Interview: Braden Ream, CEO of Voiceflow
The Ivey Business Review is a student publication conceived, designed and managed by Honors Business Administration students at the Ivey Business School.
Braden Ream is the co-founder and CEO of Voiceflow.
Prior to the amazing success you’ve seen with Voiceflow, you worked on several start-up ventures throughout your University career, including social network Flare and ridesharing app Yugo. What motivated you to begin these projects, and what characteristics do you see in yourself and other ambitious entrepreneurs that led you to continue this path?
I had three or four start-ups at university and my biggest one was Flare. Flare was what taught me everything. Yugo was a side project my friend and I built; you’d think a ridesharing system was one of those ideas that’s already been done to death, but we thought we could do it better if we had some great design operations.
I’ve always been really interested in building things, and there’s nothing cooler than building a company because it’s pretty meta — you’re building a thing that builds things. I’ve always loved the idea of leverage, which is explained in the essay How to Make Wealth by Paul Graham. You want to have a multiplier effect on your time.
I knew one hundred percent that I wanted to work in entrepreneurship after my first summer internship. I worked in finance doing sourcing procurement, and at one point my manager tasked me with reading five hundred pages and looking for discrepancies. I only had one month left in my term; given that our normal sourcing projects were four months-long I knew they were just looking for something for me to do. I immediately thought, “I don’t want someone to have leverage over a month of my time without me being able to make an impact.” As an entrepreneur, if I want to try to increase our company’s revenue, I can just make sales for an hour. I can hop on LinkedIn and talk to prospects or try to upsell an existing customer. You have the leverage to impact change, especially when you’re a small company.
The feedback loop is so tight when you have just three or four people because you can just hop on a call with customers; if they want something, you can build it over a weekend and then ship it on Monday and charge them. The bigger the company, the greater the impact of your time.
There’s an inverse parabola with respect to size and leverage on your time, because when you’re beginning you have maximum control of your time; once you get customers, they become your bosses. And if you become like Tobi — Shopify’s CEO — you kind of make it to the other end of the spectrum, and you’re your own boss again.
Beginning a new venture can be both challenging and rewarding. How and when did you ultimately decide to move on past Flare and Yugo? How should aspiring entrepreneurs think about when it’s time to move on past their first idea?
Losing Flare was tough. At the time we began to abandon the project, we had launched the previous year, and while it didn’t go that well, we still had momentum. The team was so excited because we had some initial results at Western; it got three thousand members within the first 48 hours. I got a notification on my phone every time a new user signed up, and my phone was sitting there, just buzzing. As students, however, we didn’t have the product philosophy to make improvements at a fast enough rate; Flare was all about creating instantaneous and spontaneous events to then find on campus. It got very heavy usage during O-Week when everyone was using it to meet people. Unfortunately, once people had met people the fundamental thesis of the business stopped working. People want to have fun with friends — not to go hang out with a bunch of random strangers.
I’m an introvert and I would never have used Flare the way I intended it; I just thought other people would want to use it. People used it to find parties and things during O-Week, and after three weeks it was mostly dead. I learned that great entrepreneurs must understand their target consumers well, and relate to them to understand their wants and needs.
Every app launch, the app was changed in some way; after the third launch, the engineers weren’t really working on it because they saw the writing on the wall. We didn’t want a repeat of previous launches, so we decided to create an event-based chat. The original chat feature was built on a 50-metre geofence for each event and would enable people at the party to talk to those outside and encourage them to attend. We kept trying to find new events because team morale was dying.
Once you’ve launched and failed a few times, it becomes increasingly hard to motivate the team that the next one will work. With social networks, you have to hit a critical mass of users and maintain a participation rate. Very few people wanted to be the event hosts, but a lot of people were looking for events; the ratio was mismatched. We didn’t hit critical mass and we didn’t have enough events to sustain all the listeners.
In the end, I never officially closed down Flare; I just stopped working on it. I moved on pretty quickly, starting what would become Voiceflow just two months later.
The entrepreneurship journey is both exciting and uncertain. You and your co-founding team at Voiceflow — Andrew Lawrence, Michael Hood, and Tyler Han — originally envisioned the company as an interactive children’s entertainment company. How did the company’s goal and vision evolve with time, and how did you ultimately decide to create a no-code smart skill system instead?
I view entrepreneurship as one continuous journey, constantly learning something different. The most important lessons I learned from Flare were that industry matters a lot and that you need to be solving a real problem.
It felt like a breath of fresh air after Flare failed because I could take all the lessons and apply them to a clean slate. I knew how to recruit engineers and so we were able to build the Voiceflow product quickly. If Voiceflow was to fail today, I’d start a new one, and I wouldn’t make nearly as many mistakes in the financing and managerial hiring as I did on Voiceflow.
I would view startups as a version of Maslow’s Hierarchy of Needs. The foundation is the industry, which is the most important thing by far — even more so than an idea. How big is it? How fast is it growing? How concentrated is it?
The next thing you need is the team. If you have three PhDs in machine learning, but you’re trying to start a CPG lip gloss brand, it won’t work. The industry you’re in determines the team you should have.
Your entrepreneurial idea is last. You should rarely change the foundation, frequently change the team, and change your ideas weekly. With Voiceflow, we picked the industry of conversational AI early-on. Then we picked the team: a technical co-founder, design co-founder, and business co-founder. The idea has been changing every single week since.
When people hear the word pivot, they think 180-degree pivots. True entrepreneurship should have a few 180s, but for the most part, as you get closer to a product-market fit, it should be 15-degree optimizations. How do you talk about the product? Who’s your ideal customer profile? That should be changing as you learn more. That’s my philosophy.
Within conversational AI, we initially focused on the niche of Alexa and Google Assistant. At first. We were just trying to come up with ideas for the space given our team and industry. A voice commerce app inspired by Honey that optimized price comparisons from Amazon and Google was our first idea, but we quickly realized it wasn’t technically possible. We then started pursuing interactive children’s stories on Alexa to help kids learn to read, because smart speakers are going to be everywhere, and there’s a strong correlation between educational outcomes and parents reading with their kids at night. We thought that Alexa could fill the role of parents here, and that app — Storyflow — actually did quite well.
We raised fifty thousand dollars for that idea but faced a really interesting unit economics problem which made us pivot: interactive audio content didn’t exist yet, especially for children. We had to produce all the content ourselves. You can imagine that if Netflix had to produce Netflix originals and distribute content from the beginning of the company, its economics wouldn’t have worked and it would have gone bankrupt. Given voice acting and other costs, it probably cost $3,000 — $4,000 per story and we had $50,000 total, so doing one story per week would kill us in around ten weeks. About five weeks into that, we had probably $20,000 to $25,000 left and we realized we weren’t making money at a fast enough pace — we hadn’t even introduced monetization.
What we did have was one of the top voice apps in Canada and around three thousand families listening to us every single week. It was going really well, but the economics didn’t work. This made us think, “What if we made it easy for other people to make stories and then put them on our platform?” Unfortunately, this required children’s authors to code each story. We needed to create a platform to make it easy for both us and individuals to create content so we could make user-generated content feasible.
We built the first prototype of Voiceflow in 48 hours. It was just a visual interface to put together interactive children’s stories that then get published to our Alexa app. We then started onboarding children’s authors, and we found that they weren’t very technical and it still wasn’t working. So, we just kept using the tool ourselves. I made a post in an online developer community about this platform that we built, and a bunch of developers asked if they could use it for their own projects. That was when the light bulb went off: we could become the Wix for voice apps. We raised half a million dollars for this idea from Ripple Ventures — a leap of faith from people who believed in us as a team.
We’ve been making 15 degrees every week since, and the most major change is that we’ve realized that professionals and enterprises need not just Alexa and Google support, but also a creative set of tools to design prototypes and collaboratively build conversational interfaces across any channel. Call centres, web-chats, in-car voice assistance, automated drive-through, and automated retail experiences all require really good tooling that doesn’t exist. The majority of our revenue is now outside of Alexa and Google.
We still power about two hundred million conversations a year on Alexa and Google, so we have some scale there. But the majority of our business is now shifting to becoming Adobe for conversational interfaces right now.
When we raised our seed round — $3.5 million — we could finally afford to hire senior engineers and build a proper engineering team. We’ve only recently finished the true architecture that can actually scale to power billions of conversations. A lot of it was just rubber bands and duct tape in those early days.
Your founding team originally applied for Y Combinator twice but walked away both times empty-handed despite making significant progress and raising capital. What were the keys to your rebound and success moving forward as a team? Once Voiceflow was up off the ground, how did you divide responsibility between each member of the team?
If Voiceflow was ever to break, it would have been when we got rejected from Y Combinator for the second time. We had a co-founding team that really wanted to make this work — if you do entrepreneurship for the money, you’re going to fail. As entrepreneurs, we were in this for the long game.
Here’s a framework to consider when making these types of decisions. There are two types of wealth: power and capital. Power includes what you know and who you know; if you were best friends with Elon Musk, you would just get wealthy by association and hearing information others don’t get access to through your network. Similarly, if you were ridiculously unpaid for 10 years but you ran a startup that was on the cutting edge of machine learning, you have such a wealth of knowledge that you can convert that knowledge into capital. This is why presidents like Barack Obama instantly become millionaires when leaving office. Money doesn’t convert nearly as easily into power. If you want to play the long game, don’t worry about chasing early capital. For instance, we chose to become broke university students for four years because we’re investing in the power of our education and networks.
So, bouncing back from this loss, we just buckled down the next three or four months. Now that we had this idea of Voiceflow, we were way more passionate and the new idea kept us going. We kept in mind the division of responsibilities, as doing this wrong is the number one thing that kills startups. We got very lucky in that we kind of fell into our own roles: one of us was clearly the technical co-founder, another was the design co-founder. I wasn’t either of those, but I drove them harder than they did me and I was a better public speaker, so I became the CEO.
When building a team, you want to have a local maximum for each skill set. The worst team is three people who are all sevens across the board because your quality of work is going to be a seven in all respects. Startups can only have one brand, one line of code for that particular piece of your product. Brute force doesn’t matter as much as you want to have one person who is a ten in each area; you want complementary skill sets.
The early stages, before we got external validation from customers, were challenging. Successful VCs told me it was a stupid idea. You need to have a crazy conviction that your product is a good idea because if something was obvious, everyone would do it. The whole reason VC funding works is that they’re able to pick unobvious ideas, which then can have outsized returns.
In the end, we actually got into Y Combinator on the third try, but at that point, we turned it down. I wish they’d let us in earlier. We would have done it, but it didn’t make sense at that point. We were already on our way.
In your final year of study at Ivey, you left the program to take on the challenge of leading Voiceflow. That was a big risk that’s paid off hugely! How did you think about making that decision, and how did you know when to commit full-time to the company? Did you face any resistance or doubt from yourself, your parents, your friends or classmates, and if so, how did you overcome the pressure to take a more traditional route?
My experience from Flare taught me that it’s really hard to balance school and work. You can’t really focus on doing great work when there’s an exam hanging in the back of your mind, even though you could probably manage both workloads. Since entrepreneurship is so competitive, I didn’t want to be split-brained. If you want to be amazing at something in life, you have to just do one thing.
What I actually did is I give myself an internship. Ivey lets students take a year off to go to work, so I wasn’t dropping out but rather pushing the program off a year and giving myself a year of optionality. My framework was that for every choice there is an upside and a downside — an opportunity cost. Taking a year off school provided a lot of potential upside because if the company worked, I would be financially set for the rest of my life, gain a ton of experience and do what I always wanted: build a company. The worst-case scenario was that I’d have a great resume and an awesome learning experience. That downside was that I would graduate just one year later.
I didn’t view it as very risky, and funnily enough, it didn’t feel like a big decision at all. So when I met my co-founders who shared this passion, I thought, “I have to take this opportunity, right? I might not find people who are the same age as me, super ambitious, and have the skill set to actually push this to the moon in the future.”
As the leader of a rapidly growing tech company, what key lessons and skills did you pick up at Ivey that have helped you in your new role?
The best educational experience I ever had was FinFun (Financial Fundamentals). I loved it. It had managerial accounting baked into it; balance sheets and financial statements tell stories, and you get to be a detective. By comparing financial data, you’re able to find insights about business models and determine competitive advantages.
It taught me that how you make money — your revenue model — is much harder to change than other elements of your business. You can change your logo and your name, but how you make money determines the features that you build and how you can serve customers.
I think the biggest thing Ivey teaches you is how to be at business school; there is a language and a culture to how people interact with each other, and how business is done. The mechanics of the schooling teach it. Often, the structure of how you’re taught is more important than the actual content. I might not remember much of the stuff I learned, but I remember how job interviews were done, the pressure of who’s going to get the interview, and how to network. Soft skills are the really important stuff.
Voiceflow has been an unqualified success over the last few years, serving over 60,000 users and powering 200 million conversations, and securing brand name clients like the CBC and the New York Times along the way. As your team has grown rapidly and you continue to raise new capital from external stakeholders, how do you balance rapid growth with consistent direction?
Often, it comes down to a long-term versus short-term mindset. The death of a lot of startups is a massive customer coming in, saying “We’re going to pay you a million dollars a year, but we want custom modifications.” If you spear a whale that big, you’re going to go overboard because you can’t handle a client of that size and their expectations and demands. When you’re a small team of four, taking on a huge customer means that you can’t actually work on the core product — which is what got you that customer in the first place — which means you’re not going to get new customers. You basically become an agency or service business, and service businesses have low multiples on their revenue because they’re not scalable.
Technology businesses should be infinitely scalable. You need to always know what the long-term goal is and don’t compromise the long-term for the short-term. Ensure you have enough cash and short-term wins to achieve the long term, but don’t sacrifice the long-term for the short-term. It’s a balance; you want to do the minimum viable job to achieve your short-term objectives while making sure that you’re working towards the long term.
Managing customer expectations is just one of three stakeholder relationships that you must prioritize as an entrepreneur: the two others are investors and employees. The relationship between entrepreneur and VC becomes tenuous as the company grows because VCs don’t actually have the same incentives as you; they have overlapping incentives. A VC sits on a dozen boards and just needs one company to hit a billion dollars to make their big paycheck. They can only sit on so many boards at a time, so they have the financial incentive to push you to go as fast as you possibly can to either hit a wall or become a unicorn. If things don’t work out, they’ll just place another bet. As an entrepreneur, your incentive is to take enough risk and build a big company, but not too much risk to the point where you burn out. VCs, meanwhile, want you to take as much risk as humanly possible.
You must also manage your employees well. Right now, we have a launch coming up in two weeks a new product suite. We’re behind a little bit on our sprint in our agile development cycle. Do I push them to work on Family Day for a short-term gain, if long-term one of those people might be a little bit more disgruntled and leave?
What have been the biggest challenges you have encountered throughout this journey? What have been the biggest learning experiences?
Recruiting was a big challenge because I hadn’t done it before and it was really tough convincing people to quit their jobs. Recruiting for a club has a minimal opportunity cost; you can join as many clubs as you want. But you can only have one full-time job. Also, as I mentioned before, fundraising was a big challenge.
Through this journey, my biggest learning was that your business model determines how you make money and much of the company strategy. For example, if Facebook had chosen to be a subscription-driven service instead of using an advertising model, their objective would be to create the best possible social networking experience, rather than taking all of your data.
Another thing that I’ve learned, even though it’s hard to grasp, is value creation. Why does someone pay for your service? The truth is people pay for things because they want to either achieve an ideal in their life — it’s a vitamin — or they want to reduce the pain in their life — it’s a painkiller. This understanding is fundamental to sales and branding. It’s also why Flare failed: because it didn’t reduce a pain point or achieve an ideal for very long.
Where do you see the future of interactive voice experiences and audio going into the future? How do you believe Voiceflow will contribute to and enable this growth?
I would define our industry broadly as conversational AI, and one aspect of this is audio or voice interfaces. In this sense, an interface is an input/output system.
Audio has been challenged in the digital age because it’s slower to evolve than text, video photos. It’s particularly interesting with respect to interfaces. Voice has one of the best input speeds — it’s faster to talk than type — but it has a really slow output speed. For example, imagine you are picking a Netflix movie for tonight and you know exactly what you want to watch. Typing on remotes is a horrible experience because visuals require a peripheral to accept input, whereas a voice interface can do both input and output. Yet if you didn’t know what movie you wanted, using a voice interface Netflix would list out 15 different movies and a 30-second description of each. You’d have cognitive overload because the data isn’t coming fast enough for you to remember all of it.
The distinct advantage of voice interfaces is the lack of a physical peripheral. I can be across the room and activate my smart speaker. It’s also a one-to-many interface; I can have one speaker and 10 different people all interfacing with at the same time. As a result, I expect more multi-modal experiences that leverage these complementary strengths, as well as more command-driven applications. On the multi-modal side, imagine ordering an Uber with voice using my Echo and then confirm the ride and features on my phone. On the command-driven side, think “Starbucks, order me a coffee”; it’s a direct query that takes advantage of the fast user input speed. Multi-turn dialogues and conversations are less engaging.
Relating to the rest of our business outside audio, however, I think we have a huge market to tackle. If you asked someone in 1910 how big the automotive space was, people would tell you it’s tiny because they would be misappropriating what industry they’re in. Cars were really in the transportation space, about to displace all the horses out there. They are just one medium to achieve the goal of transporting you somewhere else. People looking at Voiceflow’s Google apps do the same thing. You might think that we’re in a small industry, but we’re in the talking-to-people industry, which is pretty big. Do you know anyone who’s talked to a retail worker or drive-through worker or call center worker? Automated conversational AI is the car to the outdated model of programmatic conversation. While AI is not as flexible as humans, as the technology gets better, you will hit a critical mass and it will take over conversationally. We will impact basically any industry which has programmatic compensation — low-level qualifying sales, anything where people say the same thing a thousand times every day. These conversations can be displaced by a voice interface, a chat interface, or a multi-modal experience. The future of the industry is this platform shift for all conversations, not just Google and Alexa, which are facilitators. We’re just getting started.