Fabric AI x ChatGPT: Exploring shifting attitudes

Using Fabric AI + ChatGPT to Explore Shifting Attitudes Toward AI

Full article and case study report coming soon - but in the meantime feel free to check out the brief summary, AI walkthrough videos, and raw response data below!

Click the link here to view the “Shifting Attitudes Toward AI” case study, with Fabric AI 2.0 + ChatGPT analysis on the Fabric.is researcher platform. To see the AI output, just click the “View Fabric AI” button beneath any questions that spark your interest!

You can view the video below for an introduction to Fabric AI, the primary technology featured in this case study!

For this case study, we set out to take a close look at where general perceptions of AI are currently at, how they may have shifted during this recent generative AI tech revolution, and where people see AI going in the near future. Toward this goal, we actually employed ChatGPT to write and edit a number of the study questions, then used a combination of our proprietary Fabric AI alongside new integrations of OpenAI’s ChatGPT API to spearhead a quick-hit analysis on a relatively large qualitative video research study that normally would require a significant number of hours to code and analyze. The methodology we used is the Fabric DIY “Let Us Recruit” pathway, where we built up the study on Fabric, submitted a request for the recruit spec (two age segments of 18-29 and 40-59, then two occupational segments of Tech vs Non-Tech industries, plus an awareness and relatively strong opinion on ChatGPT / generative AI), and our on-platform recruiting team ran screeners to recruit from the Fabric / mindswarms respondent database. The N=60 respondent study closed in just 3 days, with strong response quality and insights.

The full findings of this case study will be coming soon as a complete article, but in the meantime you can view the video snippet below to see a walkthrough of using the AI to pull out some insight overviews from a couple more AI-central questions in this case study!

The Power of Emotion: Fabric Philosophy

The Power of Emotion: Fabric Philosophy

Start Here

7 min read

We firmly believe the best consumer insights are the ones rooted in an emotional understanding of people. We would argue that the vast majority of human motivation, behavior, attitude, belonging and ultimately, decision-making, is driven by emotion. 

The “why” is absolutely important as table stakes for any qualitative market research study. However, the most successful brands, ad campaigns and product designs connect with people on a deeper emotional level.

That’s why we focus on video; it’s the most emotional of all media. It allows people to express themselves more authentically, fully and genuinely than they can using the written word. Video captures tone of voice and shows facial expressions, enabling the researcher to connect human-to-human with the respondent. Video is also the most data-rich research artifact, encompassing video, audio and transcripts.

Emotion-Based Proprietary AI

Emotion is the driver motivating us to develop our own proprietary AI. Our AI is specifically programmed to be responsive to sentiment and emotion. We purpose-built our AI from scratch to be a simple, elegant ‘research assistant’ that helps researchers work faster. Our AI assistant speeds up qualitative data analysis by identifying the most emotionally engaged videos and quotes. It also tracks eight primary emotions.

Emotional Insights in the Marketplace

Our founder, Tom Bassett, has worked in advertising and product design for some of the most valuable brands in the world. This work showed clearly the importance of emotional insights in successful brands. Here are a few examples of insights he helped identify:

Olympic athletes have an intensity and near-pathological desire for victory.

This emotional insight led to the industry-famous creative brief: How do we communicate the idea that to the Nike athlete, “Sport is war minus the killing?” The fuller story is told by John C Jay of Nike’s ad agency, Wieden & Kennedy, in “Briefly,” a documentary film about the creative brief.

Working with ultrasound technology makes doctors feel like superheroes.

Despite the environment of anxiety that physicians say surround their work—the possibility of malpractice suits, digital record keeping, new technologies—we found that they all experience personal superhero moments working with ultrasound, which gives them a massive surge of confidence. This emotional insight led to award-winning product design for SonoSite, the #1 brand of portable ultrasound.

The Funnel: Structuring Research Studies

The Funnel: Structuring Research Studies

Start Here

7 min read

Goal vs. Pre-set Process

When developing a research study, there’s an interesting tension between 1) setting out to find original insights and 2) following a pre-set process. Our general recommendation is to remain focused on the goal rather than a specific process. However, if you’re newish to insights, a previously established process may be the best place to start.

Designing Your Research Study: The Funnel

A conceptual approach we recommend for developing research studies is what we call “the funnel.”

It’s important to start broad, and then narrow down in order to help frame the context for consumers—like the shape of a funnel. For example, we do not recommend starting at, say, the feature level for a product or a specific part of an ad campaign. Instead, begin with larger subjects.

Here is our 8-point framework to help you think through how to design your study.

1. Culture

Suppose your study is about launching a new car model. Begin by exploring: What’s happening in the culture more broadly? If you start with a question or two on their views about where the culture of cars seems to be, you might unlock respondents’ views on ride sharing, the role of electric cars, personal safety, image and more. These questions allow the respondent to weigh in with their opinion on the larger culture, without necessarily applying it to their personal situation yet.

Sample questions:
What’s happening in the culture of online dating? What’s your personal impression of car culture at the moment? What’s the word on the street in sneaker culture?

2. Category

What are the dynamics shaping the category your product operates in—from the POV of the consumer? By understanding the category level, you begin to help shape feedback, but not at the brand or specific product model level yet. These questions locate the respondent in relation to the whole category, not necessarily a particular brand.

Sample questions:
How has the role of cars changed in your life in the past three years? What are your views on home ownership now that interest rates have changed? What are the most important things to understand about you in terms of how you manage your personal finances?

3. Brand Landscape

How do consumers view the differences amongst brands in the landscape? Understanding how different brands are perceived helps frame the broader context for how they view the overall brand landscape. 

As we narrow the funnel here, we learn about respondents’ awareness, imagery, associations, preference, and purchases of brands. We also learn the different Jobs to be Done that respondents associate with each brand. Clay Christensen explains eloquently how the Job To Be Done for the McDonald’s milkshake (at breakfast time) was to help kill time in the morning commute (versus provides some sort of protein-based nutritional supplement).

Often, if the researcher does desk research, they’ll assess the brand landscape based on what the brand itself is trying to stand for; but chances are the consumer doesn’t see the brand landscape the way marketers attempt to position it. These consumer-based perceptual differences provide valuable insights for researchers.

Sample questions:
 Is—or how is—the Lyft brand fundamentally different from the Uber brand in your opinion? In your opinion, what are the most authentic brands in golf and what makes them authentic to you? Are smartphone brands pretty much all the same? Elaborate.

4. Product Landscape

What are the perceived product differences across your key competitors? Knowing how consumers relate to specific product-level differences—whether overall difference between two products, or more precise feature-level differences—helps you understand how specific products are differentiated in the mind of the consumer. 

Sometimes it turns out the consumer sees no difference at all. Sometimes subtle yet important product-level differences can create dramatic points of differentiation. Without packaging, for example, Dawn looks a lot like any other dishwashing liquid. But wrap the “grease cutting” emphasis around it and suddenly, it’s very different from Dove or Palmolive.

Sample questions:
When it comes to online cloud storage, which brand has the most relevant product features for you/why? What’s the biggest difference between the feature set of brand X and the feature set of brand Y for you? What are the most important product features of a home internet plan for you?

5. Design Landscape

Are there significant perceived differences in how design shapes the brand or product experience? Many brands—and VCs—recognize the power of good design in an intensely competitive environment where color, format, style, ease of use and aesthetics become major drivers of retention.

Sample questions:
How is the design of the Caviar app different from any other online food delivery app? What could be improved in the user experience of the Nike app for ordering new products? Does the design of an external battery for your electronic peripherals (phone, wireless earbuds) matter/why?

6. Advertising Landscape

How do consumers view the overall advertising landscape? Arguably, this question was a lot easier for people to answer when there weren’t so many different forms of media. It’s still important to understand the types of advertising people actually enjoy, and how they see the predictable patterns that many brands fall into. Note: most people will tell you that advertising doesn’t ‘work’ on them, so you have to be clever not to set them up to tell you that.

Sample questions:
What is the most impactful online ad you remember; what made it so memorable? When it comes to advertising for wireless phone providers, what annoys you most? Has a pop-up ad ever been helpful? Tell us about your favorite Super Bowl commercial of all time.

7. Your brand/product/ad

Now the funnel narrows down. Here is where the consumer gets to unpack all of the ways they view your offering. It could be on the level of brand, product, advertising, PR, employee, values, corporate behavior, executive leadership, sustainability, stock price, retail environment and more. 

Just be mindful that you don’t necessarily have to be loved universally for consumers to actually buy your product. Great brands are more frequently taking a stand on issues; in an era when social media is a powerful voice for the average person, taking a stand can cause flak. The question companies ask themselves is what do people actually BUY at the end of the day? 

An example is Nike’s ad featuring Colin Kaepernick. Some people burned their Nikes, but online sales jumped 30% in the following weeks.

Sample questions:
How is Polestar different from any other electric vehicle brand? How does our CEO’s recent press release resonate with you personally? When it comes to sustainability, how does our brand stack up against the best sustainability brands in the world? How does our download function compare with YouTube’s?

8. What is the opportunity?

Making your way through the funnel, look for the opportunity. What is the connective tissue that creates a unique opportunity for your brand, product, or ad? It could come from any level in the funnel: Culture (people want more beautifully designed products in their homes: Nest), Category (5,000 songs in your pocket helped a generation of iPod buyers understand what an MP3 player was), Brand (Blue Bottle coffee found a niche within the premium coffee landscape), Product (Peloton designed a phenomenal biking experience for home fitness), etc.

Following of the Funnel Framework
By following the funnel structure, researchers gain in two ways. They gain knowledge of the specific questions they set out wanting to know (How do people feel about this new feature? How do they perceive our brand and how might we change that?). They also develop a larger understanding of consumers and the current landscape, holistically and contextually.


POV: Qualitative Sample Sizes


The purpose of this content is to provide a point of view on the ideal sample size for a qualitative market research study, because that’s one of the most common questions even the most seasoned researcher asks us.

Industry norms

Adequate sample sizes have been debated extensively in the market research industry for many years, in part because quantitative sample sizes are statistically easier to measure. Often, quantitative research frameworks get applied to qualitative research because quant norms are concrete (e.g +/- variance), whereas qualitative sample sizes are difficult to prove, mathematically.

A number of studies have been published over the years on the topic of adequate sample sizes for qualitative research. Here are a few:

    • Creswell, Glaser, Morse – all pretty much agree that sample sizes of 30-50 respondents reaches a point of saturation, where adding more respondents does not significantly alter the findings
    • Springer – Springer puts forth the argument that anywhere from 5-50 is adequate, but that 25-30 is considered to be the right number
    • InterQ – recommends 20-30, or even as few as 10

Fabric POV

Qualitative research is more art than science. Bill Bowerman (Nike co-founder) tested his shoe designs on a relatively small sample size, yet was able to scale the innovation to millions of people, in large part because it was the right sample of runners.

Typically, clients turn to qualitative methods to understand meaning – the How and the Why. 

So while we believe there are minimum sample sizes that should be employed in qualitative research, there are some critical variables in the equation we consider:

    • where you are in the process; in our opinion, earlier stage projects (e.g. exploratory product innovation) can employ smaller sample sizes, because more iteration and development will be conducted as the project moves along
    • the business impact: we feel that the higher the business impact of the research, the greater the sampling (qualitative and quantitative combined) should be
    • geography: it can be as easy as domestic and international, but one critical question we consider is how broad the target audience is, geographically
    • research design: how the study is designed can have a huge impact on results (e.g. which questions to ask in what order)
    • research vendor: when you buy a qualitative research resource, you are buying a filter (like a water or air filter). One important variable is how equipped that resource is, down to the people, to field the study and interpret the results
    • quality of recruit: back to the Bowerman example, it’s very important that the people in the study be the right people; beyond technically whether they qualify, we ask ourselves whether the people we interview “feel” like the right consumer. And we will over-index the input of those whom we deem more influential (for example, a starter on an AAU basketball team in Brooklyn may be more insightful than a practice player on a D3 team in the suburbs)
    • analysis: who is interpreting the data is highly important. 
    • methodology: there are a number of qualitative techniques to choose from, including IDIs (individual interviews), focus groups, in-homes, friendship pairs, small group interviews, intercepts, observational research, ethnographies, and digital (online, mobile). In Fabric’s case, the methodology is unique in that it’s effectively “1-on-none” – meaning it’s asynchronous, and there is no moderator present. A lot has been written about the effects of group-think in focus groups, where an ‘alpha’ respondent will influence others. Similarly, the confessional style of the Fabric methodology enables what researchers have called the “online disinhibition effect” where respondents are more open to express themselves because there is no fear of disagreement or conflict with a moderator. For consumers, communicating asynchronously via video is a very comfortable and common medium

Fabric POV

Our short answer when clients ask us about the right sample size is to go with 30-50 respondents, but the context as outlined in the “Fabric POV” section above guides us beyond.

Follow up questions are welcome: please email tom@fabric.is.


4 Tips for Getting the Most out of Your Fabric Video

What’s the best way to go through video responses and glean insights? Tom Bassett, founder of Fabric & mindswarms, discusses best practices for extracting relevant data and weaving powerful stories to share within your organization.

Mobile video surveys provide powerful first-person accounts that speak customer truths. But how do you get the most from your data? Using his more than 20 years of experience working with companies like Microsoft, Virgin America, and Nike, Tom Bassett shares 4 tips to help you effectively analyze your data and draw conclusions.

Reviewing Results

When analyzing your mobile video surveys, there are a couple of things to look for. First, you’ll want to identify patterns and themes. What are people identifying with? What are the recurring problems or issues with the product, service, or experience? Come up with a list, aiming for no more than 5-10 patterns. Once you’ve identified these, it’s important to run them through your brand filter. Ask yourself: How are the insights related to my brand? Which are the most relevant for you (versus insights that are not related to my brand)?

During this stage, it’s also important to look for original insights – outliers. For example, what’s something you didn’t expect the respondents to say? What’s something unique you noticed, when seen through the eyes of the consumer? Since it’s very hard for brands to differentiate these days, the off-speed pitch is often where you find the interesting angles.

Organizing Results

Next, you’ll want to bracket your insights into two main buckets: problems and opportunities. This is a relevant way to share insights back internally at your organization. Senior leadership teams often want to help solve problems – but they also want to understand where the potential zones of opportunity are to help grow. Are the respondents articulating a problem or need that hasn’t been met? Have they mentioned a totally new idea? An expansion of idea?

Identifying Story

This is an important step because although it’s great to have lots of insight, it’s essential to find a focal point. Ask yourself: What is the overarching story? How do you articulate this? Try to hone in and articulate one story. Philosophically, we like to see things through the eyes of consumers. For example, for a Yahoo Personals project: “Women don’t want to let go of the idea that Fate played a role in finding the partner of their dreams.” Once you determine THE story, you’ll want to come up with chapters, or building blocks that help build that story. Ask yourself: What are we trying to teach or tell people? What’s the big reveal? In simple terms, there is a beginning, middle and end. The middle is usually the reveal (the point of tension, the climax) while the beginning introduces it, and the end wraps it up.

Sharing Results

When you give your presentation to colleagues, use the videos to really engage them! Videos lift heads, because there is sight and sound and motion. So, make sure to show video clips of respondents in order to get their attention. It’s worth it to put together a 2-3 minute series of clips in order to share the data in the most powerful way possible. But don’t use video to be a surrogate for PowerPoint; use video to tell a compelling story. Otherwise, if video just lists points as opposed to adding up to something singular, viewers will be confused. (One additional tidbit: we also like to sprinkle in single Fabric clips throughout a debrief deck, to help keep things lively. These clips can be links to the study matrix, or videos can be downloaded and placed into the presentation).


4 ways Fabric works with traditional qualitative research

Recruiting for ethnography.

Rather than show up at a respondent’s home with an Excel spreadsheet and fingers crossed the participant is good, Fabric has been used to recruit in advance for ethnographies. Not only does it help identify cream-of-the-crop candidates, but it also informs the research process early on, and can be used to sharpen discussion guides.

Pre-in field.

What we have often done is a wave of Fabric research in advance of doing any kind of field work (focus groups, IDIs, in-homes). A lot of clients are anxious to get projects moving quickly, and instead of waiting two weeks to begin traditional qual research, Fabric can gather insight within days from a very broad geography to begin to inform the project.

Post in-field.

Fabric has been used as a way to bounce ideas off consumers after the traditional qualitative field work has been completed. So, for example, on a project with product designers, sketches of ideas inspired on the road were bounced off Fabric respondents after the initial wave of research had been completed to help validate a direction.

Extending reach of focus groups.

In most countries, the tendency is to conduct traditional qualitative research in major markets because that’s where the lion’s share of volume comes from. But Fabric has been used in a remarkably quick way to access consumers in every level of markets to help balance out a more urban skew to the recruitment. So, for example, in the US, rather than hit the usual NY/LA/Chicago markets, sprinkling in a nationally representative wave of Fabric recruits helps provide a more representative picture of consumers nationally.