How to use chain prompting to write a blog post

Create a SEO optimized blog post with ChatGPT

Hey!

I hope your week started as productive as mine!

Been working on some automation headaches over the past 2 days and finally found out that one small mistake was responsible.

Complex automations can be tricky…especially if there are multiple AI steps included…however - it’s solved.

Let’s get to business.

In one of my recent surveys 36% answered that they mostly use ChatGPT for content creation.

The second place shortly after with 31% is using ChatGPT to write messages and emails.

(We’ll cover how to do automated AI generated email responses in one of my next How-To’s.)

So according to the survey…today it’s all about how to use AI for content creation. Especially long form content like blog articles.

ChatGPT for Content Creation

Doing research and writing on your own can take hours and hours.

ChatGPT can be a huge time saver while transforming content into blog posts.

But to get a quality blog post that is SEO optimized does not work too well with only one prompt.

I’ve tried multiple ways and using prompt chains gives the best results.

It’s a step-by-step conversation where each new step uses the output of the previous step.

This ensures ChatGPT gets the context and makes it easier for you to modify and control the results to your needs on-the-fly.

Let’s take a look at this by using an example.

We have an article we want to transform into a blog post. (Also works with multiple articles you want to combine, but in this example prompts we’ll do it with only one.)

1️⃣ First, we summarize the article.

„Summarize the following article about [topic] into a text and use maximum 400 words. [Paste Article]“

2️⃣ In the next step, we extract the key information.

„Extract the key information of this article about [topic]. Create a bullet point list with all key information.“

3️⃣ We want our blog post to be optimized for SEO, so we ask ChatGPT to extract relevant keywords from the article and categorize them.

„Identify the most relevant SEO keywords from this article about [topic] and categorize them into two bullet point lists. 1. Fat-Head Keywords and 2. Long-Tail Keywords.“

4️⃣ Now, we start with our actual blog post. We ask ChatGPT to create a title that gives a clear understanding of the content.

"Write a compelling title for a blog post about [topic] that is about [key information]. Use at least 1 Fat-Head Keyword from this list: [Keyword List]. Limit the length to 60 characters."

5️⃣ We create SEO-optimized sub-headlines for every section of the blog post and a brief description of what it intends to cover in each section.

„Draft [number] article sub-headlines for a blog article with the title [title], using this summary [summary], these key information [key information] and these SEO keywords [Keyword Lists]. Include a short description what you intend to cover in each section.“

6️⃣ In the next step, we instruct ChatGPT to write SEO-optimized blog sections based on our information and add a tonality.

„Write an insightful blog section for each of the headlines: [headlines]. Apply the tone of [tone]. Use these SEO keywords: [all keywords]. Use this summary and key information: [summary and key information]. Do not write more than 300 words per section.“

7️⃣ If you're not satisfied with the output, just ask ChatGPT to rephrase.

"Rephrase this section [section you want to change]. Use the tone of [tone] and highlight this key information instead: [key information]."

8️⃣ Once you are happy with the result you can instruct ChatGPT to compose the final blog post. Do not forget to proofread and personalize it. You don't want to miss that personal touch.

"Compose this into a full blog post with the title, sub-headlines and sections we created."

This should give you a well structured and SEO optimized blog article.

Don’t forget to edit and adjust it with some personalization!

Our automated content system

This prompt chain is a part of an automated system I’ve built to streamline content creation for a marketing agency. The system is based in AirTable and collects RSS feeds from different sources.

Then it sends the content from the scraped articles through this prompt chains and sends it back into a new field in AirTable.

From their AirTable dashboard, they can then edit, review and publish the blog post to their Wordpress website.

This saves them almost 9 hours of work per week and their website blog always has fresh content.

I’ll cover this system soon…

Until then, keep creating!

Best,

Nik