How to build an automated blog system for trending topics

July 17, 2024

Jack Marchese

15 minutes

Key takeaways:

  • This blog will teach you how you can use AI to build a content generation machine that creates high-performing blogs at scale without code.
  • Blogging still matters, even as more people turn to places like ChatGPT for their information.
  • ChatGPT and other platforms actually create more opportunity for your business to be discovered, not less.
  • They key to blogging success is making sure content is engaging, viable to rank, and writing about topics that are trending or emerging.
  • You will need accounts for TrendSwell.ai, Google, Zapier (or similar API tool), and OpenAI.

Introduction

With the invention of Large Language Models (LLM's) like ChatGPT and Claude more people are turning to AI to answer their questions instead of search engines.

You might think this means SEO efforts like blogging are becoming more dated than a 2024 US presidential candidate (pre Harris), but I'm here to tell you blogging still matters.

This article will break down why you should be blogging, and how you can create a content generation engine that will generate dozens or hundreds of trending blogs quickly.

Why should I blog anyway?

Great question. The way I see it, there's several reasons why you should keep blogging:

1. Search engines are still king

2. LLM's will index your content

3. Building a brand

Let's talk about each of these a bit more.

Google reigns supreme... for now...

While there's yet to be any kind of academic research behind this, one Reddit poll I saw while researching this topic shows that most people still primarily rely on search engines.

With companies like Apple now planning to integrate with OpenAI, it's only going to become easier and easier for people to obtain information without needing to go to the internet.

However, we aren't at to the point yet where search engines are dying out. In fact, with Google now having their own search engine have an AI assistant, it's likely search engines as the primary way to research topics, ideas, and more aren't going anywhere.

LLM's will use your content

This is probably the most important factor, so I'm burying the lead here a bit with this one. There's a lot of concern that AI will simply rip your content in an answer and give you no credit, but from my experience that's often not the case.

In fact, unlike Google, ChatGPT does not care about SEO metrics or domain authority when finding an answer, making it more likely your content can be discovered within an LLM like ChatGPT even if your blog is relatively small and new.

This matters, because when a user may ask ChatGPT a question about a relatively niche or recent topic you'd be more likely to be one of the websites sourced in a response, directly linking your website for the user to check out.

As an example, my buddy Rock has a technology consulting/implementation firm called Easie. Their main growth engine is (Word-of-Mouth) WOM, referrals, and in-person events, meaning their blog is not a central focus of their growth strategy and rarely rank organically in Google.

However, Easie has extensive expertise in extremely niche areas like domestic lithium production and when asking ChatGPT about that you can see they are directly referenced several with links to the article.

LLM's rely on content that is newer or niche when being asked questions it can't itself answer.

Easie was referenced several times alongside authoritative sites like S&P Global and Energy.gov

It should not be understated just how important it is for both discoverability and credibility for ChatGPT to surface your content alongside larger and trusted sources of information. ChatGPT is essentially positioning your content as authoritative, informational, and trustworthy.

Easie was actually able to gain a new lead when they realized that ChatGPT had surfaced this same article to a user of the platform asking about lithium.

Building a brand

Lastly, we all know the value of building an audience. The term "community" makes people roll their eyes since it was a buzzword in VC circles a few years back, but it can't be denied that having people who care about what you have to say is a powerful thing.

By regularly publishing content that is informative, engaging, and relevant to your audience, you create a platform for sharing ideas and knowledge. This not only positions your business as an authority in your industry but also invites your readers to engage with your content through comments and discussions.

When you address the needs, questions, and interests of your audience, you build trust and loyalty, encouraging them to return to your blog and participate actively. Additionally, by responding to comments and creating a dialogue with your readers, you can develop a sense of belonging and community, making your business more than just a service provider but a hub for like-minded individuals to connect and share their experiences.

This sense of community can lead to increased customer retention, WOM referrals, and a stronger brand presence.

Now that I've convinced you that blogging still matters, let's talk about our philosophy behind creating popular content.

Our content creation philosophy

When writing blogs for your business we make sure all content we create meets the following criteria:

  1. Topic fit: The audience will find this topic engaging. This is pretty straightforward.
  2. SEO fit: Topics we write about meet our SEO objectives. Your blogs don't matter if people don't see them. This will vary for everyone but as a general rule of thumb we advise that newer blogs focus on keywords with lower search volume (500 searches a month or less) and have lower quality competition. If your blog is growing you can focus on higher quality keywords with greater traffic.
  3. Trending or emerging: We want to make sure what we're writing about is trending (growing in popularity over multiple years) or emerging (has gone from 0 searches to 10+ over the last few months). If your focus is to write about trending keywords it's possible you won't be the first to write about a topic, but you'll know the topic will continue to grow in interest, extending the shelf life of your content. Conversely, if you focus on emerging topics you will likely be the first on the scene, but may not see immediate benefits from these efforts.
Example of what a trend looks like compared to an emerging topic

This philosophy may seem simple, and that's because it is. People have a tendency to overcomplicate the content creation process, but it doesn't need to be difficult.


Why not Google Trends or Trend database tools?

Before we proceed I wanted to briefly mention Google Trends. It's an excellent free tool that can help you see the relative popularity of a keyword or topic with data updated by the hour. Google Trends can be a great method to find trends, but we wouldn't recommend it for a few reasons:

  1. You're only shown what you enter, meaning true trend research is not scalable
  2. Nearly impossible to find related topics that are actually relevant and trending/emerging
  3. You're not shown SEO metrics like search volume to truly understand search interest
  4. You can't develop workflows around Google Trends because there's no public API

I'm sure you've also heard of popular trend discovery platforms like Glimpse or Exploding Topics. Both of these tools are objectively excellent with hundreds of thousands of trends having been surfaced for your viewing pleasure.

To cut down on research time you can use these tools to see what's trending, but the main limitation here are with the trends themselves.

Because these trends tools don't actually find trends for you, there's no chance for true trend discovery.

You'll see trends that thousands of other people are likely viewing and also writing about and building entire companies around.

I would highly recommend these tools if your objective is general market research and understanding specific industries and what's fueling their growth. When it comes to using them for content creation, you can do better.

If your goal is to create tons of daily content about only the most viral topics of that day, Google Trends could be right for you. However, with no API, you might have to build something that can scrape that information, which would be difficult to manage and could break Googles' TOS.

With all of that covered, let's talk about the actual process of creating content.

Three steps of the content creation process

The content creation process is straightforward and can be broken down into three parts.

  1. Research what you're going to write about
  2. Write about the topics you've researched
  3. Publish the articles on whichever blog platform you use
The blog creation process is research, writing, and publishing.

Let's do a deeper dive on each step and how you can automate each one.

Step 1: Identify your list of topics & keywords

One of the most difficult and time consuming things for people creating content is to figure out what people want to read.

The standard workflow would be to use an SEO tool like Ahrefs or SEMrush to conduct keyword research to see which keywords you might want to rank for.

Now allow me to go on a mini tangent while I explain why this method is not optimal.

  • Data Aggregation and Processing: Aggregating and processing vast amounts of search data from various sources can be time-consuming. Some keywords may not be prioritized for frequent updates, leading to outdated data. Some keywords on popular SEO platforms have not been updated in YEARS!
  • Algorithm and Data Source Changes: SEO tools rely on data from multiple sources, including search engines like Google. Changes in these sources' algorithms or data-sharing policies can affect how quickly and accurately SEO tools can update their databases.
  • Keyword Popularity and Demand:  Newer or Niche keywords might not be updated as frequently as high-demand keywords. SEO tools may focus on updating data for more commonly searched terms, leaving less popular keywords with outdated information.
  • Technical Limitations: The sheer volume of data that SEO tools manage can lead to technical limitations, affecting how often they can refresh their databases.

The truth is these tools have billions of keywords in their database, and it's simply impossible to keep all of them updated while also continuously surfacing new keywords in a timely manner. I'm not going to say you should never use these tools, because you absolutely should.

That being said, there's more efficient ways to quickly identify good topics that fit our philosophy where more relevant keywords are surfaced with more modern search data.

Let AI find trending and emerging topics for you

It's our suggestion (and yes we're biased) to let AI research your topic for you to find trends.

Allowing AI to research what's trending in your industry or niche can cut down on several hours of work and give you a higher likelihood of finding some emerging topics no one has tapped into yet.

Let's use the topic "Real Estate" as an example.

The word "Real Estate" on Google trends.

As you can see Google Trends shows the topic of real estate to be overall flat, which would lead you to believe there's probably not too much to uncover without a significant amount of leg work. You'd usually be right, but that's where AI can help.

TrendSwell.ai allows you to discover trends based on your specific use case that you can use for blog creation. Simply tell the AI your goal is content creation, your desired country, language, and enter your topic or industry.

Tell TrendSwell.ai your goal, country, language and topic.

In less time than it takes to tie your shoes, our AI was able to find over 150+ trending and emerging topics related to real estate. These can be both longtail keywords that you might find on a traditional SEO tool and topics that are highly related to what you've entered.

TrendSwell.ai results screen showing real estate trends

TrendSwell.ai works by combining AI's ability to understand natural language and relationships between words and proprietary knowledge informing the model what's already trending now.

The result is getting a combination of words displayed to you that have the highest probability to be trending.

Now that we have over 150+ keywords and topics we know are popular we can export the results to kick off the automation process.

By clicking the blue "download" button in the top right of the screen you can download all the results into a CSV file.

We'll now take this data and enter it into Google Sheets to kickoff the automation process.

Screenshot of downloaded trends on a CSV file.

Step 2: Generating blogs at scale

Now at this point in the process you could simply select the keywords you like and insert them into LLM's ChatGPT and have them write the articles for you.

Another option could be to take the CSV file, upload it to an LLM like ChatGPT directly, have it pick some topics for you and write them one at a time.

This is a cost effective option, but it requires user input and you can also be throttled by LLM's like ChatGPT even if you're subscribed (which is crazy).

We want to build something truly automated that limits the amount of human input needed. To do that, we simply take the data from the CSV file and copy/paste it into Google Sheets.

You will need to have a Google account to be able to access Google Sheets and Google Drive.

Create a new Google Sheet and name it something like "Blog generation system". We call ours "TrendSwell.ai Blog Generator". Once you have your new sheet you will copy and paste all of the data from the CSV file and input it into Google Sheets.

You'll now need to create a new column that can act as a filter to ensure the system only writes about topics you want it to based on your SEO criteria. You can make this filter as rigid as you'd like, but we'll walk you through what we did.

Since we're a small company with a newer blog we want to try and focus on keywords that are trending or emerging with a lower search volume and respectable YoY growth.

Here's the following criteria we decided was best for us when selecting topics to write about:

  1. "trending" column is "TRUE".
  2. "Search Volume" is less than 10,001.
  3. "Volume Growth" is greater than 20

TIP: If you want to write about only emerging trends you can set the "Volume Growth" criteria to be greater than 999.

With our logic written we're now going to put write an array formula with IF functions that can label our criteria. Don't worry if you're not an excel wizard, you can copy and paste this formula into the H1 column.

=ArrayFormula(IF(ROW(F:F)=1, "Criteria Met?", IF((F:F=TRUE) * (B:B<=51000) * (G:G>=20), "Yes", "No")))

Just note that if your criteria is different than ours you'll need to change the numbers in the B:B and G:G section to match what you want.

We created an array so that this formula will be automatically applied to column H without needing to drag the formula down. Since you already have data in the spreadsheet from your first search, you should be able to see how strict your criteria is by the count of "Yes" in that Column.

Look at what your spreadsheet should look like with "Criteria Met?" added.

As a completely optional step, you can also apply a filter on the "Criteria Met?" column so that you only see topics that meet your criteria.

Doing this is purely aesthetic choice and does not impact the actual functionality of the system.

By following these steps you've now completed your trends database, and thus the first portion of this system.

Moving forward you can keep copy and pasting the outputs from TrendsSwell.ai into this system in the next blank row and it will handle everything else.

We'll now go into the next stage of step two which is setting up the actual content generation engine. For this, we'll be using Zapier, but there's other options on the market that could work as well.

Zapier is a software that allows different apps to communicate with each other based on trigger events you establish. Creating these workflows are called "Zaps" and we'll be creating a Zap to take the data from the Google Sheet to create content around it.

Zapier is incredibly intuitive and user friendly, and we'll walk you through every step of our setup so you can copy what we do. Before we start building anything, let's quickly talk about the process we're actually creating.

  1. The process begins when new data is added to rows in Google Sheets.
  2. A filter is created that checks if a keyword meets our criteria. To reference the bachelor, topics who have "Criteria met" as "Yes" get a rose and if "Criteria met" is "No" they get voted off the island... am I mixing my references? I don't watch any of those shows... anyway...
  3. ChatGPT writes content based on the keywords who met the criteria.
  4. The content is sent to your CMS for further review. Our blog is on Webflow, but whatever system you use is fine.

Now that you understand what we're building... let's cook.

Building the Zap

To get started, create a Zapier account. You won't need to pay anything to get started.

Once you're logged in you'll click on the "Create" button and select "Zaps". You'll now be on a fresh screen to begin building. You can use their Copilot feature to have AI build the zap for you, but we suggested just following our lead on this one.

A new Zap will be blank when you open a fresh one.

Click on the "Trigger" box and select Google Sheets. From there you will select "New or updated spreadsheets row" as your event. From there you will follow the instructions to add your account and then setup the triggers.

Under triggers you'll select the spreadsheet you made for this system and the worksheet. Next you'll need to select the trigger column, and you'll select the "Criteria Met?" column we had made.

You'll then want to test the trigger to ensure everything ran smoothly. If the test goes smoothly Zapier should find a row in the spreadsheet and you'll get green checks on everything. Be sure to contact Zapier support if you have issues during this phase, or any other.

Make sure that when you run your test all of the columns have data in them. As you can see in the image below there are values for the keyword, competition etc.

Zapier is confirming the Google Sheets integration has been successful.

IMPORTANT: If you select a Row that doesn't have data in it (this can happen due to the array formula) the Zap wont set up properly. Zapier tries to pull data from the most recent columns, but these will all be empty other than the "Meets Criteria?" column so the tests wont run.

You can see in the image below that Zapier is pulling data from row 1000 in the spreadsheet, but there's no data in that row so there's so everything is blank.

Data will be missing if you add the array formula to your sheet.

You can account for this by entering dummy data right next to each of the column titles. To do this, click on the icon in the bottom right with the pencil that says "Duplicate to edit" when hovered over.

Creating a dummy row will allow you to properly create the system.

Once clicked you should see a new row is added called "Modified record". Simply click to the right of each purple rectangle and a field will appear where you can enter text.

You can enter whatever information you'd like into these cells as long as they would make sense. For example, you'd want to be sure to put numerical values where they make sense like competition, search volume, CPC, and Volume Growth and "True" or "False" in trending.

Enter your dummy data into the new modified record.

With the dummy data entered you can now hit continue and finalize the test. If the test is successful, the integration of Google Sheets will be complete.

With Google Sheets integrated we now need to setup our filter. As mentioned earlier, creating a filter allows for the Zap to handle what to do with keywords that meet and don't meet our criteria.

Under the trigger on the dashboard you'll see a little "+" that you can add between step 1 and step 2.

Click the "+" and you'll see a pop-up menu where you can add a new step. You're going to select the "Filter by Zapier" option under the "Popular built-in tools" section.

Select the Filter by Zapier option for your next step.

You will now be able to create your filter criteria. To do this you'll need to make sure you select the "Criteria Met?" in the insert data dropdown, select the option that says ("Text) Exactly matches" and then type yes below that. This tells the Zap to only run keywords that match our criteria through the system. Your filter should look like what we have below.

The filter ensures the zap only runs on the keywords that meet your criteria.

We're now ready for the next step. Click on the "+" to add the ChatGPT app to appear after the filter in the workflow map. Your workflow should now look something like the image below.

What your Zap should look like to this point.

With ChatGPT selected you'll need to select the "Conversation" event which will allow you to give the AI your blog writing prompt.

The Conversation option is what makes the API function like regular ChatGPT.

Before you continue you will need to go to OpenAI directly (not ChatGPT) and create an API key. To do this you'll go to platform.openai.com and click on the "API" option which should take you to a dashboard.

Select the API option to go to the API dashboard.

Once at the dashboard you'll see an item in the menu called "API Keys" where you can click and generate a key.

It's important that this key remains confidential, as it's a string that allows Zapier to know whose account they are accessing.

Click on "Create new secret key" where you can name it. Make sure "Permissions" is set to "All".

Create your API key and never make this information public.

Once that API key is generated you'll copy and paste it into Zapier so you can move to the next step which is setting up the Action step.

In the Action step is where you will instruct the AI on how you want it to write your blog. How you generate this prompt is up to you, but we will tell you what fields we filled out and what we selected. If we don't mention a field that means we just left it alone.

You'll start with "User Message. Think of this as the chat field you would usually put into ChatGPT when you use it. The only thing you'll add in here is a dynamic field which is the "keyword" option. When you click on the box you should see a dropdown of different options you can select.

Think of this field as you interacting with ChatGPT as you normally would.

Next you will select what model you want to use. What model you want to use is personal preference based on quality, time, and cost. At the time of writing this, gpt-4o and gpt-4o mini are the best models available that can write blogs quickly and cheaply. We used 4o in this worflow.

Next is the assistant instructions which will house the actual prompt that will instruct the AI on the task. Creating the prompt in this case is mostly trial and error.

To save some time and several cents we advise just having ChatGPT help you with a prompt for this and test it out there.

There are even GPT's trained on prompt engineering and blog writing you may be able to lean on to help guide you through this process.

Since our blog writes about trends, you can get an idea of what our prompt might look like in an earlier demo prompt we tested out in the image below.

Notice that you can put additional columns in the instructions as well which are dynamic fields.

In our case, the blog we write will also contain information about the search volume and YoY growth since our reads would likely care about that if they are reading about trends.

The Assistant Instructions field is where we create and enter our instructional prompt.

Lastly we fill out the "Max Tokens" section. This essentially puts a cap on how long your blog can be, and as a result, how much you'll spend per blog. At the time of writing this blog 1 token was equal to 4 characters.

We put our token limit at 3600, which would be enough for a blog on the shorter side. This would cost us about $.14 a blog if we're being conservative, but if you use 4o mini it would be around $.01 (yes you read that right).

Some models do have token limits and differing costs, so these are all things to consider. gpt-4o has a high token limit and is pretty cheap so that's the best option as of July 2024.

With your parameters set now it's time to test the integration. If it worked you should see a "Reply" field in the test window that contains your blog content.

The "Reply" tag is the output from ChatGPT. We'll need to use this soon.


If the test worked without any issues you've successfully created the Creation phase of this system and you're ready for the 3rd and final step.

Publishing your content

This last section won't be as in-depth because the CMS we use (Webflow) is likely different from yours. Just know that the main goal here is to automatically send the content that ChatGPT creates to the blogging website you use.

Your event would likely be something like "Create item" or "Create post" but this will vary on the platform you use.

Zapier can also allow you to add additional fields to your content automatically to help further automate the process.

For example, you can see in the image below we can assign written content to collection, which means the category it falls under on our website. And since I'm the only person who writes blogs at the company I can just put my name in the "author" field and that will be on every blog that pushes to Webflow.

Note that your fields may not match mine. What fields you see under the "Action" section all depends on your platform and what classifiers you have active on your website.

You  want to make sure that you put the "Reply" input from ChatGPT into whatever section your written blog goes. In Webflow, that's the "Post Body" field but it may be something else for you.

Ask Zapier support if your CMS is different and you get stuck.

The name field for us is also what the title of the blog will be called. We made it so that all of the titles of the blog are dynamic so they can be called, "New Trend: {User Request Message} which would be the keyword that was entered into the prompt.

This is just one of many ways you can get creative with automating different elements of the blog besides the blog itself.

The Title of our blogs are automated as well.

We also heavily suggest sending any automated content you create to your website as a draft and not automatically publishing it.

Before you publish anything it should always go through a manual review to ensure everything is accurate and it also has some human rizz.

While you can probably get away with publishing pure AI content if you have a good prompt that writes in your voice, we think it's better to use the written content more as a big pile of clay that you can mold to your liking.

With the integration complete you can run the final test and it should send the blog right to your website. As you can see below, all of the information we put into the automation created the new blog post as a draft and is ready for review and publishing.

A look at what our Zap created and sent to our website backend to review.

Congratulations. You've successfully created a viral blog making machine with no code and no fuss.

You're an absolute rock star. I'm proud of you.

You've now successfully created a system where your cost to produce a blog is incredibly cheap, and only gets cheaper the more you make. The image below is what the final Zap should look like.

The finalized blog creation system.

Things to consider

With the system created you'll now need to make sure you signup for Zapier if you plan on writing a high volume of blogs continuously. Depending on how much you write, Zapier will most likely throttle you on the free plan.

Since TrendSwell.ai can also find hundreds of trends per search, you may also run the risk of creating hundreds of blogs if your filter criteria is too lenient. To set a cap on how much you spend you can also set spending restrictions on OpenAI which caps your usage of the API at whatever dollar amount you set.

Also note that if you ever edit or add things to any rows in the spreadsheet this will automatically cause the Zap to re-run and could build up a long que. If you're ever going to edit the spreadsheet in anyway such as adding more columns or rows, make sure you pause your Zap first.

How much does this system cost?

Great question. Not to sound like an overpaid McKinsey consultant but, "it depends". Let's break down the cost of each tool in the system:

TrendSwell.ai: Starts at one-time cost of $149 going up to $249 depending on your needs.

Google Sheets: $0

Zapier: Starts at $20/mo with 2 week free trial.

OpenAI: Depends how many tokens you use and the model. Using gpt-4o mini you could pay as little as a few cents (or less) per article even if it's reasonably long.

As you can see, you can get started on this system right now for under $200 and keep the machine chugging for less than $30 a month for most people from there on.

Closing thoughts

I hope you this article was helpful to you on your quest to automate more of your business. If you need any help creating this system you can reach out to us or ask Zapier who also has an excellent support team.

What's great about this integration is it's both affordable and fairly easy to set up. Your blogs will generally only be as good as your prompt, so really make sure you work on your prompt engineering.

Automated trend discovery also has a lot of other broader use cases for things such as product research and competitive intelligence. These processes can also be fully automated and we'll try and make guides just like this one for those as well.

If you'd like help creating your own custom automations, or if you need help with more in-depth trend research for a much larger initiative, we also provide bespoke research services as well where we've helped corporations, governments, and small businesses alike.

We wish you luck on your blog creation journey!

Trend research, done for you!

Find relevant trends instantly with AI.

Discover thousands of trends simply by telling our AI your specific goal or industry in a sentence or less.

*14-day moneyback guaruntee
Global trends

Find trends in 77 countries and 55 languages.

Instant results

Our AI finds trends for any use case in under a minute.

CSV downloads

Download the results to develop custom workflows.

SEO data

See search volume, CPC, competition, and YoY growth.

Subscribe to our weekly email newsletter!

Check - Elements Webflow Library - BRIX Templates
Thanks for joining our newsletter
Oops! Something went wrong while submitting the form.
Join the 1,000+ people who get weekly trends.