‍Key takeaways:

• This blog will teach you how you can use discover and research trends automatically.

• TrendSwell.ai can be used to automatically discover trends for you based on your use case and industry.

• TrendsGPT can be accessed on ChatGPT and is an AI trained in analyzing trends data and gathering more.

Introduction

AI has revolutionized the process of market research. LLM's like ChatGPT can significantly speed up the time for repetitive and manual tasks like data collection, aggregation, and reporting.

‍

People are even creating custom GPT's (spoiler alert) that are trained specifically on market research tasks or trained on proprietary datasets to provide unique insights.

‍

One of the main components of market research is taping into business and consumer trends. These can be trends that are just popping up (emerging) or patterns that have been steadily increasing over the years.

‍

The problem, however, is that apps like ChatGPT aren't good at market research out of the box, and you certainly won't get any kind of unique insights compared to anyone else.

‍

This article will break down how you can use AI to expedite data aggregation for trend discovery and analysis.

‍

Can't ChatGPT just tell me what's trending?

Sort of. If you ask ChatGPT directly what's trending the first issue you'll likely run into is the lack of consistency. For example, I started with an extremely simple prompt, "Please tell me what's trending in real estate", and received vastly different responses.

‍

In the first response ChatGPT generated a python script that simulated the process of generating a list of potential topics existing within its training set. This is interesting, and will get back to this in a bit.

I then opened a new chat with the same exact prompt and ChatGPT had a completely different process and generated entirely different results. This time, ChatGPT went online and researched various different websites to pull a few topics and write some results about what it found in more of a blog format.

‍

Both of these results can be useful depending on your use case, but the code-based output is more helpful. Not only does it provide more topics (15 compared to 5) but it also provides logic for its response rather than just scraping the web which can produced bias or inaccurate results.

‍

The key to the code based prompt can be seen in the following snippet below.

‍

# I'll list common trends based on recent knowledge and prevailing patterns in the industry.

‍

By generating this output, ChatGPT is informing us that these responses are based on its training data, meaning it's seen these topics listed the most frequently compared to others.

‍

One of the more powerful elements of LLM's like ChatGPT is their ability to understand relationships between keywords. We can use this to our advantage when asking ChatGPT to potentially dig up more trends for us.

‍

That being said, how well did it to?

‍

We entered the keywords into Google Trends and looked at the last 5 years and found at the time of writing this blog that 5/15 could be classified as trending, and only three of those were a strong trend.

‍

‍

All things considered 3/15 isn't bad, especially for an entire process that took about 5-10 minutes to validate.

‍

But what if we want more results? Since ChatGPT did an ok job at surfacing 15 topics we were curious to see how it would handle a much larger ask for those who are hoping to conduct deeper research.

‍

We made the same request but this time asked for 50 results in the healthcare space. We painstakingly took all 50 topics and ran them through Google Trends to see what the results were. Let me just say... that was NOT a fun task, but I did it because I care.

‍

‍

‍

As the number got larger, we saw a dramatic decline in topic quality and relevance. Granted this was just one test on one industry, we were surprised (in a bad way) at the results.

‍

‍

So what happened? We noticed that much like before the first 15-20 words had a similar hit rate which was promising. However, as we got further down the list we noticed two glaring issues:

‍

1. ChatGPT was suggesting relevant topics but they were extremely low quality not even being close to trending.
2. ChatGPT would suggest topics that are trending, but they were less relevant topics "AI chat bots" and "Voice Activated Digital Assistants".

The issue here is obvious. In fact, ChatGPT told us what the issue was in the very first line of the first snippet of code it gave us.

‍

# I will simulate retrieving the current trends in real estate using a general approach since I don't have access to real-time data.

‍

And there lies the fundamental problem with LLM's. They are excellent at surfacing relevant topics and researching those topics, but because it doesn't have access to real time data, it fails to produce topics that are verifiably trending.

‍

All of this is to say you can use ChatGPT to help you finds trends, but the leg work involved to verify if they are trending is likely not worth the effort. As the saying goes, the juice isn't worth the squeeze.

‍

There is something else you could do, however. What if you had ChatGPT generate a list, and you import that list into an SEO tool to give you search volume?

‍

Sure! You could certainly do that! It's far faster than going the Google Trends route of entering everything manually, and you'll get actual search volume numbers rather than that 0-100 scale.

‍

So why don't we suggest it? The problem is the recency in which SEO tools often update their data. Seriously, go to your large SEO tool of course. You know the ones.

‍

Enter in a topic and go down the list. You'll be horrified to see how most of the keywords have not had their search volume updated in months (sometimes years) and it gives you know way of truly knowing the interest in a topic more recently.

‍

That's one thing Google Trends can do that no one else can. Having data updated to the hour is very powerful, especially if you care more about virality and trends that pop up and will likely last a week or less.

‍

To make things more clear to this point here's what we know:

‍

• ChatGPT is good at generating a small number of topics and researching them but doesn't have data to validate trends.
• Google Trends can get us relevant data but does not provide search volume and requires significant manual effort.
• SEO tools are good for generating results and getting SEO data but search volume can lag way behind.

So what if there was a way we could make ChatGPT smarter so it generated more trending topics, while providing some of the recency Google Trends can, while also having all of the SEO data that SEO tools provide.

‍

Meet TrendSwell.ai!

TrendSwell.ai was created to solve the exact problem that we've showcased thus far. TrendSwell.ai takes the best parts of ChatGPT, Google Trends, and SEO tools and blends them into one tool that's focused on surfacing as many trends as possible.

‍

TrendSwell.ai uses the same AI models that ChatGPT, Claude, and other LLM's use as part of it's process, but unlike those models, the TrendSwell.ai model has current and constantly evolving knowledge on what's actually trending based on recent search volume data.

‍

However, TrendSwell.ai doesn't just pull from an existing database of trends like some other trend discovery tools on the market. Our model actually takes the trending knowledge in its dataset and actively builds upon it every time it researches an industry.

‍

That's right, our AI is researching your topic or industry in REAL TIME and pulls hundreds of results it thinks have the highest probability to be trending.

‍

The end result is a trend hit rate that's much closer to 40%-50% with 100's of topics compared to just a 16% hit rate with 50 topics.

‍

‍

That means more trends, that are more relevant, more specific to you, and surfaced 100X faster than any manual process you could undertake. You'll also be given additional SEO metrics like Cost Per Click (CPC) and Competition on ranking for these topics in paid ads.

‍

TrendSwell.ai not only finds trends, but bases the research based on your specific use case.

‍

Tell the AI whether you're conducting product research, trying to understand the competitive landscape of the market, or create content, and it will focus it's trend research around this specific use case.

This means if you enter "Product Research" as your research goal the AI will focus its research most on finding products related to your search that could be trending, but will still mix in standard general topics that are related to help with writing product listings and further understanding an industry.

‍

So once trends are surfaced what's the next steps? This is where ChatGPT can take your research to the next level.

‍

TrendsGPT for trend Analysis

TrendsGPT is a newly launched AI assistant that's trained to help TrendSwell.ai users on how to use the platform and analyze results. Note you will need to be subscribed to ChatGPT to access this and other GPT's on the market.

‍

To find TrendsGPT you can click here or go to the GPT section of ChatGPT and search for TrendsGPT in the store.

‍

Once you start a new chat you can upload the search results from the CSV file and begin to ask questions about the data to help further analyze the results that TrendSwell.ai produced for you.

You can type in whatever you want or click one of the pre-generated discussion starters. To see an example of what TrendsGPT can do we'll select the option to "Make a word cloud of the best trends in my dataset".

‍

In the example above you can see the results from TrendSwell.ai were uploaded into ChatGPT. The AI has been trained on how to analyze the dataset and perform custom calculations to help you surface the best trends.

‍

This word cloud visualization can now be used to understand at a glance which trends are the most impactful and you can even re-download results into a new CSV file if ChatGPT gathers even more new information or you.

‍

Let's look at another example with new data.

‍

In this new example TrendSwell.ai found results where my use case was "Competition" and my search was "real estate". TrendSwell compiled a list of hundreds of trends and 80+ companies of all various sizes and found which ones were trending.

‍

From there, I exported the results and uploaded them to TrendsGPT where I wanted it to only analyze the companies on the list and make a word cloud and a new spreadsheet that could be worked off of.

ChatGPT was able to understand specifically what I was asking for and generated a specific word cloud I needed as well as a new sheet that had companies in order from most relevant to least relevant based on various datapoints relating to their trending status.

‍

‍

Now let's crank this research up by further leveraging the power of TrendsGPT.

‍

Now that I have identified some of the most relevant companies in the space let's take the top 10 most relevant companies and conduct some market research through automated data collection and sentiment analysis.

‍

First, let's ask TrendsGPT to get the websites of the top 10 companies into a spreadsheet so that we can have them for future reference. TrendsGPT was able to do this with ease.

‍

Now let's say we wanted to get a better idea of the sentiment around each company to identify customer satisfaction. TrendsGPT can do this for you as well. To make things more challenging, I won't give it specific directions on how to do this and let it use its best judgement.

‍

You can see I've told it I want some kind of quantitative metric to be used, even though I don't quite know what that could be. TrendsGPT will use its general knowledge on market research to search the web to get the information and get a new metric. Note that you may need to encourage it to use the internet to look it up sometimes.

‍

TrendsGPT found various reviews across various different websites and was quickly able to get qualitative information on one of the companies. That's great, but it still needs to take all of this information to create some kind of sentiment metric for our analysis. TrendsGPT created an easy to understand metric, showed the logic, and then created a new table where we could house this information.

‍

‍

TrendsGPT continued to go down the table until it was complete. Because ChatGPT puts token limits on individual responses you may need to prompt it to continue a few times for the results to be complete.

‍

The end result is an entirely new spreadsheet generated by AI that has found trending companies using TrendSwell.ai and used TrendsGPT to further analyze and add new data.

‍

Closing Thoughts

ChatGPT has revolutionized the way people can conduct market research and better understand trends. The main limitation ChatGPT has is it does not have access to actual search volume data on topics so it can fully understand what's trending and what's emerging beyond what may be appearing in its training data.

‍

By using TrendSwell.ai and TrendsGPT you can now further automate the trend research process and better analyze the trends that are found and which ones matter.

‍

Whether it's helping you pick topics for your blog, creating visualizations for your clients, or something else, AI can save you hours or even days conducting research that now takes just minutes.

‍

If you want to see how you can also you AI to create a viral blog generating machine, you can check this how-to guide on that by clicking here.

‍