Decoding Human Emotions for Deeper Consumer Insights

Quantitative market research agency Bigeye’s podcast features Lana Novikova of Heartbeat, an AI-driven tool that decodes human emotions from textual data.

IN CLEAR FOCUS: Emotions are central to human decision-making, but research has traditionally lacked the tools to accurately capture and assess them. Building on her experience as a quantitative researcher and studies in neuroscience, Lana Novikova has developed a tool that decodes human emotions from unstructured data. In this week’s podcast, we hear how Heartbeat AI provides marketing researchers with unique insights that can be applied to the creative development of advertising campaigns.

https://oembed.libsyn.com/embed?item_id=17647250

Episode Transcript

Adrian Tennant: In today’s episode of IN CLEAR FOCUS…

Lana Novikova: We market researchers have responsibility to understand our consumers and shoppers at the deepest possible level. And I wanted to build a tool that can reflect the nonbinary nature of human emotions. Heartbeat is all about emotions.

Adrian Tennant: You’re listening to IN CLEAR FOCUS, fresh perspectives on the business of advertising, produced weekly by Bigeye. Hello. I’m your host, Adrian Tennant, VP of Insights at Bigeye. An audience-focused, creative-driven, full-service advertising agency, we’re based in Orlando, Florida, but serve clients across the United States and beyond.

Thank you for joining us. Many of us like to believe that as consumers, the buying choices we make are rational – based on reviewing what’s available and considering the alternatives. The reality however, is a little different. Many neuroscientists and psychologists have concluded that emotion is a necessary ingredient to almost all purchasing decisions. Emotional responses to advertisements strongly influence consumers’  reported intent to buy a product. Studies conducted by the industry have consistently found that likeability is the measure most predictive of whether an advertisement will increase a brand’s sales. Because emotions can greatly influence or determine our buying decisions, researchers use specialized tools that attempt to identify consumers’ often unconscious feelings, for example, when exposed to TV advertisements, or while shopping in a store. Our guest today is an internationally renowned research expert who has published papers on the role of emotion in consumer behavior. Lana Novikova is the Founder and CEO of Heartbeat AI Technologies. Based in Toronto, Canada, the company’s text analytics platform measures emotions to understand the customer experience and identify the drivers of human behavior. Born and raised in the former Soviet union Republic of Kyrgyzstan and educated in the US, Lana’s career has taken her from the United Nations field office in Central Asia, to UNICEF in New York, and on to corporate market research and analytics, including several years with Nestle. Lana has designed, executed, and managed research studies for dozens of clients in a wide range of industries. She’s also a serial entrepreneur with an award-winning portfolio of research inventions. Today. Lana is joining us from her office in Nicaragua. Lana, welcome to IN CLEAR FOCUS!

Lana Novikova: Thank you so much, Adrian – thank you for inviting me to be a part of this podcast. I’m honored and excited.

Adrian Tennant: Well first, as I mentioned in the introduction, your company is based in Toronto, but you’re currently in Nicaragua. Why are you there?

Lana Novikova: Interesting question. You also mentioned that most decisions that we make in life are not rational, so I guess it’s a result of one of those emotional decisions! I came to Nicaragua exactly a year ago with my  daughter for a meditation and yoga retreat for just a couple of weeks to unwind. And that was my first vacation after five years of running a technology company. So I came here and I really fell in love with the country, and, come March. I decided to come for a longer time. So I came with my dogs and my kids and, lo and behold, COVID happened. And I made a decision to stay, instead of going back to Canada. And the longer I stayed, the more in love I got with the country. So I ended up buying a piece of land and hopefully, we’ll build a dream eco village here in this country, while still traveling to Canada a lot, traveling around the world, when the travel happens. So it’s an emotional decision.

Adrian Tennant: Well conducting background research about you for this podcast, I read some of your published papers and it’s clear that you’ve been interested in interpreting human emotions in the context of market research for quite some time. But what first sparked your interest in consumer research?

Lana Novikova: I have an educational background in linguistics and my master’s degree is in journalism and public relations. I wanted to work in PR and my first job ended up being in research. And then I discovered my natural inclination to be an analyst, and I went back to study statistics and market research methods and became a very strong quantitative researcher. We’re talking 2000, that’s when the data was still collected using pen and paper surveys and online surveys just were appearing as a new technology thing. So that was my journey in market research: six years in quantitative market research I really learned how to ask good questions, good, closed-ended questions. And then one unanswered was the “What if we ask open-ended questions in surveys? Can we learn more about consumer behavior by asking the combination of questions?”

Adrian Tennant: You have described your company, Heartbeat AI Technologies, as existing “at the intersection of human emotions and unstructured data.” Let’s unpack that. First, how do you define emotions in a consumer research context?

Lana Novikova: Back in the 1800s, William James, the American psychologist, posed the question, “What is an emotion?” And, after more than a century, psychologists and neuroscientists can’t agree on what exactly is emotion. Do animals have emotions? How do human emotions appear in the brain? And how do we express emotions? But, I’m glad you asked it within the context of consumer research and, after thinking a lot and reading a lot of literature about emotions, I very simply define it as: emotion is a conscious, explicit manifestation of that complex process we have in the body and in our mind in different parts of our brain that manifest in different ways and that’s what we call data, right? It manifests in language, how I can express emotion about a particular product, service. I can say it verbally. I can show it in my face. The emotion can go and you can see through my body, including pulse including biometrics and so on. And that’s why it’s so hard to measure emotions in the context of consumer research.

Adrian Tennant: In what kinds of ways have market researchers typically attempted to measure consumers’ emotions?

Lana Novikova: In a quantitative market research world, you know, we ask closed-ended questions because we want to control what data we get back. You can present a question, a context, “How do you feel about this product and this brand?” And give a list of emotions. You can show smiley faces and little cartoons and responders will pick the one that’s more closely related to their presenting emotion. Biometrics for example, it’s not very easy to use in field research, but the more and more they’re coming, possibilities of using  EEG devices to measure your brainwaves. And, you know, certain measurements for brain waves indicate high effect, low effect, attention, no attention, eye tracking is another way that can help you show attention. And if you measure pulse, you can see what people look at and if the pulse is increased, for example, we know that the person is in a state of affect. We don’t know if it’s positive or negative. We know that, you know, your heart beats stronger when you look at that particular brand. For example, that’s why I called my company Heartbeat AI, because there was an experiment that actually, 150, I think unmarried women were given a box that looked like a Tiffany box. that’s a turquoise kind of blue color box without even a brand name on it. But they were given a Tiffany box and their heart rate went up considerably, I think by 30% or something and like that. So their heartbeat went up. And then the reason why for unmarried women the heartbeat goes up? Because they associated the Tiffany box with a  surprise proposal of marriage. Think it’s such a beautiful story

Adrian Tennant: Why do emotions matter in market research?

Lana Novikova: Back in my corporate career days, I was running  consumer and shopper research for Nestle ice cream category. And imagine how emotional the category is. We wanted to know why loyal users of Haagen-Dazs go in the middle of the night to a convenience store and buy a small tub of ice cream for $6 or $10, and then come home and eat it all at once. Hiding it from their family. You see, this is the consumer action all driven by emotion. It’s – none of that is rational. So I wanted to know why it happens. Beyond that, you know, it’s yummy, it tastes good, but what actually drives that human behavior? So why it matters? Because I think we market researchers have responsibility to understand our consumers and shoppers at the deepest possible level. Not to be satisfied with very shallow observations, with very shallow insights, with just boxes and charts putting people into segments. So, for example, if I identify frequent users of premium ice cream, and I know what’s the segment of population, and I know their demographics and even psychographics, I still don’t know why they do what they do. The behavior that I just explained. So I wanted to go as deep as possible to understand what drives that human behavior and yeah, that’s actually one of the reasons why I left market research eight years ago to study psychotherapy and psychology and neuroscience

Adrian Tennant: Well, anyone that has used social listening or text analytics tools is probably familiar with seeing posts or open-ended survey responses classified by sentiment. So that is either positive, negative, or neutral. But I know Heartbeat AI goes way beyond that. Can you explain how?

Lana Novikova: Yes, and it came from my understanding of psychology and psychotherapy. Actually, after I studied psychotherapy, I ran a practice for two years, sitting with clients and listening. When you work as a shrink, you get presented with a lot of emotions. And that  confirmed my belief that people are just not negative, positive, neutral. They experience way more emotions than that and can experience negative and positive emotions at the same time. So imagine, we have these binary measures of sentiment in the market, but in reality, they don’t reflect how we humans experience emotions. Right? And I wanted to build a tool that can reflect the nonbinary nature of human emotions. And that tool is Heartbeat.  We took one of the most comprehensive segmentations of emotions by Gerrod W. Parrot of George Brown University in the United States. And he qualified about 136 secondary emotions going up to 8 primary emotions and the sentiment. So we scaled it down a little bit. We edit a category of body sense, which is not an emotion, but an indication how people feel in the body: whether hungry, thirsty, pain, and so on. So it’s a very useful category for some surveys or for some data. So we adjusted the psychological classification of emotions to market research: shopper insight, patient insight, and so on. And so we came up with 100 secondary emotion categories, the sub categories laddering up to 9 primary and 1 body sense emotions. And then we built a taxonomy of words and phrases. So each word and phrase in English that indicates emotion falls into one, or most likely, two or three different categories of emotions – that’s how it goes. So for example, we can show 15 different kinds of anger, 14 different kinds of joy, 5 different types of trust. And so that’s how we build it. So we can give you a little glimpse of what’s inside and that provides high accuracy of text analytics. It provides obviously, depths. Some clients still need to see a hundred emotions, you know, some categories are very dry, but it’s good to differentiate between trust and joy, for example, when you talk about your clients in your bank or your insurance company. So those are just a few examples.

Adrian Tennant: Let’s take a short break. We’ll be right back after this message.

Lauren Fore: I’m Lauren Fore, and I’m on the operations team at Bigeye. Every week, IN CLEAR FOCUS addresses topics that impact our work as agency professionals and reflects the way that Bigeye puts audiences first.  For every engagement, we develop a deep understanding of our client’s prospects and customers. This data is distilled into actionable insights that inspire creative brand-building and persuasive activation campaigns – and guide strategic, cost-efficient media placements that really connect with our clients’ audiences. If you’d like to know more about how to put Bigeye’s audience-focused insights to work for your brand, please contact us. Email info@bigeyeagency.com Bigeye. Reaching the Right People, in the Right Place, at the Right Time.

Adrian Tennant: Welcome back. I’m talking with Lana Novikova, founder and CEO of Heartbeat AI Technologies. As an agency, our typical use case is the analysis of quantitative data that we’ve collected through online surveys. Do you have any recommendations for writing open-ended questions to provide good quality data for emotional analysis with Heartbeat AI? And is there anything researchers need to avoid?

Lana Novikova: When we really care about the quality of the data and about the responder feelings, we want surveys to be shorter. What’s important really to find out? What would you want to learn from your shopper and consumer from your responder? Like in any conversation, when you start a conversation, it’s good to ask. How do you feel about this category or that? So I would put, open-ended questions about emotions at the beginning, not at the very beginning, but after pre-qualifying questions. So somewhere where people are not tired yet. We can go into neuroscience and understand where the answers come from. For example, if you ask closed-ended questions with multiple choice, it’s a very left prefrontal cortex action. So your brain needs to analyze, look at different choices, and select them. That’s what our prefrontal cortex does. When I asked, how do you feel about elections in the United States? Right away, most likely it’s not a prefrontal cortex question. It’s a very emotional question. So people will answer from a combination of the amygdala and hippocampus in different points in the brain.  And, you know, there’ve been great studies to actually show what happens in the brain when people talk about emotional subjects. Right? So I would say knowing that, ask very simple open-ended questions earlier in the survey in a context, and give people a reason to trust that the data will be used in a good way. And just like in any conversation or they will reveal deep, amazing insights. And sometimes it’s not just the lines, it’s not just the words. Sometimes people write a few phrases or if a few paragraphs even, and with the text analytics tool like Heartbeat, it’s very easy to analyze long responses, and the longer [the] responses, the more rich data you get. So very, very simple. Ask good open-ended questions early, don’t fatigue people, and, yeah, just let people trust you and they’ll give you good information.

Adrian Tennant: So just to understand the mechanics a little bit as a client with survey data, including open-ended text responses, does the researcher upload the data to an online platform? How does that work?

Lana Novikova: Once you collect your data, it’ll be closed-ended and open-ended responses. Say it’s an Excel or CSV file. So once we have a CSV file, it’s very simple to upload in our system. You know, our clients will get the password and then they can upload themselves or we can upload for them. We show all the metadata, all your closed-ended questions, your demographics will be on one side. And the open-ended responses we’ll show in a dashboard. Each word and phrase that is emotional in open-ended responses will be tagged into one or many categories of secondary emotions. You will actually see in our survey, words highlighted in blue, with anger, and you’ll see that word or phrase, it represents anger in that particular context. We use human curation to clean the data and we want to guarantee 95% accuracy. So we’ve worked with every data file. We clean it. F or example, the word cheesy. Right? So they drew a cheesy showing up in your survey responses.  When you talk about “pizza” or “pasta” [it] is not an emotional word. If your context is advertising for shampoo and you say, well, “The ad is cheesy” that’s an emotion. So things like that are full of disambiguation. So we would clean the data for those words to be disambiguated. So again, I explained to you what’s going on the back [end] and then the front [end], you see a very clean, very easy to use dashboard where you can slice and dice by gender, geography, and all your metadata. And see each word and phrase in the level of primary emotion, the level of sentiment, you see positive, negative, neutral, charts, and there’s all kinds of data mining also. You can do it through the dashboard. It’s very easy and fun to use.

Adrian Tennant: What are some of the most interesting consumer insights that you’ve been able to yield with text-based emotional analytics that isn’t typically possible with other methods?

Lana Novikova: I like examples from patient experience and employee experience because usually that’s where we get very rich data, very emotional data.  If it’s collected well. For example, a few years ago, we worked with  qualitative data, actually – [they] were interviews with patients with multiple myeloma, which is a very late stage of cancer.  A number of interviews where patients were asked their journey with their disease at the beginning before diagnosis, then later through treatment, remission, and now they’re at the last stage. So imagine it’s a lot of words, a lot of uh, dialogue. And when a human analyzes this qualitative data,  they usually pull topics and themes, it’s very, it’s impossible to quantify emotions in that situation. What happened is we split interviews into different stages, journey of that patient, downloaded the data, and it showed a very simple chart showing the emotional journey of the patient. And, it showed a very interesting spike of joy at the last stage. So there was a lot of joy, the beginning, then the patients go into that, you know, fear and anger stage, and sadness. There’s a lot of dark emotions, obviously. But then at the end of the last stage, when they kind of reconcile with the disease, joy showed up  – and a very particular type of joy – bliss. It was unexpected. We went into the data, looked into the joy and we found quite a few people were at peace with their life at that stage. And once you quantify that insight, it becomes very strong and it was beautiful, a marketing campaign for fundraising for multiple myeloma. And because of that insight, the campaign was built in orange colors with a beautiful picture of this older woman with glasses and with a big smile, instead of their, you know, typical kind of very dark, very sad campaigns. I love working with data like this, the rich data where manually it’s impossible to find it. But the system pulls it very quickly. And then the intuition of a researcher and a marketer will tell you, “Well, that’s interesting,” as something comes up, unexpected, “Let’s go deeper.” And when you go deeper, you uncover that gold mine.

Adrian Tennant: Now, you’ve mentioned the use of artificial intelligence and a combination of human and machine learning. But as you know, such systems can be biased based on who builds them, the way they’re developed, and how they’re deployed. So Lana, how do you mitigate the risk of bias in your technology?

Lana Novikova: It’s a very, very important question for all of us and for the future of technology. We mitigated from the very beginning. Our algorithms are not based upon deep learning. Our algorithms are based on supervised learning. We, you know, we engineered our system. Then we manually coded a taxonomy, the training data, I think we started with 10,000 words and phrases and manually coded them into those buckets of emotions. And that’s been done by professional, psycholinguists. So, first by one person, and then a few other people validate it, and that’s how we dealt with bias.  When, you know, a few people code the same word in the same bucket, we believe that that’s an emotion is such a… it’s not numbers, it’s a matter of opinion or the feeling. But when a few coders agree, that’s good coding. That’s how we build the system at the very beginning. And after that, we build a lot of algorithms that can take that solid first taxonomy, first training data, and extrapolate it into more words and phrases. So after that semi-supervised machine learning is used. It takes longer, but it’s worth it, because the accuracy is there. And then if we ever discover a mistake, we know exactly where to go and fix it. So I never liked black boxes, I always like open systems where I know where to go and fix it. So we just took a different approach.

Adrian Tennant: Now, how do you see the role of technology in consumer research specifically, developing over the say next five years or so?

Lana Novikova: I think it’s going to be even faster than the next five years. The transition to research companies to merge with technology companies. And you know, I’m biased, of course I’ve been running a software company for five years now, before that I was a researcher on the supply and on the client side. But now knowing the tech world, knowing how fast, how innovative the tech world is, to have a chance and having a job and career and business in market research, researchers will have to keep up with technology. As hard as it is we have to become aware, not to hide. We have to re-educate ourselves. It doesn’t mean we need to become programmers, but for example, if you are a statistician, knowing the principles of machine learning, text analytics, we’re already 80% of the way there. For example, the multivariate analysis or the algorithms that are used in machine learning, often statistically are similar to what we use in stats in multivariate. So it’s the same tool, same principles. So researchers need to kind of be brave and go and retrain themselves and just open up to technology instead of pushing against it. But it goes back to 2000 where, you know, the traditional pen and paper, quantitative researchers were saying that, “No surveys will never go online.” And lo and behold now, we know where this story is, right? In the same way, you know, embrace machine learning – it’s coming and better be friends with that than not.

Adrian Tennant: How do you keep up to date with the constantly evolving landscape of technology platforms and possibilities, and perhaps equally importantly, how do you determine what’s most deserving of your attention?

Lana Novikova: I’m an innovator. I’m very excited about anything new. Five years ago, actually, we got an award for best innovation in market research,  in Amsterdam at one of the big research events. So that sparked even more to keep looking for new things. For example, I’m very, very excited about chatbots, and how chatbots are going to influence research from data collection  all the way to analysis. And we are going to launch our first chatbot this year. I just love playing with things. So I guess if you’re curious, if you’re open, if you’re not afraid of technology, the rest is just play with that. And, I allocate time in the schedule to look for innovations, always read, keep up with innovations in market research and also outside of market research, because it’s very important to see what’s out there and what could be brought to market research. For example, Alexa or Google devices. You know, can they be used in a survey right now? We’re all at home, right? Can we use it as an interviewing device or collecting good qualitative data? For example, all of that could be done. And of course I have a team of programmers and that makes it easy for me to experiment and to program things. So I can just imagine it and my team can program, which makes it easy. So I’d say curiosity,no fear, you just go for it. And yeah, it’s, it’s easy to get lost in all these new things happening, but after a few years, you kind of develop a taste and you look for, “Okay, here’s the actual trend,” or “This is just a fad and is going to go away.” And, of course, Heartbeat is all about emotions. We’re not doing anything else but understanding emotion. So it’s really easy when you focus so deeply on one thing to do it really, really well. So any research that new things that come about emotions in neuroscience, I keep track of that with passion or obsession, I would say so. Yeah. And that keeps you going and innovating. Yeah.

Adrian Tennant: Well Lana, in parallel with your professional career, I know you’ve also been a very active supporter or there are a number of causes, but you mentioned a plan for an eco village in Nicaragua. Could you tell us more about that?

Lana Novikova: At the beginning, when you introduced me, you mentioned that I worked for the United Nations in Kyrgyzstan, and then in New York. So when I worked in Kyrgyzstan, it was 1993, for a couple of years. We ran UN offices just opened in my country. And I was a project manager. So we did beautiful work of building capacity in the poor country, building schools, building orphanages, and bringing Western knowledge and tools into a poor country to help people and that feeling of building something. For children who otherwise wouldn’t have a chance to go to school because it was destroyed by earthquake. That was one particular project. The feeling of meaning and purpose that it gave me was unforgettable enough so that I worked for many years in a corporate environment in North America. And I will always miss that feeling of fulfillment of doing something for people. So now in Nicaragua, I mentioned that I bought land and  I was planning to build a home on that land for me and my family and friends, maybe a few small homes, like a small eco village. But just a couple of months ago, we had  two hurricanes actually hit this country from the Caribbean side, and many, many families lost their homes. You live in Florida and you understand what’s happening, the more hurricanes will be coming. And  more people will be displaced by the environmental disasters. And this country by itself cannot support people who left without homes, the government is doing only so much. And the international community is doing so much. So that kind of opened my heart to think, “You know, I have the land and I can fundraise actually to build not five homes, but maybe 50 homes on that lot of land that I bought and just donate it to the village for people who otherwise wouldn’t have homes.” So that’s my dream to do it. It’ll take some time, but that’s another dream, so that’s back to the feeling of fulfillment and purpose.

Adrian Tennant: Lana, if IN CLEAR FOCUS listeners want to learn more about Heartbeat AI Technologies, where can they find you?

Lana Novikova: I would love them to find me. We have a website at www.HeartbeatAI.com. And to reach me directly, you go to Lana@HeartbeatAI.com. It’s my email. Yeah. I would love to have conversations and I would love to offer mentorship to people who really want to be in the field of text analytics, sentiment analysis, because I know it very well and I would love to share my knowledge with young professionals who want to come and maybe even work for Heartbeat one day.

Adrian Tennant: Lana, thank you very much for being our guest this week on IN CLEAR FOCUS.

Lana Novikova: Thank you so much for inviting me and thank you for your time. And I look forward to our next conversations.

[MUSIC]

Adrian Tennant: Next time on IN CLEAR FOCUS…

Eric Ortiz: How do I let people know what this product is, in a highly competitive CBD market, and also stay compliant within the ad space to make sure that my ads don’t get disapproved?

Adrian Tennant: A conversation with Eric Ortiz of Magical Brands, on navigating regulations around its CBD and cannabis-related products, that’s next time on IN CLEAR FOCUS. Thanks to my guest this week, Lana Novikova, founder and CEO of Heartbeat AI Technologies. You’ll find links to the resources we discussed today on the IN CLEAR FOCUS page at Bigeyeagency.com under “Insights”. Just click on the button marked “Podcast”. And if you haven’t already, please consider subscribing to the show on Apple Podcasts, Spotify, Google Podcasts, Amazon Music, Audible, or your preferred podcast player. Thank you for listening to IN CLEAR FOCUS produced by Bigeye. I’ve been your host, Adrian Tennant. Until next week, goodbye.

And More