• No more slogging through messy spreadsheets and a mountain of data
  • Perfect for conversion copywriters, marketers, founders and product managers
  • Takes 5 minutes to analyze qualitative survey data with ChatGPT

When was the last time you got excited to dive head-first into hundreds, if not thousands, of survey responses?

You know golden nuggets are in there somewhere, but you just. can’t. start.

You’re blinded by the sheer volume of data. Those cells are getting blurred. You’re getting cross-eyed.

Which is why you’ll love today’s AI trick. I’m going to show you how to use AI to extract insights from any survey—in seconds (fancy charts included).

Here’s what to do: Prompt ChatGPT 4 to look at a spreadsheet with your survey results and give you its observations. This is what it generated for me when I used the AI prompts I’m about to teach you:

ChatGPT-produced visuals
ChatGPT insights from survey data

Here’s how to use ChatGPT 4 (Advanced Data Analysis model) to analyze survey data and get similar results for your business or client.

Step 1: Give ChatGPT some context on its role

Before getting started, I always like to tell ChatGPT to step into the shoes of a copywriter. I find it returns better, more specific results if I give it a “frame of mind.”

In ChatGPT, paste the following prompt:

Copywriter shoes prompt for ChatGPT

Copy paste this prompt:

You are a conversion copywriter, expert in decision making, persuasion, psychology, behavioral economics, marketing, sales, UX design, customer experience, branding, and conversion rate optimization. Pretend that you are also highly empathetic and understand how people think and what makes them tick. You can easily and expertly detect human behavior, thoughts, and needs based on language.

Step 2: Share the survey data (spreadsheet) you want to analyze with ChatGPT

Next, we’re going to share our spreadsheet with ChatGPT. You don’t have to worry about the format as long as it’s either a .CSV or Microsoft Excel file. Don’t even worry about the columns and rows. ChatGPT will figure it out.

What matters here is that you know a few things:

  • The type and purpose of the survey
  • A list of the questions
  • The possible responses (if multiple choice)
  • A quick description of what insight each question can provide

Here’s an example from a recent client project (a SaaS selling proxy products).

Example of how to organize survey questions

With that in mind, in ChatGPT 4 (Advanced Data Analysis model), attach your survey spreadsheet by uploading it with the “+” button:

How to upload a file to ChatGPT

Then enter the following prompt:

Prompting ChatGPT to analyze the survey

Copy paste this prompt:

Here's a spreadsheet with results from a [SURVEY TYPE] for my client [CLIENT NAME]. This is a survey that [FUNCTION OF SURVEY].

Below are the questions, each with the possible answers and with the insight responses can give us. 

[Question 1]
[Possible responses 1]
[Insight from response 1]

[Question 2]
[Possible responses 2]
[Insight from response 2]

[Repeat for more questions/Responses]

Look at the spreadsheet and extract the insights with visuals/graphs/charts where needed and applicable. Then for each survey question extract the main finding, insight from that finding, and action step based on that insight.

Note: Here are a few examples of how you could fill out the placeholders in the prompt above…

  • [SURVEY TYPE]: “Website survey”, “Customer survey”, “Post-purchase survey”, “Exit intent survey” etc.
  • [CLIENT NAME]: just use your client’s or business name so ChatGPT has a reference
  • [FUNCTION OF SURVEY]: Here in my case I explained how my website survey worked (displayed after 6 seconds of landing on the site. Depending on your needs, just describe how users will stumble on your survey. For example, will they receive an email invite with a link to it? When? After buying or just after signing up for a free account? After how many days? etc. This will help ChatGPT get some context around the user experience.

At this point ChatGPT will have shared the graphs and charts you need in the chat. In order for you to download them, we need to ask it to provide a link. We’ll ask for a folder so we don’t have to download each image one by one. 👌

In ChatGPT enter the following prompt:

Asking ChatGPT to extract the images for download

Copy paste this prompt:

Export the images in jpeg for me to download in one single folder.

Within a few seconds, you’ll get a link to download all your rich survey data in compelling graphs and charts.

ChatGPT sharing the download link for images

You can view the full chat here. Spoiler alert: the questions and even the spreadsheet I’ve used to extract the insights came directly from ChatGPT itself! Yes, for the purpose of this article, I’ve even used it to simulate my website visitors’ responses to the survey. 😆

Note: the public link won’t let you view or download the visuals, but you can still read the full instruction set.

I tried this AI prompt for other survey types and on a real client project. Here’s what happened.

I’ve used this same exact series of prompts to analyze a real client project. And for 3 different types of surveys (website poll, free sign up survey and post-purchase survey). This is part of what ChatGPT produced for me:

Example of charts from client project survey analysis
Example of charts from client project survey analysis 2
Example of charts from client project survey insights
Example of charts from client project survey insights 2

When it comes to how to analyze survey data using AI, let ChatGPT do the grunt work. That frees you up to do the investigative work.

Keep in mind that the more material/research data you will feed to Chat GPT with your spreadsheet, the better the insights you will get with this prompt will be.

Always take ChatGPT’s recommendations with a grain of salt and triangulate with the rest of your research. But with this method you could shortcut your analysis process by a couple of days.

Not sure how to use survey data and insights to actually write copy? Head over here.

FacebookTwitterLinkedInEmail