marketresearch

Using AI for better research

While people are wondering what potential AI holds for the future, many of us are seeing benefits right now, especially when it comes to our work lives.  That’s certainly the case in the market research field.  AI has enabled us to streamline the process of finding research participants, it can help with discussion guides and moderation, and it has significantly improved report-writing capabilities.

But before we get to conducting the research, AI can help make sure we are asking the right questions of the right people. Think of AI as an opportunity to discuss your client’s research question with an expert in that specific field… actually, not one expert, but an aggregation of experts!

For example, a client recently came to us looking to understand the potential appeal of a new product in the jewelry category. They wanted to know if people would be interested in the product, which customer segments would be most interested, how they go about shopping, and how they would most likely find out about the product.

We needed some background about the category in order to decide whom to recruit and what to ask them. So I turned to AI and asked the following:

    • Act as an expert on the U.S. jewelry business. Provide an in-depth analysis of the industry, focusing on the mid-range pricing tier. Your analysis should cover the following aspects:
    • Provide a detailed description of how the jewelry category operates in the U.S., including key characteristics and market structure.
    • Focus specifically on the mid-range pricing tier. Discuss the types of jewelry typically found in this range and the defining features that differentiate this tier from budget and luxury categories.
    • Analyze the different segments of customers within the mid-range tier. Pay special attention to purchasing behavior, particularly the distinctions between customers buying for themselves and those buying as gifts.
    • Identify and explain the current trends impacting the mid-range jewelry market. Include insights on consumer preferences, popular styles, and emerging materials or technologies.
    • Explore the various business models within the mid-range jewelry market. Discuss the roles of brick-and-mortar stores, online retailers, and hybrid approaches.
    • Provide examples of key players in the mid-range jewelry market. Highlight a few prominent brands or retailers, and discuss their market strategies and positioning.

    I received a great, detailed analysis. I then followed up with questions regarding the specific product idea, as well as the typical steps in the customer journey. The input helped us design the research to make sure we were spending our client’s research dollars in the most effective way possible.

    How did I learn to write that kind of prompt? A lot has been said about prompt writing (for example, this information from OpenAI), but I also want to credit my friend and colleague Robin Algaze who spearheads an AI Community of Practice at Ignite 360 that leverages generative AI to enhance team productivity and elevate output. She provided the following template for AI prompts, which is a great starting point:

    • Act as a [specific role with expertise in the subject].
    • I need a [precise description of the deliverable].
    • You will [clearly defined task with a singular goal].
    • In the process, you should [step-by-step details or important considerations].
    • Please avoid [clear exclusions, sensitive topics, or undesired approaches].
    • Present the final result in [appropriate format for the content].
    • Here is an example [relevant and clear example to illustrate the request].

    I encourage everyone to experiment with AI prompts. The more you do it, the more you’ll find ways in which AI can benefit you!

    How can we help you conduct more effective research? Email me at info bureauwest.com and let’s discuss!

    Sources: “Prompt Engineering,” OpenAI; Robin Algaze, Ignite 360

    Posted on by Jay Zaltzman

    Getting accurate answers to research questions

    I’m excited to have been chosen as one of the industry experts presenting the Qualitative Research Consultants Association (QRCA) course series “From Start to Finish: Managing Qualitative Research with Confidence.”  I will be presenting the training on best practices for developing/evaluating discussion guides.

    Discussion guides are where the rubber meets the road when it comes to getting accurate answers to research questions.  Some clients may not realize that simply asking the research questions in order of importance usually will not get us accurate answers.  Some reasons for this:   

    • People might initially feel uncomfortable when talking to a stranger.
    • If we start by asking about specific ideas, we might bias people, and then we don’t get an accurate read about their general thoughts and feelings.
    • People have a tendency to tell us what they think they should say, rather than what they really think.

    For example, imagine a company developing a new luxury travel product.  Their main objectives are to find out if the product appeals to their target audience, if the cost seems reasonable, and how they feel about specific elements of the offering.  If we were to conduct focus groups where we started by presenting a description of the product, we would run the risk of receiving inaccurate input.

    Instead, I would recommend starting with a general discussion about how participants approach travel.  This would help people feel comfortable, and they would realize that the others in the group are people who have similar travel experiences (so, for example, they wouldn’t need to worry about seeming arrogant in front someone who is more interested in budget travel).

    Then we would want to understand how people make the purchase decision for a luxury travel product like the one being researched.  We know that people think they make their decisions completely rationally, by weighing the benefits versus the price, but we also know that’s not the case: emotional factors play an important role in the decision.  So we might utilize projective techniques to get at that emotional aspect (for example, have them imagine two friends discussing this potential purchase and write down what each of them says; then write what each of them is really thinking).

    Only after that would we actually expose the concept of the new product.  We might ask participants to write down their initial responses (to get their gut reaction before they hear others’ reactions).  When we discuss the concept, we might encourage participants to have conversations with each other, where one likes the concept and one doesn’t and ask them to persuade each other.  While some people worry about the group dynamic in focus groups, these kinds of conversations happen in the real world all the time; by listening in, we can learn what elements are likely to sway people in favor of our concept and what issues might need to be addressed.

    Those are just some examples; I encourage you to sign up for the course for a more detailed look.

    And before I go, one more tip: I like to include a final question in the discussion guide, asking participants to write a postcard to the company with their advice to the CEO on the topic discussed.  This is an opportunity for any final thoughts, and also lets people express any thoughts they might not have been comfortable saying out loud.

    How can you get accurate answers to your research questions?  Email me at info at bureauwest.com and let’s discuss!

    Posted on by Jay Zaltzman

    How to deal with low consumer sentiment

    The economy has been improving, yet consumer sentiment remains surprisingly low: consumers think inflation is worse than it actually is, and they are worried about their future.  Younger people in particular are worried.

    A recent McKinsey study on holiday shopping plans found:

    • Consumers are trading down – buying fewer and less expensive items.
    • They are looking for better prices and promotions (even more than usual).
    • They are less likely to splurge.

    (One interesting twist: many Gen Z consumers plan to pay extra for next-day or same-day delivery.)

    Here are some of McKinsey’s recommendations for marketers (which I think will apply beyond the holiday shopping season):

    • Inspire beyond promotions. As consumers grapple with whether to splurge or save this year, retailers have an opportunity to inspire their customers through storytelling. Focusing on the experiences that come with a purchase—such as crafting for the holidays, setting a beautiful table for family gatherings, or indulging in self-care—may motivate purchases more effectively than relying solely on discounts. To achieve this, retailers should tailor their communications by channel. For example, video content on social media that highlights the benefits of a product may resonate most deeply with consumers who are comparison shopping.
    • See demand generation through the “ninth inning.” As shopping journeys are becoming increasingly omnichannel, retailers should meet customers where they are to capture them throughout the holiday shopping season. Retailers are using new technologies, such as AI, and communities, such as store employees and micro-influencers, to deploy curated and personalized content online and at scale. In-store experiences can be bolstered by digital enhancements, such as augmented reality apps and cashierless shopping. Meanwhile, live stream shopping is also gaining traction, especially with younger consumers.
    • Personalize promotions for consumers. Leading retailers focus on targeted pricing and personalized offers to deepen customer loyalty, based on a comprehensive view of individual customers. This requires having a baseline understanding of when and what consumers buy and their sensitivity to price and promotions. As the holiday shopping window extends over a longer time period this season, a focus on driving customer loyalty is key. Targeted and personalized offers may also enable retailers to drive inventory sell-through while capturing incremental sales and profit.

    Let’s find out what will resonate with your customers. Email me at info at bureauwest.com and we can discuss the best approach.

    Sources: “US holiday shopping 2023: Consumer caution and retailer resilience,” McKinsey, 11/6/23; “Why are we so bummed about the economy?,” NPR Planet Money, 12/1/23

    Posted on by Jay Zaltzman

    Do better research

    I just got back from the QRCA Worldwide Qualitative Research Conference in Lisbon – there was a lot of great content packed into 2.5 days, as well as a dinner at the amazing Palacio Conde d’Obidos, shown here.

    It occurred to me that we were all there for the same reason: to learn ways to do better research.  And I think we did!  Here are a few of the highlights for me:

    Lucy Foylan gave a great presentation about the differences between conducting research online and in-person.  Her agency, The Nursery in the UK, compared the two and they found the people were more likely to work to build consensus during in-person focus groups and more willing to disagree with each other during webcam groups.  While some might think that’s a reason to conduct all focus groups online, remember that consensus building also happens in real life.  Witnessing how participants persuade one another can provide valuable insights for our clients.  Depending on the objectives of the research, we might benefit from in-person groups, webcam groups, or a mix of both – where we examine the differences between the two.

    There were several sessions about the impact of AI on qualitative research, including presentations by Daniel Berkal and Sidi Lemine, followed by a panel discussion which I moderated, with Simon Shaw, Tom Woodnut and Paul Kingsley-Smith.  Some of my takeaways:

    • Daniel talked about ways AI can be used so we can do our work better and more efficiently.  He uses Chat GPT to help with screener development, with ideas for discussion guides, and to summarize responses, and Adobe Firefly to create images for proposals and reports.
    • Sidi talked about using AI tools to recognize emotions in research participants and how they’re surprisingly accurate across cultures.  While a smile or a frown may mean different things in different cultures, it turns out micro-expressions are remarkably consistent throughout the world.  Specifically, Sidi said he likes the following tools: Phebi.ai, Emozo, Immersion.
    • While there are many great ways AI can help us in our work, our panel participants focused on what AI can’t do, and why we researchers are still needed.  One example: in a recent focus group project, participants all said they liked one of three concepts best, but I realized that was because it was the shortest concept, not because of the content of the concept.  If we had relied on AI to conduct the research, it would have taken those responses at face value and not probed further.  Simon said that we qualitative researchers are too humble and don’t do enough to explain the value we bring.  I agree!

    Those are just some of the highlights.  The Worldwide Conference reminded me of how important it is for us to keep learning and adding to our skills.  The next opportunity is coming up soon: QRCA’s annual conference will take place in Denver, January 22-25, 2024.  I recommend it!  Register here: https://www.qrca.org/event/2024-annual-conference .

    How can we add value to your next research project?  Email me at info at bureauwest.com and let’s discuss!

    Sources: QRCA 2023 Worldwide Qualitative Research Conference: “A Hybrid Future: Exploring Human Interactions On- and Off-line,” Lucy Foylan; “Navigating Qual in the Age of AI,” Daniel Berkal; “Can Emotion AI Remove Bias in Global Research?,” Sidi Lemine; “What AI Can – And Can’t – Do For Qual,” Jay Zaltzman, Simon Shaw, Paul Kingsley-Smith, Tom Woodnut

    Posted on by Jay Zaltzman