything is a topic I’ve already covered.

Here are the three main places that I find inspiration.

• Feedly: This is an AI-powered news aggregator. I feed it all the blogs, news sites and places that have information I want to know about. It then organizes everything using an internal AI that knows which articles I find the most useful.
• YouTube: I follow a bunch of great YouTube channels by other creators with different perspectives and ways of creating content. Some of the channels are heavy on the hype and others are super technical. These are great areas for my to find useful AI information that I can simplify.
• Social Media: I also pay attention to what people I follow post on Twitter and LinkedIn. If I can offer a unique perspective then it goes on to one of my cards. I find both platforms very difficult to search, so this is where I get the fewest ideas.
• Direct AI Brainstorming: The easiest source for evergreen articles is Gemini itself. I’ll simply ask it to come up with ten ideas based on all of my past articles and choose the ideas that I like the most. This is the easiest way to keep my “drafting” column full in Blue.

I always want a lot of cards in the idea phase so that every column after that also stays full.

Step Three: Research

Not every article requires extensive research. For many of my articles, I already know enough to put together a great article. When I do need research, here are my methods.

Deep Dives & Process Design:

For step-by-step guides, the research phase involves testing products, comparing methods, and designing a superior process. My article on How to Install Flux in Under 10 Minutes is my most popular article of the past years.

I spent over eight hours working on that article and designing my process. The method I was refining took over two hours. I turned a two-hour YouTube video into a ten-minute process for that article.

I recently tested a new method that was supposed to be faster…but it failed. So that is still the fastest and easiest method all these months later.

Research with Perplexity

Not every article requires me to research and test every step. Often I just need some statistics or a deeper understanding of a process. Perplexity is my favorite research AI and it always provides links to all of its sources.

It’s an amazing tool that you can master in less than ten minutes. I have some new LinkedIn articles coming out that cover all of the amazing new things Perplexity is doing, including releasing their new Comet browser.

YouTube Transcripts

One of the coolest things you can do with YouTube now is feed the transcript to an AI. I have a few different ways to do this that are all free.

You can paste the link to the video in Gemini and it will pull in the transcript. You can use Haarpa or Merlin, which are both great Chrome extensions.

You can even automatically transcribe every video from channels you follow using Taskade, if you want to automate this process.

Most of the time, I will generate the transcript using a plugin and then copy and paste the section that I want to use into Gemini.

NotebookLM

Google’s NotebookLM is my favorite AI tool on the market. It’s a tool that rarely gets discussed but is insanely valuable. When I want to learn and master a new topic, this is my go-to tool. I can paste in links to any articles and YouTube videos that I want to use as sources. I can also drop in PDFs and audio files.

Once I’ve feed everything to NotebookLM, I have it teach me using podcasts, articles, study guides, and quizzes. However you like to learn, NotebookLM will meet you there.

Step Four: Train Gemini to Write Like You

Once the research is done, I want to get the article written as fast as possible.

There are several ways to train Gemini in a matter of minutes and I’ll walk you through each of them right here.

Option 1: Gemini Knows Who You Are

Here is the exact prompt that I used to train Gemini. I also pasted in three previous LinkedIn articles that best represent how I want my new articles to look.

Option 2: Gemini Doesn’t Know Who You Are

You can easily train Gemini by uploading enough articles for it to get a sense of your style.

This is a simple prompt that will have Gemini tells you when it has mastered your personal writing style.

Future Proof Training

The third method is to provide Gemini with an index of all your past articles. For a blog like this, that is the Sitemap.

Unfortunately, LinkedIn doesn’t provide a sitemap of all my past articles and the links are blocked to AI tools, so I have to break this step into two parts.

First I create a spreadsheet with the titles and links of all my past articles.

This spreadsheet is a useful tool that I can use to train different AI models and I’ve used extensively with training Taskade in the past.

This is useful for creating links to other articles as well, so I do use this spreadsheet a lot.

I have also created a single Google document that has every LinkedIn article that I’ve ever posted on a new tab. This is a massive library of training data and far more than I need to train Gemini right now, but I will probably want to use it with a future update.

Step Five: Generating an Article Fast

This is the easiest step in the entire process and require the least work as long as you have trained Gemini correctly. I don’t need to write a style guide or teach Gemini about my signature and call to action. It has gleaned everything I know about writing and created an internal style guide.

My prompt to write an article is extremely short because I’ve already done all of the hard work. Now it is time for Gemini to shine.

If there is a lot of specific data that I want included in the article, I’ll use voice to text in my phone to provide the article prompt. Otherwise, I can usually paste the transcript of a video that I want to use or simply give Gemini the topic and let it perform research and write the article.

The more time you spend on step four, the less time you have to spend on step five.

Step Six: A Manual Pass

You are going to be tempted to skip this step…and that’s why it’s the most important step in this entire process.

If you skip this step, everything else you have done will be wasted. You want to catch those little mistakes that can embarrass you and damage your reputation., before you publish the article.

Gemini makes mistakes.

Sometimes it gets a fact wrong, it always formats the articles wrong and sometimes the links to other articles are broken.

Occasionally, Gemini will reference either me or the video transcript in the third person and that needs to be fixed before publishing.

Gemini always formats my articles with long paragraphs, even though I prefer a break between each sentence. I find that much better for LinkedIn formatting.

I usually change 5-10% of the content and this is the step that takes the longest. But never more than fifteen minutes. Usually, it’s closer to five minutes.

For this step, I copy and paste the article from Gemini into a new Google doc. Then I fix all the formatting, add in the correct headers, check that the links work, and read through the article.

I ask myself one question before publishing: “Does this look like I wrote it?”

You want to avoid publishing anything that is “good enough.” That phrase is a red flag that means your standards have dropped.

I had Gemini write the first draft of this article and it was way too casual. Then it went way to formal. I manually edited and rewrote nearly every sentence in this blog post, because I’m not interested in “good enough.

Everything with my name on it needs to look like I wrote it.

Better Living Through AI

It’s really easy to crank out trash articles and blog posts with AI.

If that’s your goal, then this blog post was a big waste of your time.

My method creates content that I’m proud of, that still sounds like me and that continues to represent my brand accurately.

Once the article is written, I have to schedule it using LinkedIn’s scheduling tool and add a header image.

When you cut corners on training Gemini, the results aren’t very good. That’s why this blog post needed so much rewriting. Gemini didn’t drop the ball, I did.

I need to write a few new blog posts that I will then use to train a better model and get higher accuracy. I haven’t written a blog post in so long that I didn’t have good examples to train Gemini.

I will use the exact same process I just taught you to create a better blog post writer and that will allow me to speed up my content creation even more.

Using Gemini I went from a post every few months to a post every single day. I’m always a week ahead with my content and have a very good pipeline.

I designed this method after trying to hire someone else to write my posts and they asked for way too much money. I got so mad that I said “I’ll just get an AI to do it.”

And here we are.