Meet the CEO Behind Noom ($3.7B Health App)

Great. Geoff, I'm excited to have this discussion together. I've been a big follower of Noom over the years and really keen to have the discussion. The first question I wanted to ask you is, what's your pre-history to Noom? What were you doing before? I understand that you had a really big outcome as well. Can you give me an idea of that pre-history and what led you to Noom?

Sure. So I founded a company back in 2005 that would go on to be called The Meet Group. I ran that for 18 years, took it public on Nasdaq, and then ultimately sold it to a syndicate led by ProSieben, a large German media conglomerate, and General Atlantic, a private equity fund. That was a long and rewarding journey. We grew to a few hundred million in revenue and a few hundred employees, and it was a $500 million exit at the end of the day. A great journey really all the way through. Our primary product towards the end was a live streaming video platform. We were doing live streaming video for a number of brands, including three brands owned by The Match Group, the large dating conglomerate that has Tinder, OkCupid and Hinge as some of its brands. So I was leaving because I had sold the company. I worked there for a couple years on the transition. As I was leaving, I was meeting people in my network, including an ex-CEO of The Match Group, whose live streaming video we did for a long time. I happened to be having breakfast with her in Dallas, Texas, and she asked me what I was up to. I mentioned I was working on a mindfulness application, and she said, "Mindfulness? You should be the CEO of Noom." I wasn't exactly thinking down the road of anything other than starting another company, which is all I've done. I'd never really worked at a company that I didn't start until Noom. But she connected me to the CEO search. She was connected to another board member who was close to it. And four to six weeks later, I was CEO of Noom.

Before we go into that, you mentioned an acquisition. Was the acquisition organic? Did you pursue it? What can founders learn from you about that process?

The way I approach M&A is that M&A is generally the result of multi-year relationships, and in this case, it's probably the result of a decades-long relationship. So you generally cultivate these sorts of relationships, which happen in the form of giving updates on how things are going, maybe once a year or more frequently than that, maybe when there's a big product launch. M&A generally only happens when both parties think the other is part of their long-term solution, and not just think it. A lot of times it only happens when they existentially believe it. I find generally M&A doesn't happen if there's any reason for it not to happen, and the reason it happens is because someone just absolutely believes it must happen. So you cultivate relationships with the people and the companies that are the natural buyers. You have to think through the buyer landscape. But it can take many years for the right sequence of events to happen. And this is my approach really on both sides, whether I'm the buyer or the seller. I've been the buyer at least five times and the seller at least three times. It's the same sort of motion, where you're not generally going to buy something you've never heard of, or just because there's some process and some banker drops a slide deck on you. Those are the ones you ignore.

And then at the point you joined Noom after that conversation, what do you see in front of you? What happens at Noom at that point?

Yeah. I was previously in social for a long time, a couple of decades. I liked the opportunity at Noom because I felt my capabilities were really in building product and thinking through strategy. In social I was doing that to improve human connection, but, a little jaded from a couple of decades in social, I liked that Noom had a large brand and took a whole-person approach to health and wellbeing. It was mostly focused on nutrition, but the brand definitely stood for more than that. I thought I could lean into the whole-person aspects and take what I'd learned about gamification and consumer app building and apply it to a footprint where, if you're using the app more, you're actually getting healthier. You're not just spending more time and generating additional ad views and maybe in-app purchase revenue. That concept appealed to me. I was looking to do something meaningful, and that's what I liked about Noom. On entering, the first thing I did was tweak the mission statement to "empower everyone everywhere to live better, longer every day." So, really leveling up beyond weight loss. Weight is certainly part of healthy aging and part of living better, longer. But saying "live better, longer every day" contrasts the abstraction of health span extension, which some players think about, with the vitality and tangibleness of feeling better today. That was one of the first actions, probably in the first month. The first month is largely listening and learning. Over time I brought in most of the team: two-thirds of the team are people I've hired since I joined. The biggest initiative in our first 12 months was building Noom into a great GLP-1 companion service. Noom was a behavior change program for weight loss, and it had good bones and infrastructure. But to do behavior change around GLP-1s, you need a different endpoint than caloric restriction, because that's not the problem on a GLP-1. I put myself in the GLP-1 user's shoes: I started microdosing a GLP-1, and you definitely feel the food noise go away. It really becomes about getting enough protein and doing the resistance training. So we created a new GLP-1 companion mode of our service, and that ended up being popular and engaging. People were actually doing it alongside the medication, which was great. Shortly after that, we brought the medication into our program and started selling behavior change plus medication combined, and that proved very successful. Those programs are now 70% of our revenue. So it's been a lot of rapid growth over the last few years.

And Geoff, on that: when you launched the GLP products, it was very early in the market. Everybody discusses GLP-1s today, maybe three, four years later. What were the hints you could see, and what kind of evidence did you have in mind to pursue that avenue?

Yeah. They were natural extensions of what Noom does, because Noom was focused on weight loss and healthy living generally, but starting with weight loss. GLP-1s are just very powerful weight loss medications that work spectacularly well. But some of their main drawbacks, like lean mass loss, are really a function of not getting enough protein and not doing enough resistance training. So behavior can really affect your GLP-1 experience. We also believed at that time, more intuitively, though over time we're getting more and more data, that because Noom was and is effectively a habit program, a program for establishing healthy habits through behavior change, GLP-1s essentially open a habit window. The way they do that is they reduce appetite, of course. They make it easier to not eat the unhealthy thing. But they seem to even do it selectively. There was a study on Tirzepatide published recently, and cravings for all food types reduced on Tirzepatide, a GLP-1, except for fruits and vegetables. So it may just become easier to make the healthy choice on a GLP-1. The thought is, when you come in on a GLP-1, you have this fresh start effect. You're ready to begin this new change program. You're embracing this idea of yourself in the future, and you have a lot of intent right at that moment that you sign up. In the Noom program, we're generally trying to address both system one and system two, so fast and slow thinking. We want to help you, while you're intentional, to architect your choice landscape to make it healthier: maybe remove the unhealthy snacks, maybe think a lot about what your shopping choices are the next time you go to the grocery store, because those will set you up in the future. But at the same time, because we're also fast-thinking creatures that don't necessarily plan perfectly logically, and we suffer from various biases like present bias and so forth, we built a bunch of gamification features like streaks and virtual currency for hitting step goals and logging healthy meals, and we brought that into the program over the last few years as well. So this is essentially how we think about it: the GLP-1 opens this habit window, and we help you make these habits. If we're doing our jobs well, that matters when you stop taking the GLP-1 medication, and most people stop eventually. Even if they later come back on, the data suggests the vast majority have stopped, something along the lines of 90% or so by year three, many by year one. And when you stop, you regain all the weight in 18 months, based on an Oxford study that was published recently. So if you want to avert that fast speed of weight regain, you have to have changed something besides just eating a little bit less during that period of GLP-1 pharmacology. That's what Noom is really all about: helping you build those healthy habits that can help you not only lose the weight but keep it off as well.

I'd like to start from you joining the company and the ordering of products over time, because we speak about GLPs, but I've also seen supplements, and there are pharma products and so on. It's very interesting to understand how you made the transition internally to offering all of these services. So I wanted to hear your view on that, and then maybe, how does everything connect together? Does the user journey start from the person wanting to lose weight, and then you provide supplementation later and so on? Is this the longer view of the user, or are these things disconnected? Because obviously discovery and connecting products internally is a very difficult process, especially in a product.

Yeah. I think the way I think about Noom's growth and where it has the right to play, the main lane that Noom has been pursuing has really been the cardiometabolic lane, right? So weight loss, pre-diabetes, diabetes: that is where Noom has real strength. We have a CDC-approved DPP, diabetes prevention program. We have a diabetes program that we've built in the last couple years. And of course, we're known for our weight loss solutions. So we think we have a natural right to extend those conditions down the cardiometabolic route. So you can imagine sleep apnea, cardiovascular disease, right? We want to be your companion not just for weight loss, which of course we have a great companion for, but for healthy aging, for heart health, and for cardiometabolic conditions generally. We also see weight loss straddling not just a cardiometabolic condition but also a condition of healthy aging, right? So it touches preventive health more generally. And when you start thinking preventive health, you're generally also thinking about energy, muscle composition, maybe sexual health, sleep, right? So we feel like those are natural areas for us to extend into. What every condition we treat really has in common is that they tend to be non-acute, high salience chronic conditions. That's kind of where we play. We don't really offer treatments for acute conditions, and we generally don't offer treatments for low salience conditions, or one-and-done sort of conditions. They tend to be chronic. So that helps us define the universe of what we offer. And I would say we're generally pretty limited in our formulary compared to a lot of players. You mentioned supplements. We have a metabolic daily supplement, and we'll have a few others launching soon, but we don't think of ourselves as predominantly a supplement maker that's going to have 50 or 60 different SKUs. It's a focused SKU. We think in terms of both add-ons and entry points. Something could be both, but some products are more natural add-ons to a GLP-1, and others might be more natural entry points. So we try to figure out which is which as we think through what to build and what our go-to-market might look like for that product.

Granted, Geoff, from the time you joined the company, you have had all of this increase in AI. I've had a number of discussions about it now. Let me preface this: I spoke to Vinod Khosla, and his statement about doctors is that AI is so much better and doctors are over. I also spoke with the CTO of Hims, and he brought up the idea that AI is empowering physicians, which is exactly the opposite statement. Where do you stand on this? What's your own view about physicians and the future, considering AI?

Sure. The primary thesis that we have is that there are a number of trends that are all converging and essentially being mediated through this layer of AI. What we basically see in healthcare right now is there are companies that are really focused on behavior change, and I would add wearables into that. So now I'm talking about Oura, Whoop, certainly Noom, Fitbit, Apple Health are aspects of that. I'm wearing an Ultrahuman at the moment, testing that out. Then you have convenient care, clinical care, and of course Hims and Ro have really pioneered this open loop, but you'd put Amazon One Medical in there too. Then in terms of the dimension of diagnostics, of course you have Function, Quest, LabCorp, and so forth, Neko Health. What's interesting about each of these is that they're not really separate domains. What's actually happening for all the players that I just named is that they started somewhere. They started really aligned to one of these axes, and we generally think in three axes. They started on one of those axes, but now they're converging into this space. Oura and Whoop add diagnostics. Noom added diagnostics with an at-home Tasso test. Ro and Hims added diagnostics. So you're seeing diagnostics become standard, but then you see Noom add clinical care as well, years ago at this point. I wouldn't be overly surprised to see wearable makers get into prescribing. Part of the reason for that is I think essentially all these players are chasing the same holy grail. Basically, the holy grail is the ability to say, "Here's what's changing in your health. Here's what's causing it. Here's what's going to happen next if you don't do something. And here's what you should do about it right now." And what you should do about it right now might be take this medication, change this behavior, take this diagnostic test. That thing could vary, but that's the essence of it. I think that's what preventive health looks like, and I do tend to think it's wearable-enabled. I think a lot of the players that I named are thinking in terms of building data assets, this closed loop where you have maybe prescribing, you have some kind of intervention like coaching and support. You're tracking via wearable, maybe the wearable that comes with the program, maybe the wearable that they happen to be wearing. You're measuring all of this data. You're titrating doses based on it. And in your coaching paradigm, which Noom has, both AI and human, you're surfacing insights that no human could possibly know, because you're not going to have a full understanding of their blood test and how that connects to their food logs any other way than through AI mediation. So the end result is you want to see this system feeding all the data into it in various ways, multifaceted data, mediated by this layer of AI, which is obviously improving rapidly, in order to produce some health insight that's actionable and that, if acted upon by the member, improves an eventual health outcome. If you can do that, you've built the ideal preventive health platform. I certainly don't think I'm unique in believing this. I think that's what falls out naturally as you start thinking about what a strong preventive health platform looks like.

Geoff, that is very interesting. It sounds like a very vertically integrated healthcare company that keeps growing. What makes me wonder is, what are the LLM companies doing? It does sound like they could be after something very similar.

Yeah. I think the frontier labs will, of course, want the data coming into the system. But unless they're able to prescribe the medication, track the medication taking, and pull the wearable signal, it's difficult to replicate what is in a closed loop data ecosystem, by definition, because it's a closed loop. We do use the LLMs in order to mediate the signal and make something surfaceable and intelligent to the user. So I think they have a role to play in it, but I don't necessarily think they capture the fullness of that opportunity, in part because there are just so many atoms moving around. I'm not talking about bytes. I'm talking about medication shipping to someone on time, a doctor prescribing the right medication, being available via chat for adverse event back and forth. This is not really what you think of when you think of an LLM, but that is what companies in the healthcare space are dealing with all the time.

Well, from a product perspective then, first of all, what was the initial introduction you had in Noom, and then where is this going? There is always this longitudinal idea that allows you to know the customer better and have more signals coming from, as you mentioned, labs and devices and so on. Obviously, that needs a specific kind of file and understanding for the user, so there are some learnings there about how do I store that information, how do I process that information, and so on. But maybe give me an idea of the product evolution considering AI in Noom, and then where is this really going?

Yeah. So we use AI in Noom in a lot of different ways. One of the first things we did a couple years ago was our "no empty boxes" policy. Essentially what that means is we have many coaches, and they should never be staring at an empty box when deciding what to message to a member. There should be something suggested, and that suggestion should really be taking into account the breadth and depth of the data that we have. But that coach is ultimately responsible for the content of the message they send. And so you can obviously improve your AI based on seeing what the coaches are doing with that message. That's one example. Another is just making it dead simple to food log by just taking a picture, interpreting the contents and all the calories, macronutrients and so forth. That was another important change. We've launched an AI chatbot that's a separate chat thread from your human coach and is obviously available twenty-four/seven to go back and forth with you in real time. So that's a small piece of what it looks like. What we're excited about is that there are new models of care coming, and one of them is being pioneered by CMS, essentially Medicare. What's interesting about the CMS ACCESS model is they're essentially paying outcomes-aligned payments. So you only get paid as the care provider, let's say Noom, if you improve hard biomarkers like blood pressure, A1C, or weight between baseline and a year after baseline. In order to prove that, it would help if you had a wearable that was FDA registered or FDA approved for, let's say, blood pressure. I think you'll eventually get to that with some of these devices, but right now we're shipping a cuff, a blood pressure cuff. What I like about that model is that the program also has a payment that goes to the clinician who is managing that patient's care. So imagine the primary care physician, instead of just saying, "You need to move more and eat better," instead says, "You should move more and eat better, and here's a bunch of choices you have to help you do that," because that's not easy to do. "Here's a Noom behavior program that comes bundled with this, that, and the other thing. I'll be able to check in on your progress with that program, and we'll talk about it the next time you come in." That's a much better patient experience than "You should eat better and move more, and by the way, maybe you should eat a yogurt in the morning." That latter feedback is forgotten the moment they step out. But if you could put something like a wearable on their wrist or on their finger or wherever, and an application that was gamified and built to encourage healthy behavior, you at least have more of a fighting chance of changing behavior. The way AI of course comes into that is the next time that clinician talks to that patient, there's going to be an AI recommendation provided in the EMR to say something about that patient's choices. You might say, "Great job. I see that you're earning your Noom coin and you're hitting your step goals, and you might want to think about more lean proteins." Something like that. I think a number of cutting-edge companies are building for the CMS ACCESS model, and they're doing it because that's going to build a bunch of data over the next year or two. Other payers are going to look at that data and say, "You know what? This one didn't make sense. This one did," in terms of whether it was worth the cost, which is generally relatively small for these payments. That's the future we see developing and that we see CMS pioneering.

Geoff, let's say we have a blank canvas here, and we design the full user journey considering AI. How many things are going to change, and how many are going to stay the same?

Yeah. In the fullness of time, it's a very hard question to answer because no thoughtful person can talk intelligently about AI a couple years out. It's moving too quickly. You could imagine a care model that's entirely AI-driven, but you would need a fair amount of regulatory change to enable that, and maybe you'll get it or maybe you won't. As we think about evolving, what we're really thinking about is how do you take all this data, because there are mountains of data, and make an actionable insight with it? For the clinician, if you're talking about an EMR integration, the last thing they want is to see a bunch of data. You get a few lines, right? And those lines had better matter, because that clinical visit is probably five to 10 minutes long. So you need something that allows for one minute of it to take place in a way that improves a potential outcome down the road. A lot of it is digesting into something that is actionable and easy to work with. I think that's a challenge for a lot of players, because as you get deeper into AI, a lot of times you also get this complexity that tends to snowball and is the opposite of making things easier. So we're trying to focus on that. You could obviously imagine even greater change as you think through robotics and different care models that could be AI only. But we generally assume our job is to say, well, there aren't enough physicians in this country to treat everybody unless care models evolve. Asynchronous care is a model patients actually want because they can chat with their clinician in real time, and that clinician is able to scale their care across more patients than they otherwise would. How do we get maximal benefit from AI for that while keeping the clinician in control of the patient's health and ultimately responsible for the contents of their communications? That's generally how we think about it today.

Maybe a couple of questions on peptides, Geoff. What kind of peptide are you really excited about today, or maybe convinced by, beyond the GLP-1s?

Yeah. Today on Noom we have hormones in our formulary: we offer HRT. We have the peptide sermorelin, which is relevant for muscle composition for some people on a GLP-1. We have the coenzyme NAD+. We have a few other non-GLP-1 anti-obesity or related metabolic medications, like metformin. But right now I'd say our formulary is fairly focused. I do think you'll see continued expansion of it. It wasn't long ago that none of those things were available on Noom, and we've taken a slow and steady approach to formulary expansion. We're paying attention to what's going on in Washington on this front, and of course we're familiar with the PCAC hearings that took place. Ultimately, it's our chief medical officer and the medical directors, working with me and the product teams, who decide when something should enter our formulary. For now, what I just described is what we offer. So in terms of peptides, sermorelin is the only non-GLP-1 peptide currently in our formulary. There are only so many non-GLP-1 peptides compoundable in the United States today. You also have oxytocin and maybe PT-141, but it's really a small list of peptides that are compoundable in the United States today.

What's your opinion about the market for peptides over a five-year timeframe?

It depends on whether you see a loosening of the regulatory approach, with some additional peptides added to the Category 1 list. If that happens, I think you'll see a lot more clinical studies and a lot more evidence develop quickly. Right now you can't run those studies, because you'd have to go through the new drug application process to run that trial, and you'd be talking about tens of millions in investment to make that happen. So I think you'd see studies take place. I'd imagine that if there were a loosening, the regulator would need some kind of surveillance system in place, so you'd get a lot of signal. Right now there's a limit on how much data is out there beyond mechanistic study data. In a lot of ways, what I just said is also true of the supplement business, and has been for decades. To some extent, I look at some of the newer peptides as asking what prescription-grade wellness looks like, or what a supplement in injection form looks like. There, what's probably all-important is the safety signal, and that it's very clean. But it's a situation with a lot of nuance that you want to move slowly in. The players that go in quickest are generally the ones with the least oversight and responsibility. At the same time, you have to balance the fact that patients are seeking many of these interventions. If the stance is to force them into the gray market, or what I'd call the black market, because I don't think it's particularly gray, people are getting compounds from China in the mail and reconstituting some kind of powder in their living room. That's an extremely dangerous situation, and apparently a multi-billion dollar market today, based on what I've seen. That has its own risks. So if you're able to establish a safety signal, moving that into a regulated market could offer some patient benefits. But that's the nature of the trade-off the regulator has to evaluate.

I can't help making the correlation with creatine 20 years ago. Back then, getting creatine was very difficult: you'd need to order it from somewhere and it would arrive at your house, and the news treated it like a steroid, which back in the day was a really bad word. So I'm wondering how much the government is a lagging indicator in these things, and also what role companies play in helping the government understand all of this.

Maybe one of the questions I wanted to ask you, Geoff, if not the last question, let me just ask a double question on this one. Granted that AI changed so many things, what kind of shift did you make internally in the company so you can be more AI native, so people understand AI better, and so on? Was it more of a training of the existing people? Did you bring in a separate function? And then the last question I would add to that is, say we speak three years from today again, what does Noom look like?

Yeah. On the AI front, I've brought in a strong CTO who's attracted a number of AI architects around him, and has built tools so that things that used to take a long time, like creating new intake flows, are something you can do without code, right? And then democratize the building of that beyond just a few people in a particular group to maybe the whole marketing team. So that's the nature of the changes that need to happen. As I look forward into what happens as that continues to evolve, it really comes down to how much signal there is in some of these wearable devices, right? Are you going to be able to get a blood pressure signal off of the device without any kind of intrusive cuff or something? Is there a low limit for what useful data can be gleaned from the device? Or is it actually a high limit, where you get all sorts of stuff off that device with just different colors of light and different measures that are all non-invasive? And I think if you believe it's the latter case, then you're about to enter this golden age of health, because you're going to be able to interpret signals off the body that feel like magic. But if it turns out there's not much you can do through the skin beyond heart rate and HRV, then of course that's a lot less interesting. I don't think we know which future we're in right now. But if I had to bet, I'm an optimist: the sensors are getting better, we get better at knowing what to look for, and AI gets better at interpreting the signal. So I tend to look at the wearable companies as really well-positioned in this space right now, because they're the closest to the patient, or the user.

Super. Geoff, it was a great discussion. Thank you so much.

Yeah. Thank you.