September 27 2018

Words by Hilary George-Parkin

Eli Katz knows his employer’s vision is grandiose. "In the absolute best scenario,” he says, “we change the way the world consumes knowledge.” We’re sitting in the 12th-floor NoHo offices of Wonder, the research startup Katz joined just over a year ago, in a conference room called Milo. It’s named after the kid protagonist of the The Phantom Tollbooth, Norton Juster's classic paean to curiosity.

Katz is the company's technical lead—though when I ask, he jokes that he's just there to play with the dogs that come into the office every day. He talks about Wonder's value proposition in much the same way one might talk about computing. "Right now, if you think about people, we're still in this stage where we're both knowledge machines and processors," he says. The way he sees it, “we’re using too much space to keep memory inside” — space that could otherwise be used for processing information, recognizing patterns, and generating new insights.

We may be past the days of scouring dusty library shelves, but even the internet, with its promise of infinite, accessible knowledge, takes time and a knack for search terms to navigate well. (How many pages of junk did you click through the last time you had to find a fact more esoteric than, say, a celebrity’s height, for example?)

To Katz, this process is maddeningly inefficient, which is a key part of the problem that he and his colleagues think about obsessively: how to make research faster, better, and smarter.

That's the end goal. To get there, Wonder pairs clients with a network of thousands of researchers (or "analysts," in the company's lingo) available 24/7 to answer any question at all, be it a competitive analysis of cloud storage companies or the mechanisms through which a peanut butter-and-jelly sandwich evokes feelings of nostalgia.

The company now has about 50 employees, most of whom work from its NoHo office, sitting at desks beside a long hallway that doubles as a dog run for Conrad, a Golden Retriever, and Kai, a toy Australian Shepherd, and any other stars of @dogsofaskwonder that happen to be visiting. It's a sunny space with vaguely collegiate touches — industrial light fixtures, a cushy chesterfield, and an eclectic library spanning a couple shelves of books, the product of a company tradition that requires every new employee to add one to the collection.

Danny Park, a software engineer who joined in March, says he chose "Dealers of Lightning," the history of the pioneering researchers of Xerox PARC, who over the course of the '70s invented the first personal computer, ethernet, laser printing, and image editing software, and more, along with the corporation that failed to capitalize on them. The real library the company is thinking about, though, is the one its analysts are compiling with each research request they fulfill. Eventually, the team hopes to turn it into a searchable (and valuable) knowledge database.

A view of the Wonder office space

Achieving the highest possible quality research in the fastest possible time is a goal that falls to Wonder's operations team, headed by Ally Sprague, a fast talker who joined the company two years ago after studying supply chain management and working as a buyer for Bloomingdale's. At the time, the company had only six employees, and every research request was handled by a single analyst from beginning to end, a system that's since changed radically.

Today, the research process is broken down into individual components. This allows the system to split up more general queries into their component parts so multiple analysts can work on one project simultaneously. The internal team used to handle all client communication, but quickly realized it wasn't scalable, so they offered up the responsibility to members of the analyst network. Wonder charges clients — who can range from consulting firms to startup founders to TV producers — $75 per question, plus a monthly subscription fee depending on the size of their company and research needs. Wonder says the same research would cost at least ten times as much to do in-house.

Ally Sprague on top of Wonder's rooftop in Manhattan

How did you end up at Wonder?

At my old job, I didn't feel like I was impacting the world in the way that I wanted to impact the world. I felt like I was putting more expensive shirts in a bunch of different stores across the country for people with a lot of money to buy. So I got on this education kick because I felt like in my own education, I had been put in boxes. I was really good at math, so people were like, "Do business." And I was semi-creative, so people were like, "Do fashion." So I ended up for two years — such good brain-time of my life — at this retail company and job that I didn't really like. I was like, "Fuck this! I don't want to be in this box. I don't want anyone else to be in this box. So, I'm going to change education." And then I went around and I talked to a hundred million education tech companies — every ed tech company I could lay my hands on. And I didn't feel like anyone was doing anything game-changing. I don't want to build Blackboard. Then a friend of mine from Bloomingdale’s introduced me to Justin and we went on a walk and I was sold in the first ten minutes. It was about helping people explore and learn what they want to learn. Education today is not at all Socratic — it's like: learn these ten things. I wanted someone to look at me and be like, "What do you want to learn?" And to me, that's what Wonder represents. You come to us with a question and we take you down your own learning journey. We're like a learning sherpa. That's freaking cool, right?

How would you describe your role now?

Our goal is essentially make the process as efficient as humanly possible and produce the highest possible quality research in the fastest possible time. Those are directly inversely correlated, so there's a bit of a "pick your favorite child" thing going on: do you want it to be really fast, or do you want it to be really good?

How do you make that decision?

There's a lot of moving around. Different sub-goals become important at different times. There are four different product operations people, and each of them own a metric: how much our analysts are earning; liquidity (how well are we matching the demand of the jobs that we get to the supply of analysts that complete the jobs?); analyst net promoter score (how happy are they? How much do they love working for us?); and quality (how trustworthy? How credible? How much do clients feel like they can come to us when they're in a pinch and we'll definitely hit it out of the park?)

How has the product evolved since you joined the company?

A lot! Research is this really nebulous task. I don't know if you've ever tried to do hardcore Internet research, but it's hard. There's no exact equation for it. It has a lot of ill-defined components and the research journey can take many different routes depending on the topic, the question, the analyst. The nebulous nature of it means it can be pretty inefficient. So over the past couple years we've gotten a lot smarter about those paths and making our machine, our research process, our analysts, efficient. We've gotten a lot smarter about defining that research equation in our product.

Most of our product changes start with talking to analysts and digging into the data. When you have thousands of conversations with analysts about their research process and then line up those conversations with the data, you can start to pick out trends in where we're inefficient and why. And then, something that felt really foggy and nebulous before, starts to feel clearer and more structured and you can actually start to base decisions on it.

We run lots of tests in Google Sheets. So we might look into: what if we could schedule our analysts? Before we build that into the product, we'll probably just test it in a Google Sheet. What's important is being as transparent and open and communicative as possible with our analysts because ultimately we're building things in collaboration with them, we're not building at them.

You must have built a pretty significant archive of research by now. Are analysts able to draw from that when they're working on requests?

We've found similar information online thousands of times — hundreds of thousands of times. One of the coolest things we're working on is how to tag all of our research so that the next time someone asks something about Bitcoin, which they've asked a thousand times in the past, we don't say "Hey Joe, why don't you start from scratch again and google Bitcoin?" We can say, "Here's what Bitcoin is, here are the 400 sources we've found about it in the past, here are ten insights that we've pulled out from past research, use these to get started."

It seems like there would be huge efficiency gains by having that information highly indexed and replicable. Do you anticipate eventually automating some of the work analysts are doing now?

Our core belief is that there is innate value in a human brain doing something. We aren't trying to replace fact-finding with this past database — we always want to be expanding it, and on the research journey itself you find new information that helps you problem-solve. The way I like to think about it is we are basically building tools for our analysts to be faster, stronger, and better than they are today — not to replace them.

How do you think about the future of the product?

We often play this game: if we're at this many analysts now, what would happen if we had 300x that number? What would happen is my phone would spontaneously combust with Slack messages. [laughs] But we think a lot about how we can have a dynamic or not-flat analyst community. Can we have self-management? Can we have admins? Can we level up our most senior members to moderate the community? We think a lot about retention. The way that our analysts talk about our community, they're not like, 'Wonder made this bad decision. I should find a different community.' They're like, 'Wonder made this bad decision. I should tell them why it's bad.' We have analysts that write pages and pages of plans for what we should build because it's worth it for them. They want to be here a long time and we listen to what they say.

Justin Wohlstadter likes people who ask good questions. "Why?" is practically a mantra at Wonder, which he founded in 2012, a little more than a year after finishing grad school. The idea for the company stemmed from the thesis he wrote there on the future of education. In its earliest iterations, it was a free educational service provided to students and researchers, with the analyst role filled by volunteer librarians whom he cold-called after completing the first hundred research requests himself. Between then and now, he says, "we wandered in the desert for several years building multiple versions of the same thing in much worse ways until we finally figured it out. But that period of wandering in the desert built a lot of great muscle and understanding of what we wanted to achieve and what we wanted in people in our organization."

What do you see as the company's fundamental values?

In our own internal team culture, which of course leaks over into our entire product, curiosity is probably the most fundamental value. You have to be curious and you have to ask why. I grew up in an Orthodox Jewish background and there's a saying, 'two Jews, three opinions.' That's very much the ethos we have here, which is we debate to find truth. We're very opinionated. We do it by asking, 'Why, why, why?' And fundamentally understanding the reason from a perspective of curiosity. I think it's definitely not for everyone. I studied the Talmud from a young age, which is basically like Supreme Court case law on steroids. And the purpose of it is to get you thinking and get you to have an opinion and debate that opinion in a way that doesn't prove that you're smarter than everyone, but that comes to truth and understanding, or a greater combined truth. And so, yes, we debate things, we're very logical, we are not afraid to express opinions, and we keep throwing back and forth an idea until we come to an understanding of what the right decision or answer is.

What mistakes have you made so far?

There are thousands — I as a CEO have made a ton of mistakes in terms of hiring the wrong people and firing and managing and that kind of stuff. We as a company have made mistakes in terms of what we focus on from a product perspective. How we communicate with our analysts, how we treat our clients. We have screwed up probably every single possible aspect that we can, but we move very quickly, so when we realize a mistake, we try to course-correct and fix it. Part of the downside to rolling out changes so quickly and moving so quickly is that if you don't have very good communication mechanisms in place, people feel left behind, and when you have an analyst network of over 6,000 people all over the world, it becomes even more difficult. We had a situation last week where we hadn't communicated where we were going in terms of how different jobs should function, who we were looking to do different jobs, and so we had many upset analysts. So we've taken a step back in the past week and built a document for what are our principles for communicating with the analyst network. How do we see ourselves in relation to the analyst community? What do we prioritize? What do we not prioritize?

How do you see yourself in relation to the analyst community?

Ideally we see ourselves as part of that community. We have a community that, like our company, is inherently curious, and love the fact that every day they get to learn about the craziest, most random things. The more that we can treat everyone as a cohesive unit and the same, I think the better off we are. But it's really hard, because there are analysts who are there for different reasons. There are people with different cultural backgrounds

How do think that relationship differs from other companies in the so-called gig economy?

We do a bunch of things that most other companies purposefully do not do. And I don't know whether we're right or not. [laughs] For example, all 6,000 of our analysts are in one Slack. I guarantee 98% of gig economy companies do not want their supply base to communicate. Uber, specifically—not only do they not have chat, they don't have forums. They have nothing. So their people self-congregate in private Facebook groups.

We feel like we're a community. Not only do we have a different cultural perspective, but the work we do is collaborative. When someone drives you in a car or delivers you a package, it's an individualistic thing. We have a global, beautiful community, and they help each other. We actually provide people with fulfilling, intellectually stimulating work. Analysts at Wonder are proud to put it on their LinkedIn. You don't often want to put that you're a driver for Uber on your LinkedIn. Not to mention the fact that every hour you spend in your car, you're wearing down your car, you're wearing down your back. Every hour you spend on Wonder, you're learning, you're generating credentials.

We did an analysis of our top 50 analysts, our youngest is a 18-year-old  in Wisconsin. Our oldest is 84 in Dallas, Texas. A 18-year-old is working alongside an 84-year-old to produce a piece of knowledge. That kid goes to college, applies for an internship at Bain or McKinsey, and can be like, 'Look at the market sizings I've done.' We can credential people and give them incredible ways to learn and earn, and I think that's so powerful and something that's very rare in the flexible work environment.

What's the biggest challenge for you right now?

Growth! It's really hard. I've never done this before, and I think communication issues increase exponentially with linear growth in people, so every few weeks we have to figure out what are the best ways to communicate things across the company. We have a very cohesive team. We're very clear on who we hire and why we hire them. You can ask anyone what our cultural values are and they would be able to tell you. We don't have ping-pong tables and foosball tables. We don't have stickers everywhere with sayings and quotes and whatever. There's a reason we're not in San Francisco. [laughs]

So New York was an intentional choice?

Definitely. I'm from Dallas, Texas, and I'm sure we will have a second office at some point. Maybe multiple offices. I would love to be in a place with more grass and trees. But New York fits very well philosophically with how we feel the company should be built. It is the knowledge capital of the world: all of the incumbents, all of the knowledge companies, research companies, every major industry, the biggest concentration of the top universities is in the Northeast — so in terms of diversity of thinking, obviously this is incredible stuff.

What are you looking for in an employee?

When we're hiring, we're looking for people who are really thoughtful and genuinely curious. We have an infinitely complex product, so if you can't come up with at least a few really good questions, it either means that you're not interested or you don't know how to think about these questions. And for tenacity, you can also see it in what they've done in the past: have they really gone deep in whatever they've done? Have they shown that they can work somewhere for a consistent period of time and get promoted in that job, no matter how shitty it was, and at least try to squeeze that lemon juice out of that lemon? We're very focused on solving a problem around knowledge accessibility.

We find people we believe emulate our core values around curiosity and humility and tenacity and positivity, and we stress-test the hell out of that for every person we hire. It's taken a long time to get here. I genuinely love every person in the company. Yes, they're smart and they're hard-working, but they're also just great people.

"We feel like we're a community. Not only do we have a different cultural perspective, but the work we do is collaborative."