DIY with AI: How to Streamline Your Qual Research
Editor’s note: This article is an automated speech-to-text transcription, edited lightly for clarity.
AI is revolutionizing the research landscape, especially for those embracing DIY methods. But mastering AI to streamline the collection, analysis and sharing of qualitative insights efficiently and cost-effectively requires some guidance.
On September 25, Erica Dinger, director of research services at Voxpopme provided some guidance to attendees.
Read the transcript below or watch the recording to learn all about how to streamline AI use for your organization.
Webinar transcript:
Joe Rydholm:
Hi everybody and welcome to our presentation, “DIY with AI: How to Streamline Your Qual Research.” I'm Quirk’s Editor, Joe Rydholm. Thanks for joining us today.
Just a quick reminder that you can use the chat tab if you'd like to interact with other attendees during today's discussion. And you can use the Q&A tab to submit questions to the presenters and we'll get to as many as we have time for at the end.
Our session today is presented by Voxpopme. Enjoy the presentation!
Erica Dinger:
I'm Erica Dinger and I'm the director of research services for Voxpopme.
I've been in the market research industry for 25 years, both on the brand side and on the services side. I've worked both in the nonprofit space with AARP, conducting member research, public policy research and designing and evaluating programs. I've also been on the for-profit side at Under Armour and on the services side as head of full-service research for North America at Qualtrics.
I've worked with organizations across almost every field, so believe me when I say I have been where you are sitting right now.
For those of you that haven't heard of us, Voxpopme is an AI-powered qualitative insights platform for researchers at many of the world's largest companies. Our team is at the forefront of applying AI to qualitative research to make human stories more scalable and accessible, which is an incredible transformation.
Fortunately for me, as Voxpopme's Director of Research Services, I get to deliver full service qualitative projects for insights teams at major brands and agencies across the U.S.
But before we get into Voxpopme's, AI-powered tech and how it helps us analyze qualitative data, let's cast our minds back to November of 2022.
In November of 2022, OpenAI released ChatGPT. It was the first time just about anyone could easily interact with AI and we refer to this AI as generative AI. Now, what was so special about generative AI?
Well, it responds to us in a way we'd expect a colleague or other humans to respond, and it understands the nuance and context of our conversation. Generative AI in particular has taken artificial intelligence from a lofty ideal used only in difficult to access systems to a practical tool that businesses can use for tangible benefits today.
So, you would have to be living under a rock to have missed the hype around AI in the past year. From writing e-mails and articles, to creating art with just a few prompts, AI has been hailed as the greatest productivity booster since the Industrial Revolution or a sign of impending human doom.
For market researchers, we have to ask ourselves, is AI friend or foe?
A recent Qualtrics study of 250 market researchers found that 93% of researchers see AI as an industry opportunity with 80% saying AI will have a positive impact on the market research landscape.
Now, I am fully in this camp, and I think it is especially true for qualitative research.
No one wants to give up having real conversations with real consumers, especially as we battle bots and fraud in qualitative research and quantitative research. But the dilemma has always been that all of that rich qualitative data takes a long time to wade through and analyze, and even then you're always worried that you might've missed something.
That's where AI comes in, not as a replacement for you and your years of experience, but as a tool to help you understand and interrogate your data.
It's time to dive deeper into what's possible with AI powered tools that are available today, just 20 months after generative AI burst onto the scene. Let's explore how it can accelerate your understanding of consumers on a human level.
I'll do that today by showing you how I've been using AI to deliver research projects for insights teams that work with Voxpopme services.
When and how can you use AI in qualitative research? Truly at every stage of the journey.
Now, I love writing surveys and I'm never going to want AI to do it for me completely, but there are days when you're staring at a blank screen and you need a prompt or you want AI to fill out a basic discussion guide template for you so you can go in and customize it, or your stakeholder gave you a question in three parts that uses all sorts of insider jargon. You need to make it consumer friendly.
All of those are great uses of AI in the survey and discussion guide design process. But where I think AI really shines is in the analysis phase.
Again, you might have hours of interviews or hundreds of short response videos to go through, and AI can be an incredibly useful assistant to help you understand and analyze your responses and put them into a comprehensive way of understanding them for you and your client.
And last but not least, AI is going to help us share our findings with our internal stakeholders, helping us build showreels and reports in less time.
Now, we might've seen some of the analysis and show real tools, but I'd love to take a moment to show you how AI can accelerate the creation of questions for both live interviews and video surveys.
We just use these prompts, and I'll show you how you can fast track discussion guide and video survey creation with just a few prompts.
So, here we're going to use the objective in the discussion guide generator. You can see, you just put your prompt, what you want to learn from your interview into that generator, tell it how long you want the interview to last and then just click on ‘discussion guide generator.”
It's going to generate an introduction, questions in different parts of the interview with timings on them and then some closing information as well.
Now, the nice part about this is again, you can go in and edit these questions to either change them completely or just maybe add a few probes that you really want to ask your respondents, and then you can download that guide and you're done.
Amazing. Let's focus in on AI's role in transforming the analysis of qualitative research. Let's zoom in and see what transformative effects we see AI having.
First, we see AI replace traditional time intensive processes that underpin analysis. You no longer need to manually watch, read and code qualitative data. AI can interpret your feedback, analyze transcriptions and give you a first pass at that analysis. And after that first pass, it can augment your work.
As you begin a deeper analysis, it'll present you with summaries, key themes and recommendations backed with supporting evidence that you can use to verify the work of the AI and AI will even generate suggested further research based on findings. And this is where you can really start to feel AI making your analysis process much more efficient.
And finally, AI transforms qualitative analysis beyond anything we've experienced by making it possible to interact with and ask questions about your data to reveal more nuanced insights or to ask probing questions that connect your hypotheses. And that works on single projects or across entire repositories of qualitative data, which is a tremendous advancement and tool.
At this stage, it really brings the power of people and technology together to make qualitative possible at quantitative stales.
Now let's look at exactly what that means inside the Voxpopme portal using a coffee project. The AI analysis of the transcript highlights three main areas of focus.
The first being personal experience and service comparison of major chains like Starbucks and Costa and the personalization of coffee orders. Here we get a breakdown of the first key theme.
We've used AI to identify that customers appreciate personalized coffee experiences, and the AI has also provided video clips to back up these findings. You can click to preview each snippet of the video.
Now, if you wanted to quickly create a show reel based on this theme, you could add all of these clips to a show reel, or if you just wanted to add a couple of these clips into your show reel and not the whole set, you can click on the plus icon.
Here we can see a respondent talking about their customer service and personalizing their experience. We wanted more context around this response. We can click this icon here, and this will show you the quote within the transcript.
Now at the bottom there, you're going to see the AI will also provide some suggestions for follow-up questions. These are questions that you can further ask to expand or clarify information that you've been given.
Now we know the incredible value AI can add to our analysis process, and I want to ensure you know how to get the best results from your new research assistant. And that's very much how I think of AI, as a research assistant.
I've been using AI to assist the delivery of analysis projects in the new services division of Voxpopme. So, we've learned some common missteps in how to avoid them, and we also have some key takeaways to remember as well as what we've seen already today.
AI is incredible, but luckily for us it is some time away from being our all-knowing robot overlord. It has limitations. So be diligent.
If you've used ChatGPT or any other popular generative AI, you'll know it can occasionally miss context or possibly even hallucinate. One of the reasons it does this is because it is formulating its point of view from a vast general database.
Voxpopme’s integration of ChatGPT limits the dataset analyzed to your specific research. This minimizes the likelihood of hallucination because it has a very focused set of data to pull from. But remember, you're still the domain expert and I definitely recommend paying attention and validating its findings.
How is that helpful?
Well, recently I was using AI to help identify some themes and find some key quotes for a project I was working on. And as I did so, I realized that some of my quotes around sustainability and my ecological concerns section of my report were actually about price sustainability and not about green energy.
That's an important distinction. And without the supporting quotes provided, I could have overstated how important the ecological factors were in this study.
So, I was able to of course, be a little smug that my research expertise was still very much needed. But the best part is that because our generative AI provides those snippets, provides those clips of backup information, I was able to dig into the evidence and understand what our consumers were saying.
Another important use case is interrogating your qualitative data. There are many uses for this of course, and as you delve deeper into your themes, you might want to see if anyone mentioned a specific topic or area. Old school searches would only bring up the very specific words you searched on. But with AI chat, you can ask questions more broadly and get deeper insights.
Here are some useful prompts that we have found work very well in qualitative research.
So, for example, ‘I'm trying to answer this business question. What are the habits of millennials that buy ship to home meal plan services? Write me a summary of this that addresses that question.’
I love formatting my questions for AI chat in a way that tells it how I want the answer presented, whether that's a summary, an executive summary or a bulleted list, think about how it would be most helpful for you to see that information.
I also love how it can save searching for something.
So, example, ‘give me an examples of respondents talking about X.’
I love this because it saves you those hours of searching for something, especially if maybe there was a minor point in a research project that you wanted to highlight. And then on your first go through, you didn't pull any quotes from people saying that. This is a great way of saving those hours of searching.
Last but not least, I would be remiss if I didn't mention AI's role in crafting your deliverables. With each round of analysis, AI will present you with video evidence to support your findings.
So, you can see here I've asked AI a question and it's giving me a text response based on its interpretation of my data. But crucially, it's given me these clips of real people telling stories that support those insights. And when I'm creating a deliverable for insights teams, I can simply add those clips to a show reel and build a show reel of AI sourced insights verified by me that's ready to share with colleagues.
We've been through a lot today, but if I had some key takeaways, it would be this.
Scale your research. You can now think beyond what was previously possible with qualitative research.
AI opens doors that had been shut because it took too much time and attention to do. You can fast track every stage and free up your time to offer your unique skill and expertise to the process and let AI do some of the wading through all of those interviews.
And again, think of AI as your research assistant. You are still the domain expert, but AI is here to help you move faster and make it easier on you.
Now, if you're not comfortable doing all of this yourself, we love talking about research, whether you just want to talk about what the AI capabilities are, you have a research project planned or you don't have a research project planned, and you just want to discuss what's possible. We are here.
Schedule an informational meeting to discuss your research agenda and how Voxpopme can make it easier for you. If you do have a project in mind, we can map it out together.
So, take a moment, scan that QR code and set up a consultation and we'll discuss more.