Building a Successful Product Business: Five Key Lessons Learned by Canvs Founder Jared Feldman

Tremis Skeete
Jul 8 · 15 min read

People today consume content through more channels than ever before, and they’re also using media channels to talk about the TV shows and movies they watch. And when they’re talking about these shows, sharing their favorite moments, quotes, reactions, opinions, and creating content for fellow fans to respond to — a lot of feelings are being shared and exchanged.

Canvs Founder and CEO, Jared Feldman believes in the importance of understanding how people feel on social media as they consume content. By leading the creation and development of groundbreaking qualitative analysis products, Jared is dedicated to helping companies like Netflix, Sony, and Disney understand how their content impacts audiences everywhere.

Jared Feldman, Founder & CEO of Canvs

At a fireside chat at Meetup HQ in New York City, Jared sat down with Greg Battle, Chief Product Officer at finance intelligence company, Laso to discuss Canvs products, the world of qualitative analytics, and what it takes to build a business dedicated to understanding the power of empathy.

Building Products by Understanding Empathy

Jared, from a young age, was an aspiring musician with dreams of working in the television industry. By the time he arrived at NYU to study music and technology, it was instilled in him that the reasons musicians perform is to engage with the subconscious within people and elicit emotions.

His appreciation for music and emotions is one of the things that inspired him to create Canvs:

“When we started Canvs, it was founded with the mission of making the world more empathetic. And what makes someone empathetic is to really understand how the other person is feeling. We want customers to have the ability to understand modern dialogue on social media as they learn how consumers engage with content.”

Automating ways to understand how consumers feel about content is a product specialist’s dream at companies today. With so many opinions being shared, there is no limit to the possibilities in data, and leveraging the power of natural language processing, computational linguistics, and other technologies are only the beginning of what Canvs does for their customers.

Understanding the Impact of Emotions in Social Media

Emotion is a valuable, yet ambiguous element — especially when it comes to understanding the impact of video and social media content. And when emotions intersect with language, expression and meaning; providing actionable information based on feelings people share online is the challenge Canvs takes on for its customers, as Jared explains:

“We analyze emotional reactions and communicate insights based on the feelings people experience while consuming content. These insights are designed to build empathy, because our product and marketing customers want to understand and make sense of these emotions.”

Now understanding emotions is not a simple task, especially if you lack context, for different people have different ways to share their feelings. That’s Jared and his teams focus on identifying the core emotions that people use when they express how they feel online.

“There are 42 different core emotions we recognize as the ways people share their feelings online. We’re also teaching A.I. technology to learn and understand grammatical nuances, various spellings, slang and other language variations that occur online. Teaching A.I. to understand how sentiment and context can affect the tone of content is also a part of the process.”

These 42 emotions were derived from comprehensive studies led by Canvs Chief Data Scientist, Dr. Sam K. Hui; guided by research on human emotion from noted psychologists, Dr. Paul Ekman and Dr. Robert Plutchik.

Dr. Sam K. Hui is the mind behind the Canvs machine. Sam earned his Ph.D in Marketing from UPenn, a MS in Statistics from Stanford, and a BS in Mathematics and Computational Science from Stanford and has written over 20 publications.

Dr. Paul Ekman is an American psychologist and professor emeritus at the University of California, San Francisco and a pioneer in the study of emotions and their relation to facial expressions. He is the world’s deception detection expert, co-discoverer of micro expressions, and the inspiration behind the hit television series, Lie to Me.

Dr. Robert Plutchik was professor emeritus at the Albert Einstein College of Medicine and adjunct professor at the University of South Florida. He received his Ph.D. from Columbia University and he was also a psychologist. His research interests included the study of emotions, the study of suicide and violence, and the study of the psychotherapy process.

Measuring Emotional Reactions of Consumers

Experts for years have been using surveys and other tools to learn about how people feel. While this is true, producers like Sony and Disney historically were not concerned with how people feel [emotions] about content. It’s understandable. Reactions are easier to capture than emotions.

“As Canvs evolved in its capabilities, what type of metrics have you found are effective in measuring emotions and empathy?” Greg asks Jared.

“When it comes to understanding consumers, companies are more interested in learning about what people do because of how they feel [emotional reactions],” Jared replies. “And now we’ve developed and released new types of metrics for measuring emotional reactions from the customer standpoint. These metrics are included in what we call a ‘social empathy meter’. Companies like Facebook and Disney use this meter to see how consumers are connecting with and reacting to their shows and other products emotionally, whether positively or negatively.”

“Could you share with us an example of how your customers use the social empathy meter?” Greg asks Jared.

“Will Smith in the Disney movie remake ‘Aladdin’ is a scenario we’re studying,” Jared says. “As most of you know, Will Smith plays the genie, and there have been many conversations on social media about how unusual he looks. Disney wanted to find out if promoting the genie on social media would positively or negatively impact their marketing strategy. Now they use Canvs products to understand how people on social media feel about Will Smith’s appearance.”

Will Smith as the Genie in Disney’s ‘Aladdin’. Source: Essence.com

With the social empathy meter, customers can go a step further, and identify why people feel and react a particular way. This is important because while everyone experiences similar emotions, those emotions can be expressed in very different ways.

Analyzing Emotional Data Using Two Approaches

“Historically, product people have analyzed emotional data in terms of ‘positive’ and ‘negative’,” Jared explains. “This approach works well for operational data, but it’s a bad approach for emotional data. At Canvs we believe positive and negative feelings are context specific so we’ve developed tools designed to help our customers get as close as possible to the real reasons why things happen.”

When analyzing emotions, Canvs works with two fundamentally different types of data: operational and experiential data. And the research methods have two different approaches: confirmatory and exploratory.

According to Jared, at Canvs they view ‘Insights’, among other things as the ways to make connections between experiential data and operational data. To make connections successfully, they start with the confirmatory approach when data already exists and it needs to be contextualized.

“With the confirmatory approach, it requires faster validation, because we’re usually trying to see if the hypotheses we have in mind can be supported by data. In the events where there’s little or no data to support our hypotheses, we switch to a more exploratory approach.”

The exploratory approach, however, happens less often and is more involved:

“With the exploratory approach, it involves using asymmetric insights which can lead to unexpected outcomes, because we’re no longer trying to validate hypotheses. Instead, we’re trying to figure out what the data suggests. Sometimes, we want to identify which sentiments show up the most in the data.”

For example, if expressions of nostalgia appear the most in the data you’re analyzing, more exploring is required to determine the right situational contexts. Because feelings of nostalgia can cause some people to be happy and other people to be sad.

Another example is some companies want to find out if commercials they make for their products are eliciting the kinds of emotions they expect from consumers. With Canvs products they can learn whether or not a campaign that was meant to elicit happiness, ended up eliciting anger or disgust.

Sometimes analysts notice trends in emotional responses to particular shows, or to particular characters featured in shows. For example, viewers may express hate for characters in the Bravo TV show ‘the Real Housewives in Atlanta’. While consumers based on their social media activity ‘hate’ some of these characters, they still watch more episodes of the Real Housewives.

The Real Housewives of Atlanta. Source: MadameNoire.com

It’s like the lesson Jared learned earlier on — companies are more interested in learning about what people do because of how they feel. A person who says they hate a Real Housewives character is not the same as a person who says they hate the Real Housewives show. Context is critical when analyzing qualitative data, and consumers who hate Real Housewives characters is seen by Bravo producers as good for ratings.

Building a Product Business: 5 Important Lessons Learned

There are key distinctions between product-based and service-based businesses. While both types sell a product, in service-based businesses, customers buy the services provided by experts, such as a web designer or accountant. Product-based businesses provide tangible and/or interactive products that are similar in their presentation and quality for each user, making for a consistent customer experience.

Canvs didn’t start out as a product business. In 2013, Canvs was known as Mashwork, and they provided services in the form of analyses and insights for companies. They also had a beta product.

Things started out very well. TruTV was one of their first customers. They recognized the power of their technology, and they decided that they were going to transition those customers from Mashwork to Canvs.

Source: Canvs.ai

They hired engineers to build out prototypes and started raising venture capital. Things didn’t go quite as planned when it came to timing though. He assumed that the entire transition would take around six months because he didn’t know any better. In reality, it ended up taking around eighteen months.

He has no regrets when it comes to making the transition, but the journey was far from the straightforward path he imagined. Here are five important lessons Jared learned.

1. Test Your Assumptions Before Writing Code

An important part of growing a startup is understanding the market. One of the most common reasons that businesses fail is because they didn’t go after the correct market. A new business’s success relies on that business’s ability to learn, grow, and be nimble. A way to do that is to test assumptions.

“There is often a trade-off between speed and assumption testing. A product can either be rushed out the door and into the market as soon as possible, or it can be tested to make sure that it will succeed. Every idea that comes along, every decision that you think you should make, no matter what it is, is an assumption before it is tested.”

This, Jared admits, is an area that he has experienced some missteps in.

A product manager asks Jared, “With regards to the UX journey of Canvs products, what were your biggest lessons learned as you were building out these services for your key users?”

Source: Giff Constable

“One of the biggest missteps I made is thinking that I am the customer. I grew to realize that as someone who wants to build great products, I will often believe that I have a great idea and want to see it fully executed. However, that idea may not be what’s best for the customer.”

He admits to Greg that there have been many instances where a product has failed because the ideas and perspectives weren’t fully realized. As Jared says, “It felt like a good idea, but in hindsight, nobody knew why it was a good idea. We had to learn how to ‘unearth the why’ when we talk about product ideas.”

“My best learning came from massive missteps,” Jared says to Greg. He tells a story about a spreadsheet product they developed:

“Our product was minimalist and modern-looking. However, when we did our first demo, our more senior customers who did most of the user and market research, they had trouble reading the on-screen text because it was too small. After getting a lot of feedback from our senior customers, we ended up having to remake the product significantly. This caused us to lose several weeks of valuable coding and development time.”

It was then, he was reminded of what his software engineers were repeatedly trying to teach him: “Any design decision you make without testing is an assumption. If it’s an assumption, it does not matter, until you can prove it works by testing it on users.”

And now, Jared does not let his engineers code anything before building and validating his ideas with prototypes:

“Testing your assumptions builds confidence and allows you to invest deeply and make bolder moves on what gets validated. To not write a single line of code until you get validation is so critical when you have limited resources to spare. Big companies waste money all the time, but when you’re a small company, and you have 12 to 18 months of runway, you don’t have wiggle room to mess things up. It’s one of the biggest things I’ve learned.”

Source: Canvs.ai

2. Hire Designers Before You Hire Engineers

The ultimate assumption test, according to Jared, is to ask customers directly if they would pay for a product or feature. These words of wisdom, combined with what he learned about validating assumptions before coding; inspired him to make a choice that would make Lean UX fans proud.

“That’s why my first hire was not engineering. It was in design, because I learned to value the power of validating MVPs before building anything,” He tells Greg and the audience:

“When I decided to transition to a product business, I recognized there was an ‘imagination gap’ between myself and the customers. After being burned with not testing assumptions before coding earlier on, I realized that it was best to at least build prototypes on paper and show to customers, because they won’t understand your vision unless you show it to them.”

3. Co-Design With Your Customers

With the Canvs Surveys product, Jared admits that it was not his idea. They instead stumbled upon the idea as they co-designed other products and prototypes with their customers.

“We were being empathetic, co-designing services with customers like Amazon and NBC, and they shared with us that surveys was still one of the major ways companies gather consumer data. We decided to explore automated technology for survey data, testing product and design assumptions along the way. Eventually, we came up with a viable concept for Canvs Surveys.”

Co-design is a huge part of how teams at Canvs collaborate and get things done. “We don’t commit a line of code without testing assumptions, and we believe in creating a work culture around that practice,” Jared says to the audience. “We believe in taking a step back before taking a step forward, talking to customers, doing interviews and observing their struggles. The objective is to deeply understand what we can learn from these activities in the process.”

4. Let The Talent Make Decisions

When a startup is founded, the founder can end up taking on a number of different roles. They can be the head of sales, chief analyst, CTO, or anything else. The founder takes on whatever roles are needed to serve the business until talent is hired for those roles. When that talent is hired, it’s also important for the founder to pass the baton and let them take the reigns.

“A well run organization is an organization that scales. As the founder, I need to make sure that critical decisions are being made, but I also needed to learn to let the new talent make the decisions. It’s one of the reasons why I placed a priority of finding a competent product leader I can trust to make the key product decisions as Canvs continues to scale.”

Ultimately, he learned that the most important thing that a product leader can do as their business grows is say ‘no.’ Canvs teams and products need to be able to ingest volumes of data and figure out what information will be helpful, what will be possible, and what needs to be thrown away — all to ensure they remain focused on the right needs of their customers.

Source: Canvs.ai

5. As You Learn, Adjust Your Capabilities

As a CEO, Jared has to live in the future often. He is always trying to see where his business can go. Unfortunately, he admits that his vision can sometimes go a bit too far and create a gap between what the business wants and what the business is capable of.

“Figuring out how to articulate the vision is just as important as the vision itself.”

Canvs began as a service business but their core operation has always revolved around modern dialogue. He explains that technology companies have historically based their empathetic metrics on dictionaries that contained definitions of feelings and emotions. And what Jared found while working at Canvs was that this was inconsistent with how people actually expressed themselves. They adjusted their capabilities and Canvs became an insights company going forward.

Some of Canvs’ customers are companies with access to over 20 years of social media and consumer survey data. They have also invested in machine learning technology in order to generate value from this data, and they’re willing to work with Jared’s teams to discover even better ways make this happen. He realizes however that to take on the challenge of ingesting and contextualizing data sets of that magnitude, the company will have to greatly expand their capabilities.

“Becoming a product business is a balancing act between being focused on the big product ideas and staying interested what we do everyday as a company; while constantly thinking about what can be productized and monetized based on market trends, and serves real customer needs.”

Jared continues to talk to his teams about evolving products into automated services that people could use to understand emotional data and each other as human beings. His dream for example is to have something that could receive simple text and translate that to feelings in a way that’s simpler than what exists today. This is something that they’re moving towards every day.

Source: Canvs.ai

Good Data Leads to Better Products

In Canvs’ first three years, they were a profitable service business that would subscribe to third party social listening tools. Through this, they were able to discover what tools were and were not useful. They also learned how marketers, advertisers, and researchers needed to receive information in order for it to be useful to them. Even before they were a product business, accuracy always mattered.

Jared shares with Greg a story about a meeting with the American media company, ABC. He explains that at ABC they have a system that takes years of historical data and they use it to predict ratings for future seasons of content.

“And now they have been using Canvs products to enhance their predictions and accuracy in their decision making,” He says.

When he reflects upon how Canvs products was first received by customers, he humorously compares the experiences to what he imagines it would be like to mass-produce flying cars.

“As intrigued as they were, initially, people were skeptical and preferred to stick with what they were familiar with. However, once they actually used the products, opinions began to change.”

Now customers like ABC can’t imagine going back to the old way of doing things. “This is the essence of a product driven by experience-based selling, versus a product driven by needs-based selling,” He says.

Contextualizing and understanding the valuable aspects of emotional reactions is at the core of the Canvs mission. To do that work well, Jared’s teams have learned that building products that help customers recognize the strengths and weaknesses in emotional data, is tantamount to building empathy; because behind that data is people who love sharing how they feel as they watch great stories.


Product Management Insider

The exclusive data and insights that enable 12,000+ product managers to win. Subscribe via email at productmanagementinsider.com. We are powered by Alpha.

Tremis Skeete

Written by

Senior Product Analyst at Paralect. To find out more, visit tremisskeete.com.

Product Management Insider

The exclusive data and insights that enable 12,000+ product managers to win. Subscribe via email at productmanagementinsider.com. We are powered by Alpha.

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade