Brands must understand their audiences to find success. But it isn’t enough to collect the data, you must understand it. And when it seems like the data is too much to handle, artificial intelligence just might help.
In the past year, marketing interest in AI has boomed. Every discipline is looking for ways to employ the latest technology. Market researchers have even found ways to uncover human insights with AI.
Researchers are employing chatbots to brainstorm research ideas and conversation prompts. Once their surveys, focus groups, and interviews are complete, they turn to unstructured data processors to make sense of daunting mountains of content. While these technologies aren’t what make human insights special, they are cutting down laborious and repetitive tasks, leaving professionals with more freedom and energy to elevate their work.
To get a hands-on understanding of how researchers are using AI, we sat down for a Q&A with Timothy Sauer, Ph.D., director of research and account strategy at (human)x and Rachael Roland, (human)x consultant and cofounder of Enable AI.
This blog is the final installment in a three-part series called “The Human + The Tool” on the real human applications of AI with subject matter experts at (human)x in search engine optimization, UX, and market research.
*Author’s note: This interview was edited and condensed for clarity.
Brands must understand their audiences to find success. But it isn’t enough to collect the data, you must understand it. And when it seems like the data is too much to handle, artificial intelligence just might help.
In the past year, marketing interest in AI has boomed. Every discipline is looking for ways to employ the latest technology. Market researchers have even found ways to uncover human insights with AI.
Researchers are employing chatbots to brainstorm research ideas and conversation prompts. Once their surveys, focus groups, and interviews are complete, they turn to unstructured data processors to make sense of daunting mountains of content. While these technologies aren’t what make human insights special, they are cutting down laborious and repetitive tasks, leaving professionals with more freedom and energy to elevate their work.
To get a hands-on understanding of how researchers are using AI, we sat down for a Q&A with Timothy Sauer, Ph.D., director of research and account strategy at (human)x and Rachael Roland, (human)x consultant and cofounder of Enable AI.
This blog is the final installment in a three-part series called “The Human + The Tool” on the real human applications of AI with subject matter experts at (human)x in search engine optimization, UX, and market research.
*Author’s note: This interview was edited and condensed for clarity.
What is your background with AI?
TS: With AI, my role on the team is toe-dipper. AI has emerged in popularity quickly in the last year, and I’ve had to stay on top of its constant development and try things out.
Integrating AI into our work process has involved a lot of experimentation and looking for efficiencies that can be gained. I look for ways to integrate cutting-edge AI programs in appropriate ways. For example, I can’t use ChatGPT as a resource database. I can’t ask it “How do people currently use Tik Tok?” The data might be two years old and inaccurate. But I have used it for efficiency, like generating prompts or jumping off points for writing prompts. I’ve also used ChatGPT to help process documents and data. It’s cut down the number of working hours on a lot of projects for my team.
RR: I started the company Enable AI a little over six years ago. We were working with unstructured data, such as text and images. Many of the projects I did related to product reviews. I would scrape Amazon, Walmart, and other retailer websites, and pull reviews on my client’s products to understand what people were talking about, what they cared about, what was driving one-star reviews or five-star reviews, and so on. We also did competitive analyses with social media i.e., Facebook versus Instagram.
As Timothy mentioned, about six months ago AI rose in popularity and the world shifted under my feet. All the things I had been doing the old-school, machine learning way, I could suddenly do in ChatGPT. When I say old school, I mean techniques from three years ago — that’s the olden days now.
Rachael, to the average person, it feels like just six months to a year ago AI exploded in popularity. What inspired you six years ago to start working in AI?
RR: At that time, AI wasn’t a common term. It was often referred to as the “deep learning revolution.” My interest was piqued when I connected with friends at Georgia Tech who were taking online courses in computer science, machine learning, and AI. Through them, I got a glimpse of the exciting developments in machine learning and the potential of working with unstructured data. It was clear that this field was rapidly expanding, and I saw an opportunity to be part of it.
While conducting research, what AI programs are you using?
TS: ChatGPT is the hub for me. I use several plugins, including AskYourPDF and WebPilot, as well as the advanced data analysis tool available in GPT-4.
We have to be careful with using ChatGPT and open-source AI because we work with some private client insights. A lot of research is taking primary data, owned data by a client that we’ve gathered with focus groups, surveys, or some other research. We don’t want it going into ChatGPT itself because of privacy and security issues.
With Rachael’s help, our team is essentially taking the ChatGPT model and building our own on Microsoft Azure that is self-contained, and we can engage with. All the data that we put in and everything it puts back out to us is contained on our server. Security is our main concern. The last thing we want is for our clients to lose trust in us.
Can you describe a research project where you used AI?
RR: With ChatGPT, I discovered the ability to extract summarization. For example, if you get a long email and you say, “I don’t want to read all this,” you can tell ChatGPT to read it and give the main bullet points.
One of (human)x’s current clients sells fire starters and fire logs. We scraped online reviews of their products and processed it with AI. We uncovered major topics of interest of concern. It’s a long-lasting product, but some people think it smells plasticky when it burns. It’s a quality product, but it can be expensive. Businesses want to know about public opinions, and by utilizing AI we were able to present that information quickly and easily.
When I started Enable AI, we worked with a nationwide, long-term healthcare company with about 10,000 employees. They would do an employee survey every year with multiple choice questions and three open-ended questions at the end. Every year, they would get the surveys, analyze the multiple-choice answers, but do nothing with the open-ended answers, because it’s daunting to read 10,000 responses.
The first thing we did for them was utilize our program, designed to make sense of unstructured data, to analyze all these open-ended responses. We came to the meeting with the major themes our program had gathered. We told them: “black scrubs.” We found out that there were hundreds of occurrences in the survey results of people saying they’d prefer to wear black scrubs to hide stains throughout the workday. They didn’t know that their employees wanted this until we really started delving into those open-ended responses.
One of the fundamental things that we’ve done with unstructured data is pull up unknown unknowns. There are things that you don’t know about. You can find these things if you really look. Machine learning and AI is powerful in its ability to pull out those sort of things — even simple things like topic extraction, sentiment, named entity recognition, and figuring out people, places, data, times, and things that seem simple but can be daunting and time-consuming when done manually.
Timothy and other research specialists at (human)x may go through transcripts from interviews or focus groups, reading all these things to annotate of label them. It can be time-consuming. That’s a great use case for ChatGPT or some other AI program, because it’s tireless. AI is happy to sit there and read 10,000 responses or an eight-hour transcript.
TS: Building on what Rachael said, the “aha” moment in the research comes from us probing and asking important questions in a focus group. The beauty of AI is that the 40-page transcript that our eyes gloss over, ChatGPT can go in and pull information for us to help elevate our work.
AI not to the point yet where we’re using it for hard research. It might get there. We’re mostly using it to help prompt conversations, build briefs, or pull major themes.
In research, there are projects where I know our work is 98% confirmation for what our clients already know. But there is that 2% of information they don’t know. That’s what I want to find. That’s what I want to bring them.
Tell me about (human)x’s process of adopting AI.
TS: There is a community atmosphere of learning in our team when it comes to AI. (human)x, and the entirety of Leap Group, is invested in it.
We have dedicated Slack channels for talking about AI and sharing our experiences. There is so much to say and learn that no one person can keep track of what’s happening. We use that channel quite a bit to gather all our progress.
There is an AI ethics committee. Some team members across disciplines have created a separate AI club, where we meet about once a mother to explore how we will use AI. We have developers, strategists, researchers, creatives, and project managers gathering to see how our teams will approach new challenges with AI.
(human)x just launched a new data-enrichment product that uses predictive technology. Can you tell us about humanView?
TS: I had a conversation with ChatGPT to help me describe humanView. After some back-and-forth with the chatbot, I came to a version that best explains our new product. Here it is:
ChatGPT: humanView™ goes beyond the traditional boundaries of data enrichment. While its core feature revolves around appending first-party data by adding hundreds of relevant psychographic and demographic variables, it delves deeper into understanding an audience’s nuances. This enriched data provides a multi-dimensional view of an audience, painting a more detailed and holistic picture of their preferences, behaviors, and tendencies.
But what truly sets humanView™ apart is its adeptness at harnessing the power of AI and machine learning. In today’s digital age, a significant portion of valuable data comes in unstructured forms, like customer reviews, feedback, and open-ended survey responses. Processing this unstructured data manually is not only time-consuming but also prone to oversight and errors. humanView™ utilizes advanced AI and machine learning algorithms to sift through this vast sea of unstructured data, extracting meaningful insights and patterns. By analyzing customer reviews, for instance, businesses can gain a deeper understanding of their customers’ pain points, preferences, and sentiments. This capability transforms raw, often overlooked data into actionable insights, empowering businesses to fine-tune their strategies and offerings based on genuine customer feedback and experiences.
AI + Market Research with (human)x
Timothy Sauer and Rachael Roland’s experiences showcase the opportunity AI can afford to researchers interested in exploring new technologies. Here is what we learned.
- As this generation of artificial intelligence is still new to the user, marketers and researchers are still going through trial and error to determine the best ways they can use AI.
- AI isn’t necessarily a new thing. It’s gone by different names over the years: deep learning, machine learning, predictive modelling, etc.
- This technology is quick-moving. Techniques and programs from a few years ago are now outdated.
- As a researcher, open-sourced AI like ChatGPT can be a concern when working with owned and private data from a client.
- (human)x researchers aren’t using AI during the research process. Instead, they are using it to brainstorm ideas beforehand, and process information afterward.
- Those “aha” moments in research come from human moments between the researcher and the subject. AI cannot make up for that. But it can take on laborious tasks and give the researcher more time and energy to find the “aha” moment.
AI offers new opportunities for innovation not just in market research, but in search engine optimization (SEO) and user experience (UX) production. The first installment of The Human & The Tool covered AI and SEO with (human)x Senior SEO Analyst Russ Allen. Our second blog discussed AI and UX with (human)x UX Producer Jake Zastrow. Both Q&As are available on our blog.
If your brand wants to learn more about how AI can elevate your market research, (human)x can help. Alongside our in-house, Ph.D.-led research team, we launched humanView, a first-party data enrichment product that uses predictive modelling to make sense of your data and better understand your audience. To learn more about our service, visit our website or contact us for a consultation.
