Building Instagram for Doctors

I had made this group of friends, and we were investing a little bit of money in the stock market and a few other securities as a group — until they changed the law that fall and made what we were doing illegal, just the style in which we were doing it. So we had to stop and basically give everybody their money back. And in the last quarterly update, me and two of the guys were talking about the sort of things we might do as people who wanted to do small amounts of investing. Were there companies we wanted to invest in, or start?

And I mentioned this idea, and it was one of those moments where you hear a clattering of a fork and then silence as people look at you, and you say, "What did I say?" And everyone's like, "I'm pretty sure that's a good business idea." And I was like, "Thank you — but what do you want to do about it?" So my developer friend said, "I think I know how to build a very basic version of that. I could do it in probably a month." And my other co-founder-to-be thought that if Richard, the developer, could build an alpha version of the program, he could show it to a few family friends and potentially raise enough money for us to get a small office and maybe hire one person.

So we both did those things. We sat down and decided what the value proposition of this tool would be, how we'd describe it to people, and where the possible market for it could be. My friend was able to raise the money. We incorporated the company, got a small bit of office space at an accelerator, and then six months later the app was in the App Store. I think we had 20,000 users in the first month.

Great question — you've asked not just one question, you've asked all the questions rolled into one. You're right that there are lots of things that are useful once they exist, but there's no clear path to getting them created, because there's no capital or time you could devote to creating them. When I meet with medical entrepreneurs, a lot of the editing of those ideas has to happen at that very early stage, and often doesn't.

You've got a longer education than most people already, and you've still got lots more to go. But a lot of folks come hot out of their undergraduate degree, launch themselves headfirst into production, and when they're done they've got a version of their app, a deck, and a suit, and they come into the office ready to pitch. And you say, "Who's going to pay for this?" And they're like, "I don't know, we've got to do research." My man — that goes first.

So we learned there are essentially two ways you can raise money, two types of success, either of which is sufficient. You can raise money because your product is growing and finding product-market fit very quickly — you've made something with obvious appeal that people want to use. Or you've made a thing that makes a lot of money. Something's got to be growing — either your revenue or your user base.

Our concept to grow the company was to do it on the back of user engagement — whether there were signs of product-market fit. We decided we'd try to grow that first, because you have to decide early. We had 10,000 users in our first week and 20,000 after the first month. Growth was very, very quick. So we leaned on the growth of our network to provide network effects — to improve the quality of content on the platform — but also, that was how we were going to raise money, certainly for the first few rounds.

Your unit economics for user acquisition have to be correct when you launch a growth campaign — you can't be growing for more than it costs to make your money back from those users. But since our growth was very inexpensive, and healthcare professional users tend to be very valuable, we were happy to defer revenue for the first bunch of years. And our investors were satisfied with that plan too, as long as it was working.

It depends on what business model you select. Figure 1's business model now is that we build custom educational content for non-governmental organizations and biopharmaceutical companies. This was new to healthcare advertising when we created it. Rather than use a monetization avenue that was well-trodden but not very lucrative, we wanted to build something less capital-intensive to run that would still earn us a sufficient amount of revenue.

We went to the market — these NGOs, like Doctors Without Borders, and biopharmaceutical companies — and said, "We'd like to create this educational piece that relates to this disease you want to educate people about." So we create these programs which you can call ads, cheaply, if you want to.

They're more akin to sponsored content, the native ads you've seen on Facebook and Instagram for almost a decade now. We basically created something new for healthcare. Before, healthcare ads were only banner ads online — there was no engagement content, no making sure the content is interesting or delicious to physicians' brains.

That was a lot of how we got our early traction. We created a podcast — it's now one of our big revenue drivers — called DDx. We made the first season out of our own pocket to see if we could create an interesting medical podcast, and it was the number one science and medicine podcast every week it was released. So we took that and said, "We'd love to do sponsored podcasts on this topic or this disease," and some of our partners said, "Yeah, we'd like to get into that." Now I think we're about to launch season five, where seasons two, three, and four were all sponsored. It helped grow the revenue by focusing on a new way to deliver content that's supposed to be interesting to physicians anyway.

For a brief period of time, probably. One of the things important to the Figure 1 ethos is that we know physicians don't have a lot of time to spend on the platform. If we're going to give them adverts, they should at least be good, interesting, thoughtful, or educational — make it worth their time. And what we found is the physicians who see these ads tend to like them as much or more than other content on the platform. There was a period where the top three cases on our platform were all sponsored — which isn't surprising, because they have better artwork and better writing than most of the regular posts.

Sure. A couple of days before launching — maybe the week before — we called all our local media and got in touch with whoever might be willing to write a piece about a new healthcare startup. We explained what we thought the value proposition was. When we launched, we wanted Figure 1 to be a layer of physician and healthcare-professional discourse around the world.

Starting in Canada, fine — but anywhere you can download an app, because you've got internet and you've got a patient in front of you, you can send that patient's clinical information to any physician in the world and get feedback immediately. We have the technology to do these things, and we could make healthcare instantly better anywhere there's a silo — whether that silo is socioeconomic, geographic, or simply that the hospital on this side of the street is run by a different corporation than the one on the other side. Removing those borders and giving people the opportunity to talk about cases with their colleagues felt like a very creatable way to open up the discussions in healthcare.

It helps for real-life healthcare — there are lots of examples of that happening. But it also creates an educational platform where two students can watch and listen to expert cardiologists text back and forth about whether the QRS has an elevated J point, and whether that's a significant ST elevation needing intervention or just a tachycardic young person. Those one-off conversations where there's a real pearl in there — you get to watch them unfold if you're a student or a learner. I remember being in residency and wanting residency to just be that: show me my attendings when they disagree with each other, and I'll learn everything I need to learn.

Well, this was 2012. The iPhone had only been introduced in 2007, so it was a five-year-old product. Instagram was maybe only two years old. If I'd said "It's Instagram for doctors," I don't think most people in 2012 would have known what that was. And, honestly, until it was being shouted at full volume on the Today Show, I don't think we really embraced the idea that "Instagram for doctors" is the right way to describe Figure 1.

That's probably because our mission was to democratize the knowledge of medicine — not to be superficial in some way.

We were very cautious about how people were going to perceive our brand. We knew people could upload gratuitous or eye-catching, gory images that wouldn't be appropriate to share — either because you're sharing something purely for the spectacle of it, or because these are the most vulnerable moments for patients. Having a modicum of modesty to apply was something we wanted to figure out how to message correctly. So, no, I didn't love "Instagram for doctors" as a description — but that's what people call you, so you live with it.

It's really interesting you say that, because I've definitely led you to believe we had early and very robust product-market fit. We had early adoption — a lot of it. To the point where we were making bets it wouldn't be that high early on, and lots of people lost those bets. There were people at Figure 1 who earned a trip to Paris because the co-founders all felt there was no way we'd get that many users. But we did. And then a few years went by and growth kept up. We were on the front page of BuzzFeed at one point for some of the images shared, and that nearly broke our servers with the amount of downloads it caused.

I think it gave us an overly optimistic view of what our product-market fit was. It wasn't even that we had a high churn rate. It was more that sharing cases isn't something people do as much as we'd hoped in an unprompted way. So there's been a lot of product work still being done today, five years later, to shore up and retain the interest of those valuable healthcare professionals. We thought we were doing amazing, but I think we just had a really good story that did better than our product. Then we had a lot of product debt to catch up on and make sure we were actually fulfilling the promises we were making with our marketing.

Well, some of it is that the problem is maybe a different problem than we thought. The problem isn't that doctors aren't sharing enough cases — it's whether they feel safe and secure in the privacy of sharing their opinion. So some of it was creating an onboarding process that reminds and reinforces our privacy boundaries, and helps people appreciate who their audience is. For a while we were letting in almost anybody who wanted an account. Now it's licensed healthcare professionals only, and you have to prove you've got a license before you can sign up. People were worried there'd be lawyers, politicians, and journalists lurking to find stories, and they'd be exposed in some way. So some of the problem was fixing those privacy concerns.

Some of it was giving people a reason to upload — what benefit do they individually get. And some of it was just recognizing that we're not going to have a population where most people are uploading cases most of the time.

That's not the case on any media-based platform. Even on YouTube they talk about the pyramid, where 90% of people are lurkers, 10% will comment and like, and only 1% actually create the content. You have to be okay with that kind of ratio — but sometimes it means you don't get the content you thought you were going to get.

There are two or three most common reactions. The first is probably people saying, "Oh yeah, I thought of that idea too." Okay, fine, great — where's your account? The second is, "How do you deal with the medical-legal stuff?" The answer is: we didn't start creating this product until we knew what to do medical-legally. We made sure we've got consent forms built into the app, that we're not handling any protected health information, that any identifying details for the patient are not on our servers.

Just like the clinical images in the medicine section of the New England Journal of Medicine, we're showing you a medical piece of information — a picture of a wound, a lesion, a surgery, a finding — disconnected from that person's personality and life. Their personal details aren't on our server, and we're not permitting them to be. We built something called the queue — a privacy queue — where trained moderators manually examine every case before it's made live, to make sure privacy violations don't go public. Every case can be flagged immediately, taken from circulation, and re-examined at any time. We built literally every precaution we could think of into this. So most people say, "Yeah, that does sound pretty good" — but it doesn't mean they'll feel comfortable using it. It just means they can't think of any other ways to make it stronger.

No, it's not something we've thought about or considered.

That feels like what I've learned to call an edge case. Give me a call when that software works and is out there in the public, and we'll see if that happens. The very first time we think it might be having an effect on our privacy, we'd probably intervene — but only then, not in advance.

By the way, you're mentioning one of my least favorite techniques of practising medicine, which I think is also a fine way of practising in business. It's the practice of asking, "What do we do if this happens?" But really, unless you actually have a problem, you're not meant to be doing anything, because all you're doing is adding risk. In both business and medicine, if you add layers, spend time, spend money, use resources, and expose people to medications, you're creating risk that didn't exist before — because you think there may be a different risk-benefit relationship in the future. You're altering your current practice to optimize for a future you don't know.

And most of the time you're wrong about the future anyway. So why not just wait until it gets here and then solve the problems then?

It was happening. And one of our board members — one of our lead investors, from MedRounds, and a big champion of our site — basically applied the philosophy in his Twitter bio to our network, and it solved the problem. The line from his bio is just this: real names beat proof. You put people's real names on there and they settle right down.

But yeah, we did have a few. There was actually a small episode a few years ago where we tried to enforce some of the community guidelines, and this small group of people decided they didn't want to be bound by them. This was their home. And, by the way, it fills me with both pride and fear simultaneously — you feel amazing that you've built something where people feel at home and feel like they belong there more than you do. Great, it means you did a good job building a community. But they want to make inappropriate jokes, be lewd, say things you wouldn't say in front of a patient about the patient. And it feels like now you're crossing the boundary of professionalism. We do expect our users to act the way they'd act in a work environment. If you don't think you should say something in front of a patient out loud, why would you think it's okay to say it in a publicly viewable forum that anybody could use?

So we wanted to enforce our community guidelines, and a lot of people got very upset. A bunch of people quit the platform. Some of those users are still around, but most of that group left. It was tough, because these were some of our best users — really engaged, in the app every day, leading conversations. One of them was the best radiologist I think we ever had on the platform. But if he wasn't allowed to make sexually suggestive comments to other people when they got chatty deep in some threads, he wasn't going to stay and teach. So we just had to part ways. A fair difference of opinion.

Some of the worst money we spent growing was spent on social-network ads — Facebook, LinkedIn, Instagram, that sort of thing.

"Sign up for our email newsletter and see interesting cases," or "What's the diagnosis of this x-ray?" It would perform pretty well as an ad, but those users would show up once and never again. Probably a million of those users showed up through ads, and if 10% of them stayed, that's probably an over-exaggeration.

So we learned that method works in bulk but doesn't get you users who'll help your revenue grow. It's great to be able to boast that you have a lot of users, but anybody you meaningfully need to convince is going to ask, "When were they last on the platform? When did they last open an email? Are they all verified?" Well, no — a lot came from countries we couldn't even track because they were using VPNs, so we didn't even know what country they were in. Out of the three million users, probably a third will never be useful to us in any meaningful business sense.

The hard part about growing a physician-engagement business is you basically have to go out and get the physicians nearly one at a time — meeting people, showing them the platform, getting them to onboard. We can do that at conferences, but not really anywhere else. We built an ambassadorship program that was very helpful for growth early on. Ambassadors were medical students — I think at that time just medical students — representing their class to Figure 1. They'd compete for who could get more users from their class to join their class group. Doing that won them swag, like a beautiful Figure 1 t-shirt, or stickers, or, if you got to the top, an Amazon gift card or something.

That ambassador program — having people who knew the platform very well bring it to peer-to-peer discussions — is where most of our growth came from. So don't underestimate getting people to get others to sign up. That seemed to be the most helpful thing we ever did.

I think you're right that if you're writing marketing copy, you want to sell the benefit of your product, not the use of it. If Figure 1 were writing that copy, it would be something like selling you the benefit of being the doctor who has seen a case of everything — always know what to do, because you've seen everything. There's nothing new under the sun. See one case of every disease. So to me, the benefit of the platform would be never being surprised, never feeling anxious, never not knowing what to do — because that's the insecurity that drives a lot of doctors to want more education. That's what it'd be off the top of my head. But, like anything else, I'd want to test to see what would work.

That's too exciting. Too exciting.

We've just shipped something I wrote on a blackboard in 2016 — the plan for this piece of technology. It's now in the app, shipped. So anyone hearing this can go see what I thought the future was going to be six years ago. The tool is that we've built a database so that when you upload a case — take a picture, type a caption — we scan the caption for keywords, match it to the medical subject headings database, and draw from PubMed the cases that are similar to yours. So now, if you upload a case, you get medical literature recommendations for what pieces of literature go well with it. I'm excited to refine that to the stage where, if you need to know the guidelines or a trial that showed the usefulness of a particular drug, you can just upload the case and we'll tell you what it is.

The step after that is to actually provide specific treatment recommendations. But there are already lots of apps that do that, like UpToDate or Pathway. What I think is powerful is being able to unite the knowledge required for understanding this particular case. I'd love to do some stuff using machine learning — we actually started a machine learning project a bunch of years ago that isn't finished yet, but I've still got hopes. You take a picture of a cardiogram, and the AI in our platform could give you the diagnosis, or at least a reading consistent with what somebody of your skill level would have.

Everything's right there except for the "lots of money" part. There are businesses that do this — you just have to look as far as TigerText, and a bunch of others, but TigerText was the one I thought about most. We shipped a direct-messaging piece in our platform. It was there for about three years, and it was one of the least-used features we ever shipped. It took a year of development, and maybe a thousand people ever used it to share cases directly with each other. We didn't go to big health systems to try to sell it, only because we had advice — essentially from our competitors — that if they had the opportunity, they'd have picked a different business model. Because selling subscription software to healthcare systems is both slow and not that lucrative.

Maybe in the UK the NHS is just throwing money at software companies — that doesn't sound right, by the way, I don't think that's what's happening — but there just didn't seem to be an easy way to enter that market. We certainly weren't getting spontaneous traction for that behavior. And just like going around collecting each individual doctor, now we'd have to go around and collect each individual hospital system.

Whereas with our business model now, where we're essentially selling physician engagement, the physicians show up spontaneously. The audience we need is already there. If we had to go out and get physicians to fill the app, and then go out again to collect deals from the market to make money, you're essentially creating a whole separate company on top of the one you've already created — a whole new workflow and department whose job is to collect business from people who maybe aren't that interested in giving it to you. It just didn't seem like the right direction.

That's basically why we started Figure 1 — because people were doing it by email. But it didn't replace that behavior. And I suspect it's because Figure 1 is not easier to use than WhatsApp for sending stuff.

Yeah — it's the knowledge I've gained, either from experience or from having the opportunity to receive the experience of others. One of the things I coach young entrepreneurs on is to ask where the demand for their product is. This comes from one of my favorite investors, Bill Gurley, the general partner at Benchmark, one of the most successful Silicon Valley firms. He's got a great blog he's written for years. He wrote a piece about how the difficult part of business is accruing demand, not supply. A lot of people reverse that — they get excited about an opportunity where they can create a lot of supply and it'll be easy to serve the demand. But that's the hard part: getting the demand. Making sure you know where the demand is in your market really changes how you build and model your business.

I've also been very fond of the methods people use to build minimum viable products. Whenever we've tried to test, I've recommended we use the most minimal version of those tests to see whether we should build anything further. We did not do that with direct messaging — we built the whole thing first and then found out nobody wanted it. But there was a one- or two-month period where we wondered if Figure 1 was going to be a telemedicine platform. So we built a web page for a telemedicine platform, spent a few hundred dollars on Facebook ads to get it in front of people, and it was a simple onboarding process that even asked people for their credit cards.

At the end of the process, it informed them the service — which was not built — didn't have any availability in their area, and that we'd let them know when it did. The purpose was to see how many people actually clicked the button and gave us a credit card number, indicating they wanted to use the service. We didn't keep the credit card numbers — they went to a secure server, and we deleted the account and expunged the numbers. But the point is we tested it and found out no, not a lot of people want this product. So we didn't do much more than build a superficial website. You can fake it to see if people want the thing you're going to make — and then you'll know if you should make it. Most of the advice I give people is things like that, because they restructure how you build your business and can save you a lot of time and money.

I'm a very impatient person, and that's definitely influenced my choice of career — being an intensive care doctor, where you can jump in and make a huge difference as quickly as possible. The same is true of me being an entrepreneur. I really wanted to get the product out there, build it, get it in front of people, test it, and push things forward quickly. Having that sense of urgency associated with the work I do in hospital works well in an entrepreneurship environment, because as a business person I'm allowed to shoot first and ask questions later — which is not something you do as much in medicine.

I once heard our product was going to be featured in a talk at a medical education conference in the same city as us. So I grabbed a piece of Figure 1 swag, threw on the t-shirt, ran from the office, and ran across the city to find that talk. I stood at the back with my t-shirt on, and when it was time for questions, the presenter saw me and asked me to come up and answer questions with them right there in the room.

So that impulsivity has been beneficial on the entrepreneurship side of my life, where it might be deleterious in my other professional life.

Yeah. You have to do unscalable things first, because then you know which pieces are worth it. You can scale anything if you build the correct infrastructure — the hard part is that it takes a long time to build infrastructure. So if you can do it manually first and show you get the output you need, then you can build the infrastructure and scale, provided the unit economics are sufficient.

That's how I'd advise people to build any business. Do one version of your business first, and do it manually — the hard way — and see if you can get a cycle all the way through. A friend of mine called wanting to know how he could raise money to make a carbon-fiber bicycle business, because carbon-fiber bikes are expensive and becoming more popular, and he thinks he can sell a lot of them at his bicycle store. I said, "Sure, tell me about your business." He did, and I said, "How many bikes have you built?" He said, "Well, we were going to raise the money and then build a factory." I said, "Okay — but build one bike first. Prove you can even build a bike, and get somebody to buy the bike you built." You're not going to have a factory and suddenly a thousand bikes. Have somebody make the first few by hand. If you can sell those, then maybe you sell ten next year, and fifty the year after. You're not going to sell a thousand next year — you're going to sell one. So make one bike first.

My recommendation is not to listen to doctors about those things, but to listen to builders. Being a doctor and being an entrepreneur or operator is great, but the reason you're good at it isn't because you're a doctor — it's because you're a person who found it sufficiently interesting to gain enough skill to be good at it. So I'd advise you to stay away from things specifically aimed at doctors if you want to learn how to build things. Because if you're building it, no one cares that you're a doctor. You're just doing two jobs and saving yourself a hire for somebody who'd give you medical ideas. That's why not a lot of medical companies end up hiring a doctor unless the doctor is the one who started the company — they're kind of expensive. You need someone working full-time on the product. So it helps if you're lucky enough to have both jobs. But I don't think there are any resources that made me feel like, because I'm a doctor and an entrepreneur, these two things go hand in hand in one special resource guide.

I think it's just being good at both things. I didn't mean to offend your podcast, by the way — which I guess is for physicians who want to build things. But I guess this is more of a place to get resources than a resource itself.

There definitely are. I have to unpack it, because some of the things are that I'm bad at some things I should be good at as a doctor, but I'm good at them because of my entrepreneurship. The easiest answer is that there are transferable skills. Being able to analyze somebody's presentation and understand whether they correctly represented the data in their findings is something I do without thinking about it.

That comes naturally because I've sat in thousands of rooms with PowerPoint presentations where someone asks a basic question about where the data on a slide came from, and the presenter easily answers. When you do that on a sales call and they have no idea where the data came from — now you recognize who you're in the room with. Being able to scratch the surface where you think the answers are is something every doctor learns to do.

There's a famous expression about an old bank robber — I have to remember his name, because they named a law after him. In the famous interview, they asked, "Why do you rob banks?" And the answer was, "Well, that's where the money is." Being able to know where the money is, is something you pick up as a physician, because you interview people every day, you examine people every day, you read notes that folks have written to help you understand. And when those notes or people or discussions are incorrect, you find out, and it means something really important — you've discovered something important about that person's health. When you have those discussions in a business context, people are much looser with the truth. Being able to recognize when people are being looser with the truth helps you be more critical about your own decisions. The robustness of your decision-making, through a science- and healthcare-based education, gives you a much stronger sense of where you can trust the information you get. So that's one of probably many things I've learned to appreciate about being in both business and healthcare.