The Role of Emotion in Marketing: Understanding the 98.8%

At Element Human, we believe measuring only what's easily measurable is one of the biggest frustrations in marketing. Instead, companies should focus on measuring what's truly valuable, including the human experience. This is especially important in the creator economy where over 200 million creators interact with over 4.2 billion people daily. We're building the "empathy layer for the internet" in part to help brands and creators form deeper, more meaningful relationships with their audience.

Traditional metrics, like clicks and impressions, fail to capture the full impact of a campaign. Only 1.2% of viewers take an action that can be directly attributed to an ad. What happens to the other 98.8%? To understand this "invisible" majority, we use biometric data to measure emotion and attention, providing insight into the human experience of content.

In Sima's conversation with our Co-Founder and CEO Matt Celuszak, Matt highlights some of the key insights we’ve uncovered:

• Attention data is crucial for effective advertising. It is essential to understand whether viewers are actually paying attention to an ad, as impressions and views don't necessarily translate to engagement.
• Authenticity is vital in the creator economy. Younger generations are especially adept at recognizing inauthenticity. Brands should prioritize genuine connections with creators and avoid overly scripted messaging.
• Emotional response drives action. Element Human research suggests a strong correlation between emotional engagement and purchase intent. Evoking positive emotions can lead to increased conversions.
• Behavioral insights offer a deeper understanding of consumer behavior. Element Human leverages behavioral data, including facial coding, biometrics, and attention tracking, to gain insights into consumer reactions to content in real-time.  This provides a more comprehensive understanding of campaign performance.

Looking at creatives and advertisements through the lens of human experience adds depth and nuance to how one sees their marketing effectiveness.

In this spirit, we're dedicated to building AI-powered tools to make these insights accessible and actionable, enabling brands and creators to develop campaigns that resonate deeply with audiences and drive results across the entire marketing funnel.

‍

‍

‍

Key Highlights:

(03:39) AI-driven insights help marketers understand how human emotions affect engagement.

(07:15) Attention and stopping power are crucial for creating lasting brand impressions.

(11:03) Combining human behavior with data drives personalized marketing strategies.

(14:24) Emotional response correlates with memorability and purchase intent.

(19:00) Real-world behavioral responses provide complementary data to standard metrics.

(23:15) Balancing creators’ authenticity with brand messaging is key to successful influencer marketing.

(28:10) The staying power of content is as critical as initial engagement in brand building.

(31:30) Understanding the emotional impact of content helps brands connect more effectively with consumers.

‍

‍

Transcript: 

The Role of Emotion in Marketing with Matt Celuszak of Element Human

[00:00:00] Sima Vasa: Welcome to another episode of Data Gurus. I'm so excited to welcome Matt Celuszak, who is the co founder and CEO of Element Human. Welcome, Matt.

[00:00:15] Matt Celuzsak: Thanks for having me here, Asima. Lots of interesting stuff to talk about.

[00:00:19] Sima Vasa: Yeah. And I love the shirt that you're wearing. It inspires goodness.

[00:00:24] Matt Celuzsak: Yeah, it's a colorful, I don't want to be stuck in a box.

[00:00:29] Sima Vasa: Usually I wear black, so I'm wearing a little bit of different colors. You got your

[00:00:33] Matt Celuzsak: summer colors out today.

[00:00:34] Sima Vasa: There you go. There you go. I'm so excited to talk about some really critical concepts that your company's addressing and also based on some key. Perspectives that you believe before we dive into that, I'd love just to talk a little bit about your journey as to how you got here.

[00:00:55] Matt Celuzsak: Yeah, sure. Thanks for that. I got started at university, I was doing my degree in human kinetics, which is around anatomy, physiology, biology, and very grounded in the world of sport. And One of the challenges that was brought up that kind of hooked me into this idea. So Elements Human is a human data company.

Yeah. And what hooked me to this idea around, can we use technology to understand ourselves and each other better? Was when a cross country coach came to me and said, my cross country team just isn't training. They're drinking, they're doing the university thing, but they're not training. And that's when we've all been there.

And so what was really interesting about it is he said, can you help me with this? I said are you sharing the VO two max data with them? They're like lung capacity data, and we were using sensors to try it out. He goes no. And I said why not? He said cause I need to know that. And then I translate it for them.

I said have you ever thought maybe if we just box up the translation and hand it to them, they might do, they might respond a little bit better because we're finding a good correlation with lung capacity and performance,

At the races. And so he he said, all right, great. And so he tried it and then he came back about four or five months later saying, okay, I now have another problem.

I said, what's that? And he goes, They're not training too hard. They won't stop trying to compete on lung capacity. I was like, so I think it just goes back to people manage what they can measure. And that's how I fell into the world of measurement. Then I went to the first job was an internship and then ended up working at IMG to help build sponsorship measurement platforms for brands because they were spending.

Millions to hundreds of millions of dollars on people like Tiger Woods. And how do you put a value on that and a return on investment? So lots of sponsorship deals and seeing how money can translate for measurement, which was pretty cool. Then I went over to wanted to learn how to do business development.

And that was with a company called vision critical, which was in the market research space. We're building online community panels now called Alita. I'm still there, still going strong. And I was doing new market development with them. So I worked across a number of different markets. And so I learned how.

Measurement not only changes, but also the insight and value of continual feedback with different groups in different countries can lead to different types of success and outcomes within businesses. I thought that was super interesting. And that's where I got a real first flavor of how a market researcher and a data guru could actually play their role within an organization to serve a number of different stakeholders.

And then that's where I first met somebody around neuroscience where we got into the. What if we could use video cameras, webcams, and we're not using it on this call. We all use it with permission, but how do we use video cameras, webcams, microphones, that type of stuff to get body language or behavioral AI today is what it's called.

And 11 years ago, my co founder Diego and I started on that journey and I started to bring [00:04:00] emotional models to everyday practitioners. In which it's now currently being used to help build standards in a standardized measurement in the creator economy, which is a heavily fragmented space of influencer marketing.

And it's been an interesting journey, but yeah, we love it.

[00:04:18] Sima Vasa: There's so much newness and cutting edge stuff that you're working on. Creator economy, one could argue it's not new, but it's accelerated a lot in the last few years. And in terms of, yeah, it

[00:04:32] Matt Celuzsak: has eerily. So crater economy is about, let's roughly 500 million people talking to 4.

2 billion people on a daily basis. So when you look at it from a data transfer, a human interaction, a human experience perspective, I thought that's probably the most amount of human interactions or experiences of any industry on the planet. There's very few that have more touch points. That actually have a machine interface in the middle.

Now, by putting machine interface in the middle, you can capture a lot of information. Got it. And that's what's really interesting and information that has a lot of inference . That we don't have in other data sources, like surveys or things like that. Now, capturing is one thing, making sense of it as a whole other thing.

Oh yeah. So that's been the 11 year journey, I would say. What also is interesting, what strikes me is interesting about the creator economy. So we support companies like Whaler, Omnicom, Influencer to build measurement systems. So what really strikes me is the eerily similar problem and use case to sponsorship of celebrities and individuals that we had at IMG in the early noughties.

So same type of thing, I'm going to spend a lot of money on this creator, on this influencer, and I need to measure what happens between the view and the click. And I have no idea how it's contributing to my brand. And so that, that's a big question. That's hard to mind marketers these days.

[00:06:01] Sima Vasa: Yeah. And I think also, it's about the click, but it's also about the people who don't press the click, right?

And there's very few

[00:06:12] Matt Celuzsak: categories that go to click right away. Most categories are, or most of the marketing funnel that's click based attribution is right at the tail end of building the brand.

So sometimes you get trended stuff and some categories move fast. Sometimes certain deals like where there's exclusivity in a market or something like that, where you can only buy a hundred of these vehicles, you see that but equally in automotive, you also see one to two year buying cycles from first exposure of an electric vehicle from Ford all the way through to buying the Mustang EV.

[00:06:49] Sima Vasa: And that's actually would be an interesting something to track, like going from your standard car to getting an electric, electrical vehicle. Interesting. Okay. So let's talk about, you have some a very strong perspective about research. Can you please share that? Yeah.

[00:07:09] Matt Celuzsak: Because I'm not a traditional researcher and I've grown up around them where we worked in this whole market research industry that a practice that's 65 years old as to as a number of governing bodies but you, so Mark being what big global one.

And then you've got a market research society and the UK and then a number in the U S and. And across different cultures, you have different research practices that are ethical guidelines and methodologies. What's very interesting is with the emergence of technology, what we're seeing, and we saw this early days at vision critical was this thing called a rogue research, where you'd have a salesperson adapt a SurveyMonkey tool and then just fire in a poll to their friends, and it's okay.

But there's a lot of bias in how you ask that question. You asked a question that was going to be answered the way you wanted [00:08:00] it to. And bye. The reality is that curiosity isn't owned by any one role within an organization. And particularly in marketing where there's a lot of questions getting asked, research is more of a function, not a role.

So our premise and one of the guiding principles of our business is how do we just put the people first, always in it and our people, we work a lot with research teams, but a lot of them like working with us because. One person can have a lot more scale and impact within an organization because a lot of the tooling we provide them with can be used by someone who's making 25, 35 grand a year has to just make a decision and they want to make it from an informed stance so they don't lose their job or so they, they can look good in front of their boss or whatever it is.

And so that really is actually the problem that anybody, I believe in the insights and research space and the data space, that's our job, effectively, We're effectively managing a marketplace and our job is to make that data as easily accessible and usable as possible by people who are non researchers.

So that, so anyone can ask their question. And if you get the curiosity of the masses, you can open up huge conversations and discussions and massive opportunities. And so that, that's why, that's how we view it as a research is more of a function on a role. And I think you'll note the top business leaders in the world have an element of curiosity and always know how to do really good research.

[00:09:25] Sima Vasa: Yeah. But it's an instinct

[00:09:27] Matt Celuzsak: for it.

[00:09:27] Sima Vasa: Yeah. And I also would say that curiosity doesn't just lie in research, right? There's curiosity. Across the board and finance data and sales data and everything else. But I agree with you. Like curiosity could be, I have a question. I want to go get it answered. I don't have to go to five other people to run a study and then get the answer.

I think if I interpret what you're saying appropriately. Yeah,

[00:09:51] Matt Celuzsak: a little bit. Exactly. A little bit like that. The other one is, and there's so much value in knowing the right question to ask, which is, this is. This is where I see, so we're seeing a lot of our clients and some of our some of the stuff that we saw in the past that we're seeing in the industry is this divergence.

You have researchers who are moving towards insights and strategy. So they're like, I'm going to be really good at getting the right tool in front of the right people. And then helping those people learn how to ask really good questions that will get them real actionable information. Cause research isn't, the research is a means to an end.

It's a tool to answer the

question. Yeah, exactly.

[00:10:34] Matt Celuzsak: But if you ask the wrong question, you're not going to have much to action on the back end. So it starts with the briefing and a lot of our. We've had clients come to us and say, great, you do behavioral AI, attention, emotion, memory, data. Great. I need attention data.

I'm like, okay, why do you need attention data? And they're immediately like, I don't know. Because I need to know beyond the view. What are you trying to know beyond the

[00:10:58] Sima Vasa: view?

[00:10:59] Matt Celuzsak: There might be better ways to do this, and quite frankly, the most expensive ways to do this or faster or whatever it is.

But what's the business problem here? And so teaching people how to assess the business problem and ask the

[00:11:10] Sima Vasa: right

[00:11:11] Matt Celuzsak: question, that's where. That's where I think researchers become really impactful and effective and that's where the role then empowers the function within everybody else.

[00:11:21] Sima Vasa: Yes. So just keying off that, why do people want to know about attention data?

[00:11:28] Matt Celuzsak: Oh, yeah. Quite frankly our managing director, Hamish, Dr. Hamish McFarland ex BBC, CNN built kind of the measuring programs there in, in athletes of the commercial groups. And what, He likes to quote something called the forehorsen of the ad apocalypse.

[00:11:49] Sima Vasa: And

[00:11:49] Matt Celuzsak: it basically is, and I love it because

yeah,

[00:11:54] Matt Celuzsak: Just because you have an impression and a view doesn't mean somebody looked at it.

[00:11:58] Sima Vasa: And

[00:11:59] Matt Celuzsak: quite frankly, [00:12:00] we're incredibly visually stimulated as human beings in our day to day lives. And we need to first understand if somebody looked at it, then we need to understand, did somebody actually take it in, react to it? Did they remember it? And we sound a lot of correlation between the intensity of the expressional response.

[00:12:19] Sima Vasa: Yep. Yeah.

[00:12:21] Matt Celuzsak: Expressional response. I'll come on to a motion in a second. The intensity of the expression of response to the to the memorability and then the memorability to purchase a ticket. So it's about actually saying, 99 percent of ads get viewed.

[00:12:40] Sima Vasa: Yes.

[00:12:41] Matt Celuzsak: But don't get clicked. So the real question is, what's happening with the

[00:12:45] Sima Vasa: other 90%?

What's happening? Yeah, exactly. What's happening

[00:12:48] Matt Celuzsak: to that whole funnel there. And real world human behavior response gives us a glimpse into that. It's a complimentary data set to the viewability, the click metric and the survey metrics. What's actually happens is it's the qualifiers.

It's what you would get out of a focus group or an interview or some sort of observation but massive. So our job has been again, make it available for the masses. So massive scale. So now people can, in our case, they can load up a piece of content. I have, it's a good one. I had to demo to a client and they forgot to send me the campaign.

They wanted to see it show off. So they sent it to me 10 hours before the presentation and I had, it was three studies in Australia, was able to upload it, send it across, get the response out, the report done by the time the presentation happened just to show off the power of the tool. And so we've worked really hard to just package that all up and make it super fast because in influencer marketing, you need under 24 hour turnaround times.

So that's just cause after that.

[00:14:03] Sima Vasa: It's a

[00:14:05] Matt Celuzsak: barn

[00:14:06] Sima Vasa: the trend just goes dies off. It's like a hula hoop. It goes up Maybe not a hula hoop, but like trending

[00:14:13] Matt Celuzsak: stuff but again, a lot of influencer marketing, like these brand collaborations and that for a brand dollars represents a huge portion of the commerce for influencers and creators.

Okay. So they need to know, there's also a brand protection element. Is it going the wrong way? Did we miss, did we say, did the brand message dictate too much? And the creator themselves lost authenticity with their own audience. Which audience is performing better. So we use attention data and kind of attention is one aspect of one output of our behavioral data.

So we look at on one axis, we look at percentile rank.

[00:14:54] Sima Vasa: Yeah.

[00:14:54] Matt Celuzsak: Of we call it stopping power. So when people go into a feed, we've built private feeds. So they look at it, tons of distractors in there. When they go into a feed, how much attention do they pay to it? So it doesn't stop them in feet.

And then the staying power is very much it goes along the other side, which is around if it did stop them, how memorable was their content for those that stopped. And that's survey

[00:15:20] Sima Vasa: data?

[00:15:22] Matt Celuzsak: It's observed with webcam data. So it is a recruited audience into a private environment. It's a mock environment.

It looks and feels like Facebook or TikTok or Instagram or whatever your environment that you're producing on. Yeah. Yeah. And then the audience looks exactly like the audiences that you have and we've done side by side tests with the audiences you have. So the weights are in the right place.

And and then that allows a brand to say, okay, great. Here's my stopping power. Here's my staying power. And here's what's working really well. And of course every single test that's added in from all of our partners, like it's aggregated to a benchmark. Today, [00:16:00] how you're doing. And if you want to test next week, you can test how you're doing.

If you want to give the diagnostic feedback to the creator to say, really nailed it with the storyline, but not for the brand,

[00:16:10] Sima Vasa: you might want to capture all the attention.

[00:16:13] Matt Celuzsak: But can you get the brand front and center and if you do, we can give you more media dollars.

That's cool.

You

[00:16:22] Sima Vasa: are in the activation space.

You are helping media buyers determine how to tweak. They're messaging, how to work with creators,

[00:16:35] Matt Celuzsak: content in context. So where most attention providers play and they're really good at, quite frankly, I'd say the best ones out there are not us. And they're very much in the media buying space.

So they will be able to give you a weighting based options based options of the inventory value. They're very focused on the contextual element of that. We focus on the content. Inside that context. So we focus on the actual story, the actual interaction, where people are looking, where they're expressing, what are the key moments, cut it down, extend it, all that type of stuff.

So we focus on the real nature of the creative and the content inside the context of TikTok or Instagram or Facebook or YouTube or wherever.

[00:17:22] Sima Vasa: That makes sense. So even

[00:17:23] Matt Celuzsak: this, if we wanted to put this on, we could take a five minute clip of this and run it make sure we get the right type of puck and reel that we want on there.

[00:17:32] Sima Vasa: Oh, that'd be fun. To

[00:17:33] Matt Celuzsak: boost viewership.

[00:17:35] Sima Vasa: That'd be fun. Why

not?

[00:17:36] Sima Vasa: Yeah. Actually talking about that, I was like, I know we said you collect a ton of data, but Like on a YouTube, how many data points on an individual would you collect privacy? Obviously it's anonymized. You're not like give me a sense of how much data does that involve?

[00:18:01] Matt Celuzsak: When a creator looks at a YouTube video and they'll get a bunch of feedback points. Yeah. Drop off points and

Yeah.

[00:18:07] Matt Celuzsak: Likes, comments, all that type of stuff. And it's very rich data, right? All this platform interaction data is very rich from an engagement perspective, but it's not contextualized, again, how is the audience actually experiencing it?

Are they just trying to get through it? Are they skipping? What are they doing and or are they even looking at it or is it on a second screen and they're actually just looking out there the whole time. Yeah. .

So what we offer is for every kind of there's about 10, 000 data points per individual.

And then there's 150 to 250 individuals that look at the content on our test. So you're into some pretty serious data points. Very rich data very rich data. Cause you're taking. About anywhere between 256 to a thousand data points, a frame at 30 frames a second. And so you just like this incredibly, and it's something that even as a human, even as a human moderator observer,

[00:19:03] Sima Vasa: you can't pick it up really hard.

Like

[00:19:06] Matt Celuzsak: you can be the top trained FBI psychologist out there. I know. Can I miss this minor tweak? Of course you did. You're right. That's not what we were brain to do. What you can do is interpret it. And humans are very good at interpretation. Much better than machines. Machines are much better at looking for signals that correlate with other things and looking for signals that predict other things.

[00:19:28] Sima Vasa: So the interesting point I think is that you collect all this data, but the result is quite simple, not simple, but it's robust in its, in, in terms of the data that informs the results, but it's pretty easy to digest and consume for somebody who's looking at the data.

[00:19:47] Matt Celuzsak: It only took 11 years.

[00:19:49] Sima Vasa: Yeah. That's entrepreneurship, right? Yeah.

[00:19:54] Matt Celuzsak: You shouldn't see in the initial data graphs that we had going through there and it was just like, you look at [00:20:00] it, go what's even.

[00:20:01] Sima Vasa: Yeah. What's happening here.

[00:20:03] Matt Celuzsak: What's. So have you ever seen when you see like a voice note or a song that has all these kind of, and you see it on, yeah, so that's what visual data can articulate as one visualization of kind of the emotional time series of expressional time series.

And that's why when he comes on, like I said, I've come on to emotion. Emotions themselves are still in a bit of an interesting space of development.

[00:20:29] Sima Vasa: Okay.

[00:20:29] Matt Celuzsak: There is. A little bit of an agreement and disagreement around universality of emotions, like we all have the exact same expression. I disagree with that personally.

I haven't seen any data to prove that. And so we don't necessarily focus on some customers ask us to classify into emotional states, these six general Ekman states. I think what I really appreciate about that model is that it got everybody talking about it and made it adoptable. Problem with that models from a data perspective, it just doesn't.

Doesn't provide the consistency, way too much variability. So where we started in, and I'm not saying we have the answer yet. I don't think anybody does. And I haven't seen a really good. Use case out there. But we started really focusing on the individual and normalizing on the individual and looking for how much their expressionism deviates from the norm.

[00:21:19] Sima Vasa: That's interesting. And we

[00:21:19] Matt Celuzsak: find that to be incredibly Yeah. Insightful. Much more so than and that's why we focus on expression first, and it's the general expression. It's less about what the emotions classified about and it's much more about the intent.

[00:21:32] Sima Vasa: That makes perfect sense. 'cause as soon as you said that, I think about ethnicity and culture.

We all right, we're taught differently to express ourselves.

[00:21:40] Matt Celuzsak: And even, the best example of this, I was initially introducing this capability in the UK and I was showing it around and immediately just had expats who'd come back from overseas and say, oh, this would never work in Asia. So I said, okay.

So I joined the UK trade commission and went to Xingdao in China. I'm probably butchering that name. So apologies to do all the Mandarin speakers out there. So we went out there and then I literally ran it and the system had no problem picking him up. But absolutely correct. Did us did I have an issue picking it up?

Yeah, I did.

[00:22:24] Sima Vasa: Yeah.

[00:22:24] Matt Celuzsak: But it's just because I haven't grown up around that particular individual. And so it just went to pre, even within two, three seconds, you have a much stronger understanding of somebody than somebody who's seeing them for the first time, and this is where the whole business model is built off of.

If we can provide a very safe environment for people to interact safely with machines with a responsive AI, we can build a layer of empathy with inside that interaction. But more importantly, we can also empower this data back to the individual, like we did those cross country runners. So then they can manage their time.

They can, I, they can have an alarm come up and say it's doom scrolling on TikTok and this is where it gets really exciting. This is where it gets really exciting.

[00:23:13] Sima Vasa: It's like nudges, right? Hey, listen, you've been on, you've been doing this.

[00:23:17] Matt Celuzsak: This is what I'm observing. No judgment, right? With it, what you want to do with it, that's fine.

I'm observing every time you're on this call with an individual, and you see this expression happens on the other person.

[00:23:28] Sima Vasa: Yeah. Or this

[00:23:28] Matt Celuzsak: low happens, you tend to do X.

[00:23:31] Sima Vasa: That's amazing.

[00:23:32] Matt Celuzsak: I would. I personally would love that. So selfishly, I'm building this for myself first and

then

[00:23:38] Matt Celuzsak: if anybody else wants to join like the community.

[00:23:41] Sima Vasa: I know. I love that. That's great. So let's talk about this. Cause I, I know that, we talked about two things. Number one, younger generation really being focused on authenticity and you can, they can sniff out authenticity or [00:24:00] lack of. Immediately. And it's a really critical component.

How are you seeing that play out in the data? Are you, give us a little bit of a sense of some of the trends

[00:24:12] Matt Celuzsak: yeah. So the biggest one is so we started tracking authenticity three, four years ago, trust, trustworthiness, and authenticity were the two brand traits we were looking for and we put them in as implicit brand traits.

So it's a implicit reaction time test.

[00:24:30] Sima Vasa: Okay.

[00:24:31] Matt Celuzsak: That we've baked into the tool. So Tinder for brands, as we like to call it.

[00:24:34] Sima Vasa: Yeah. Got it. Yeah.

[00:24:37] Matt Celuzsak: And it's very interesting to see how authenticity and engagement kind of start to link together and also how authenticity and some of the brand metrics like uplift that type of performance starts, starts to link together.

But we have seen and Hamish has again, has done an expose on this around kind of the power of authenticity. And look. I think everybody and their brother in the marketing industry that can like, okay, authenticity is, it is, it's been beaten to death. I actually really hasn't.

And I, and what I think has been missed entirely is that as a marketing industry and a media marketing industry, We have flooded the market with so much content, particularly in the influencer marketing space that again, I go back to 500 million people daily talking to 4. 2 billion people. Not so many actual interactions are happening.

So what has. Notably happened for those that spend the most amount of time on social media is they're getting real good notes

[00:25:48] Sima Vasa: or

[00:25:49] Matt Celuzsak: sense. I call it a sense.

[00:25:51] Sima Vasa: Thanks. What's authentic and what's not. Yeah.

[00:25:55] Matt Celuzsak: Yes. They're getting a real good kind of sense for what's authentic, what's not. And particularly, especially as we start to see advertising standards get into the influencer space, cause we're seeing that as well, with some of the more, more recent stuff that's happened with Stephen Bartlett. He's actually an investor here and he's also promoting the product. He's not allowed to do that or he hasn't been very clear on that. It's interesting. Cause I'd say his content is very authentic. He believes in the product.

He bought the product. He bought shares in the product and he went ahead and did it. But. It puts a huge question mark around authenticity. Do we moderate for it? How do creators make sure they don't lose themselves when they take a brand brief? How do we teach brands how to brief so that the creator's authenticity goes through, because the greatest number one driver is still audience growth and engagement.

[00:26:44] Sima Vasa: Because

[00:26:45] Matt Celuzsak: without it, they don't have a brand. They

[00:26:47] Sima Vasa: don't have it. They have not. Yeah. So when

[00:26:50] Matt Celuzsak: a brand comes in, then it says, cool. Cool. Some agency manager has to go, okay, I'm thinking of this great strategy. And then they literally hit up and I've, I talk a lot with creators and the creators go, so then I just get this force fed what I'd have to say because of the brand guidelines

[00:27:06] Sima Vasa: and

[00:27:06] Matt Celuzsak: it is so jarring with my audience because they have no hope, like they're like, this is, you're clearly paid to do this.

Just tell us you're paid to do this because you need to make money and that's cool. We'll accept it as such, I'll probably skip it, but we'll accept it as such.

[00:27:20] Sima Vasa: Yeah. I didn't even know that they had awards, I learned recently they had awards for people who read their ads, read live ads on podcasts.

The best I had no idea. I was like, wow, they said that anyway, I won't mention the podcast. I listened to a few, but they were celebrating the fact that they won an award for it. I'm like, wow. I wouldn't even have thought of that. It's paid ad. They felt authentic in the way that they read the ad. So let's talk about, guardrails around this whole digital world.

You mentioned that [00:28:00] empathy before, but How do you we're in a two way dialogue virtually. Versus being in a physical room. How do you view the differences there?

[00:28:14] Matt Celuzsak: When you actually dig into psychological definitions of emotion, one of them is it is environmental. It's that sense, that feeling you have in a in a space.

And if you've ever been in a, if you've ever been close to a crime or I've been in a crowd that's gotten riotous. You could feel it. You literally, there's an electricity in your hair. You could feel, yeah, you're like, Ooh, something's going on. I actually don't even know what it is, but something's going on.

Yeah. And, or, just around spatially. So there is that element. I think there's a few things that I think are going to be true. One, I think. The human in five years is not the human today. And we're going to be drastically augmented, whether it be through external devices or internal devices, one or the other.

We, humans will always selfishly try to advance their capability and by, by any means possible. And. And so I think from that perspective, though the machines that are here to help us still relatively crude and I say that kind of with the knowledge of what LLMs could do and what these chat bots could do and and that type of stuff and frankly, they're crude, we've made incredible strides and breakthrough, but we're now getting to the point I'm also the firm belief that we will always make kind of technology in our image.

So there is an element of that. And then finally this social media and social interaction and creating these kind of public town halls where you can hide a little bit behind your opinion and you

Kind of deescalation effects starts to happen. I believe two things. I believe we've lost the arching discussion.

[00:30:06] Sima Vasa: Fair. I agree with that.

[00:30:07] Matt Celuzsak: In in, in a sense that we've forgotten how this. And it's incredible what happens when you learn how to listen, because the reflection moment takes sense through space and you tend to see, just look at any of the top performing teams out there, they are acutely good at both listening and feeding back.

And what we haven't done is we haven't embedded that feedback loop into our devices. So we're interacting with these devices. We're interacting through these devices, but we actually had we've removed this tactical, emotional sense that allows us to have a real strong bond and empathetic response to somebody on the other side, but It doesn't, this current device doesn't allow us to do that.

Yeah. This device is, I will show you how you have to do all of it. So I use my hands a lot. We get colorful, we get expressive, we put blurred on the background so you can focus on the face, all that stuff. But the reality is that we haven't done that. And my firm belief, particularly in the world of artificial intelligence is highly generative, which is great.

It helps and frankly from a thinking and assistive tool for what I'm trying to create. Oh my God. I love it. It is so good. It has saved me hours of help me synthesize what's happening. Here's what I think is happening. What would the counterpoints be? How do I negotiate something? Here's where all that type of stuff is great.

And it just, but it's not great because it doesn't work. It's great because it changes the way I think. And I think that's really good. So now the second step to that though, not everything's logical. And that particular [00:32:00] type of artificial intelligence, it's very big model oriented. You're talking, billions of data points that have to be crutched to get a good model out.

What about the individual effect? I am still different than you. We can both use an LLM, but I'm still different than you. And what something emotionally means to you will mean something very different emotionally to me because our memories are different. Our experiences are different. So how do we build some guardrails around this AI saying this is the way?

Rather give the AI a feedback loop that says this is the way for this person, but actually didn't land so well for that person. So keep iterating. Like you got a job to do. If

[00:32:36] Sima Vasa: you

[00:32:36] Matt Celuzsak: wanna be this most generative person in the room

[00:32:40] Sima Vasa: Sure.

[00:32:40] Matt Celuzsak: Be that.

Yeah.

[00:32:41] Matt Celuzsak: But know that you're having an impact on somebody. And this gets us away from, I think we should be a lot more sensitive as organizations, particularly as we have more and more reach and Brent Yeah.

To the individual consumers who are using our capabilities and a lot more respectful to the exposure that they're having, particularly with their information, their data. And I believe that, that is where the emotional guardrails come in. So I am going to, for the rest of my life, always work on getting better and better understanding of humans for the individuals.

And yes, that'll ladder up for corporate interests and all that type of stuff. But as long as the individuals have the control I'm happy with it. You feel good about

[00:33:29] Sima Vasa: it. You feel like, listen, you're enriching. I can go to sleep at night.

[00:33:35] Matt Celuzsak: Speaking about that, that, that trick. Yeah. To China, it was a, it was quite an eye opener because at that point, some of the Chinese companies in the biometric space were not around in their early doors and we were asked to go take a look at some of that stuff that ended up in social credit scoring.

And I'm like, that is not where we need to be going. Yeah. And that is not a rising tide raises all boats. That is in my view. And so that's why we came. That's why I said to my co founder, I said, we got our I'm trying to figure this out, but for, from a different way, and how do we, if we can give this to everybody, then everybody's empowered with their own VO two backs, if you will, but for emotions.

[00:34:18] Sima Vasa: That's cool. So let me, I just want to clarify something. So people who participate in the research, if you will, do they get feedback on themselves as well? What?

[00:34:30] Matt Celuzsak: Yeah. Nobody gets that feedback. So what we do is we take the person identifiable data, and

then

[00:34:38] Matt Celuzsak: we create our models, create a bunch of vectors underneath that.

Got it.

[00:34:44] Matt Celuzsak: Then those vectors are not identifiable again. Got it.

Yeah, of course.

[00:34:49] Matt Celuzsak: And then we look for commonalities in those vectors. And then what we do is from there customers like Twitch who publicly talked about this. They then link it to things like video completion rate.

And so they say video completion rates important to us. We say, here's the survey data. Now here's what we're also pulling from the biometrics. But it's all learning feedback loops. So whatever organizations we work with, it's what data is important to you. We're going to provide you the standard stuff.

You get the survey data and all that type of stuff. And whatever other data you want to bring in, and we're going to bring this behavioral data as well

and

[00:35:25] Matt Celuzsak: Make that a learning ecosystem.

[00:35:28] Sima Vasa: We are truly in the world of big data.

[00:35:31] Matt Celuzsak: We are, but this is actually what I call small data because not, I call it

[00:35:34] Sima Vasa: everybody does.

Yeah.

[00:35:36] Matt Celuzsak: Yeah. We're I would say this is the misnomer about emotions. It's okay. This isn't a lot of the emotion providers out there are still stuck in this kind of, you have to classify up to trade up to boil the ocean. I actually, we take a little bit of a more individualistic approach. You are your own algorithm and you are your own model and we do just need cross comparative ability across the [00:36:00] models.

So we look at it from bottom up rather than top down. So it's all big data down to small data. We look at it from small data, ladder it up into big data.

[00:36:07] Sima Vasa: Got it. Got it. Matt, I congratulate you on all that you're working on. It sounds fascinating. Yeah. And I'd look forward to keeping in touch with you.

[00:36:17] Matt Celuzsak: Yeah, I would look, I appreciate you having me over here. It's a very fun topic.

[00:36:22] Sima Vasa: Honestly, I feel like we could talk for hours. I appreciate you taking the time. I know I

[00:36:28] Matt Celuzsak: can. Yeah. But my wife tries to tell me off about

[00:36:30] Sima Vasa: that. Take care. Thank you.

[00:36:34] Matt Celuzsak: All right. Thanks, Sima.

‍