🎧🍌 PhD to $100M in 4 Years: Rebuilding SMB Lending with Sahill Poddar, Co-founder and CEO of Parafin
Discovering the Higgs boson particle at CERN, why access to credit is broken for SMBs, and the growth trick that Robinhood got most of its userbase
Sahill Poddar and the team at Parafin has quietly built to nearly a $100m revenue run rate in four years. And they did it in an industry that's become a Silicon Valley graveyard: SMB lending.
This latest podcast episode with Sahill goes inside how they partnered with other marketplaces, vertical SaaS, and point of sale providers to offer financial services to SMBs at scale, landing DoorDash as their first customer before building the product, and advice for technical teams learning enterprise sales.
Sahill's a fascinating founder, starting his career getting a PhD discovering the Higgs boson particle at CERN. We talk about physics, he explains how the Large Hadron Collider works, why physics is just real world machine learning, and all the lessons he learned on the growth teams at Facebook and Robinhood (including the way Robinhood acquired most of its userbase!)
Thanks to Ramp for supporting this episode, and to Hans Tung at Notable Capital, Nick Shalek at Ribbit Capital, and Mahdi Raza at Pathlight Ventures for their help brainstorming topics for the conversation.
Presented by Ramp
If you're running a finance team, you know how much time gets wasted on expense management. Chasing receipts, categorizing transactions, waiting for expense reports. It adds up quickly.
Ramp handles all of this automatically. Its a corporate card and expense management platform that over 40,000 companies, like Shopify, CBRE and Stripe use to streamline their financial operations.
But here's what makes their corporate card different: every transaction gets automatically categorized and matched to receipts. No more wondering what that $47 charge was for three weeks later. You can set spending controls, get real-time alerts, and even block certain merchant categories. It's like having a finance team member embedded in every purchase.
The platform integrates with your accounting system and ERP, so everything flows through without manual data entry. Whether you're issuing cards to a few employees or managing spend across departments, Ramp gives you visibility and control without the paperwork.
Stop chasing receipts: check out ramp.com/ThePeel to get $250 and see what a corporate card can actually do for you.
Time is money. Save both with Ramp.
To inquire about sponsoring future episodes, click here.
Timestamps to jump in:
4:06 Lending to SMBs inside marketplaces and platforms
9:39 Why SMB lending is so hard
12:50 Three ways AI is changing Fintech
16:47 Silicon Valley’s graveyard of SMB lenders
22:44 Getting a PhD in Particle Physics
26:15 How CERN's Large Hadron Collider works
31:49 Discovering new dimensions
34:10 Building billion user data sets at Facebook
39:53 Working with other physicists at Robinhood
50:29 Growth lessons from FB + Robinhood
1:00:57 Starting Parafin, embedded SMB lending
1:06:09 Why credit is the biggest problem for SMBs
1:10:53 Raising a Seed from Ribbit
1:13:25 Landing DoorDash as the first customer
1:16:51 Mastering B2B sales as a technical founder
1:22:58 Lessons from Vlad at Robinhood
Referenced:
Careers at Parafin
Prior episode with Charley and Mahdi
Check out Julius
Find Sahill on X / Twitter and LinkedIn
👉 Stream on YouTube, Spotify, and Apple
Transcript
Find transcripts of all prior episodes here.
Turner Novak:
Sahill, how's it going along with the show?
Sahill Poddar:
Yeah, thanks for having me, Turner.
Turner Novak:
I'm excited to do this. I think our mutual friend, Mahdi, prior guest of the show, episode with Charley and Mahdi from Pathlight, Mahdi introduced us. He said, "You got to talk, you got to have him on." So, I'm excited for this. And I know you haven't done very many podcasts. So, for people who are not familiar with Parafin, the company that you built, can you give us a real quick rundown on what you are, what you're building, state of the business, and then we can go from there?
Sahill Poddar:
Yeah, absolutely. Parafin's mission is to grow small businesses. Our execution on that is fairly straightforward. Small businesses flock to software platforms every day to start, run and grow their business. These platforms process payments, generate demand, solve a lot of operational challenges for small businesses. As a result they sit very low in the mass loss pyramid of needs for small businesses. They have an established trust, and trust is really the bedrock of any financial transaction. And so, we believe that these platforms have the right to serve more financial services to small businesses, and Parafin's goal is to accelerate that eventuality by enabling platforms to do so in the most simple manner as possible. The biggest need as it turns out for small businesses is getting easy access to capital.
Turner Novak:
Really?
Sahill Poddar:
If you think about the different ways a small business, or for that matter any business can fund itself, it's either through earnings, through debt, or through equity. Equity is not really an option for small businesses, and payment solves for the earnings use case. Parafin comes in and solves for the debt use case. We essentially tie data that platforms have on their small businesses, the existing relations they have, and use that to start extending credit to small business sellers.
Turner Novak:
Maybe a dumb question, but we've had banks for thousands of years, even in the past. They get the modern age. There's a lot of banks, don't they have small businesses that they lend to?
Sahill Poddar:
They absolutely do. But one of the biggest problems in small business credit in the United States is that it really is a consumer loan or consumer lending products masquerading as small business lending products.
Turner Novak:
So, what does that mean?
Sahill Poddar:
What that means is that banks and credit card programs typically rely on personal FICO or personal credit scores or require personal collateral, even if the person trying to take the loan is trying to do so for their small business and not for personal use. And that basically deters women-owned businesses, minority-owned businesses, new forms of business lines. You can be a profitable seller on Etsy or Amazon and selling profitably for several years, but you may not pass the EBIDTA threshold for the bank or you may not have the right FICO number for the bank. Meanwhile, if you look at the business performance, there's many reasons to be able to extend credit to them. And so, we break that mold and solve that by relying on data and machine learning and AI modeling in order to extend credit. And so yeah, banks do play in the space, but they have archaic means and archaic methods to do so, and we're here to reinvent that.
Turner Novak:
And so, what, I guess before we get really deep into it, what's kind of current state of the business when someone's like, "How's it going? How's Parafin doing?"
Sahill Poddar:
Yeah, so Parafin is doing well. We funded somewhere between 30 to 35,000 businesses, small businesses. Are nearing in on 100 million dollars of gap revenue run rate, and we launched our product just over four years ago. And so, the state of businesses is pretty good. We're growing pretty rapidly and we're excited about growing our existing set of products, but also launching newer products.
Turner Novak:
I think you mentioned before, there is, in every neighborhood in America there's probably a Parafin customer, you phrased it something like that. Is that the case?
Sahill Poddar:
Yeah, so we have a map where we can see where we are actively making offers to small businesses. Today we make over 10 billion in offers to small businesses on a daily basis. And so, there's probably a Parafin offer available to a small business, no matter which part of the country you're in.
Turner Novak:
When I say make an offer, so this is like if I'm on my Etsy seller dashboard, if I'm in my DoorDash restaurant management thing, is it in the top of the screen it's like, "You qualify for a $50,000 equipment." How does it usually play out?
Sahill Poddar:
Yeah, that's directionally right, so for example, if you're a restaurant on DoorDash, if you were to log into your DoorDash Merchant Portal, Parafin has essentially looked at your data, underwritten you for an offer. And so, it's more than just a pre-qualified offer. We like to call it a pre-approved offer, because there's very few steps, essentially on the back of a few clicks you could be accepting that offer and getting paid as soon as the same business day.
Turner Novak:
And that's because you have an integration with a platform and you have all the data and you are just constantly as part of the UI based on the data that you're getting it just floats it to the merchants or to the businesses that are in it?
Sahill Poddar:
That's right. The way we typically work with platforms is we approach them, we start ingesting their merchant sales data, and that passes through our underwriting models, which has trained on over a million small businesses here in the U.S. And based on that we either decide, we decide whether a merchant is eligible for an offer, and if they are, what the size and price of that offer is, and start rendering that offer to them on the DoorDash Merchant Portal. A restaurant could log in today and see what their offer is, and if they like it, if they like the terms and if they have the need for that money to grow their business, they could accept the funds and get that in their bank account as soon as the same business day.
Turner Novak:
You've done other FinTech financial services in the past, we'll probably talk about that a little bit later, but why lending specifically? Because I think a lot of people might say, "Lending, there's pros and cons to a lending business." Why do you think lending was a good idea?
Sahill Poddar:
A few different reasons. First and foremost, I grew up in a small business family in India, so I've seen firsthand what embedded lending, which primarily happens through social networks there. You know someone who knows someone who gives you money for you to grow your business.
Turner Novak:
That's like offline social network?
Sahill Poddar:
Offline social networks, yeah. Offline social graphs, I think is probably the better way to put it.
Turner Novak:
You log into Facebook and your news feed it's like-
Sahill Poddar:
No, these are offline social graphs, and how access to capital for SMBs can both enable and the lack of access can hinder the growth of small businesses. In some ways I've lived that experience. And secondly, going back to my earlier point, businesses can earn money in a few different ways, or fund themselves in few different ways, either through earnings on sales or through getting access to debt. The debt markets historically have been relatively shut for small businesses, and there's a vast opportunity to enable that for them. And at the end of the day, if you want to be playing in a large market, you have to be solving real customer problems and real customer needs. And if you were to survey even the restaurants within a one-mile radius here, or for that matter most small businesses around the one-mile radius here, the vast majority of them would say the biggest blocker in them growing their business is getting access to financing.
Turner Novak:
Really? Because you think it's profitable that you should be able to borrow money. It's that the, generally speaking, we're making some vague assumptions here and maybe generalizing a bit, but it's just they're too small for it to be worth somebody who does a lot of manual work on a lending model, building a cashflow model to, can they pay it back and they have a sales team that needs to spend time. Is it just some of these businesses are just too small for it to really be worth it for a non-tech focused software-first platform or traditional lending company?
Sahill Poddar:
Yeah, that's exactly right. One other way to think about it is, what are the different forms of leverage that exists? There's capital code, labor and media or content, banks are in the business of using capital and labor in order to access their markets. We are in the business of using capital and code. And with code you can scale things much faster than you can with human labor. And so, getting the human out of the loop in an underwriting process and doing that and replacing them with code that is making smarter decisions and is able to ingest more data and have better sort of decisioning along any risk-reward curve is the way to access this market at scale. Because you can only get so far if you were to leverage labor versus code.
Turner Novak:
And speaking of leverage using technology, are you seeing any ways that AI is changing how FinTech companies operate? I don't know if you're doing anything specifically inside of Parafin, but ...
Sahill Poddar:
Yeah, absolutely. It is certainly a paradigm shift in many ways. At Parafin, my background is in machine learning. I was leading the machine learning team engineering at Robinhood, and I got into the world of machine learning for a PhD in particle physics. So, in some ways, data science, machine learning, AI models have been very close to me and the kind of work that I've done in the past. And the same is true for my co-founders at Parafin. We've been using machine learning and AI since pretty much day zero, or since the very early days in our underwriting models. And that's one big area of investment in AI and AI models is in order to keep improving our underwriting and get smarter around identifying bad borrowers, good borrowers, pricing, and removing customer friction in their flow.
The second category is improving our own operational teams and equipping them with AI tools that make them faster and more efficient. For example, the teams will often look at if the borrower, the business borrower has any related businesses that have filed for bankruptcy. Is this business undergoing any court cases? And almost prepare like a report card given a business owners, borrower, business. And before the advent of LLMs, we'd have to go and do that pretty manually. Now we've built a tool that just enter a business and you can get that information out pretty quickly. So, automating these type of processes helps us be more efficient. And so, we've in-housed a lot of that sort of intelligence around underwriting funding and collecting repayments from a business using machine learning and AI.
The second category is using vendors, and you obviously want to be smart around using fast-growing vendors or solving problems. So, we use a vendor for customer support, we use a vendor for coding, et cetera.
The third category is just on a personal productivity, but work-related personal productivity. If I'm working on a task, how could I be more productive and get this done much faster using the best of what LLMs can offer? And so, those are kind of the three surface areas. In-house products that make our core offering stronger, being very thoughtful and intentional about which vendors we use, and using it as often as possible in order to get efficiency on projects that I'm working on or other team members are working on.
Turner Novak:
What's been the most efficiency that you've gotten from something like that? Any specific tool or trick that you found?
Sahill Poddar:
Yeah, I would say o3, and in general the reasoning models have been really good in giving long-form analysis to run. For example, one analysis that we run very often is IRR computation on a loan debt, given a set of borrowers and given repayments, compute IRRs and portfolio performance. And yes, surely we have our own code that kind of goes through that, but o3 is able to kind of back out into similar numbers pretty fast. And so, using the mathematical intelligence of something like o3 to run a deep analysis on data has been quite fruitful. And I don't have to take time away from someone internally on what they're working on, and I can just do that myself.
Turner Novak:
Sure. Have you tried Julius? Julius AI?
Sahill Poddar:
I have heard of it. Of course, the founder was a celebrity for a day-
Turner Novak:
He still is a celebrity.
Sahill Poddar:
He still is a celebrity, and I haven't used the tool myself.
Turner Novak:
Oh man, you got to try it. He's also a prior guest of the show, so you got to support The Peel ecosystem, all the prior guests.
Sahill Poddar:
Awesome. I'll try to give it a try.
Turner Novak:
I think it's a really interesting thesis around building this, but I feel like there's been a lot of people that have sort of tried this in the past. Is that right? Or some approach of like SMB lending using software?
Sahill Poddar:
Yeah, yeah. So yes, that is right. Silicon Valley is a graveyard of SMB lenders.
Turner Novak:
That was a way better way to phrase what I just said.
Sahill Poddar:
Silicon Valley is a graveyard of SMB lenders. There's no hiding from that fact. I think what each of the folks prior to us have gotten wrong is that they've gone direct to the small business and tried to lend to them. Why is that a mistake? That is a mistake for a few different reasons. First and foremost, when you do that you are fundamentally not differentiating yourself on product, but rather, who can directly reach a small business as fast as possible or make the most attractive sort of acquisition pitch to them as possible?
That is fundamentally a marketing business, not a product-led business. And that's a business that in the long term the ad networks like Google and Facebook win, because the only way to get access to small businesses directly at scale is to run source of some sort of performance advertising. But guess what? Businesses that click on ads to borrow money are probably not the ones you want to lend money to anyways.
Turner Novak:
Like, "Help, I need a loan."
Sahill Poddar:
So, you're kind of fighting on the edges of adverse selection and you're constantly trying to stay away from adverse selection while being very close to it because of this direct to SMB approach. Secondly, there is naturally a CAC associated with this, which eats into the unit economics, and CAC is not predictable. You can go and acquire 100 users on Facebook at 20 bucks and be like, "Great, my CAC is 20 bucks." And guess what happens when you try to do that with 10,000 users? Your CAC probably blows up to 200 or 2,000, and you have a fundamentally very different unit economics to what you had started out with. And so, these are the sort of problems that play direct to SMB lenders. There's no differentiation on product. You attract adverse selection, and you're fighting CAC for unit economics-
Turner Novak:
Did you specifically say, "These are the problems, how do we solve for this?"
Sahill Poddar:
That's right. We focused on each one of these, and let's look at them one at a time. First and foremost, by not going direct, but relying on platforms where SMBs are earning their paycheck, a restaurant earns their paycheck on DoorDash, an E-commerce seller earns their paycheck on Amazon, they sit very low in the master pyramid of needs.
This established trust is much harder for a small business to walk away, for a restaurant to walk away from DoorDash than from Sahill Capital or Turner Capital. And so, by relying on those relationships you can essentially reach many small businesses once you have integrations with one of these large platforms.
Secondly, by using data, highly proprietary siloed data that's sitting in these platforms, you can start to underwrite and cherry-pick which businesses you want to give loans to, versus which ones are going to click on your ad and then get rejected in your flow.
And then thirdly, because you know what the size of the opportunity is, you can price deals with these platforms in a way that your CAC is highly predictable and scales from one to 100,000 customers and does not change along the way. So, it's highly predictable, highly scalable cost of getting a business. In fact, there's no CAC. It's kind of-
Turner Novak:
Like the CAC is acquiring the platform, and that is-
Sahill Poddar:
That's right.
Turner Novak:
... much more affordable, much more and more repetitive and sustainable.
Sahill Poddar:
And there's essentially zero incremental CAC for getting every new small business to use the product. And that tends to be powerful. And going back to my second point, that's a very powerful point in using the data pre-approving offers, because the best businesses who you want to lend to are not actually actively looking for money. However, when they log into their merchant portals of their dashboards they're using to run their business enough times and they see a pre-approved offer, there is a non-zero probability that they will accept it. Simply because you started to plant an idea that, "Hey, here's $50,000, it's a few clicks away." And then in two weeks later that subconscious mind has spun up ideas how to use that $50,000, and the business comes back and accepts the offer. So, that dynamic is an impossible to crack when you're going direct to small businesses, which is why many have ... Any provider who's tried this in the past has not been successful, and why Parafin is different.
Turner Novak:
It's like the same analogy with VC and startups. The best startups probably don't need the VC's money. If you're constantly only relying on the founders that are coming inbound pitching you, it's like, well, maybe there's a little bit of adverse selection there versus the guy who's just building, crushing it, doesn't need your money. And then you have in the venture side, it's like everyone trying to talk to that guy or that girl to give them money. It's like maybe similar dynamic, different profile of business, but it's some parallels.
Sahill Poddar:
Yeah, no, absolutely. And VC, I fundamentally believe is a cottage industry and you can actually be very successful in the cottage industry if you're very charming. So charm, the best venture capitalists are very charming people and that's how they get the right to win the deals and they can be successful in the cottage industry. But me being charming is not going to help Parafin acquire the 100 thousand small business that has to come through scale, through code, through mathematics and engineering more than of anyone individual on the team.
Turner Novak:
All right, I do want to talk to you about charming people and getting big customers and then maybe some fundraising stuff too. But I do want to not skip this, and I kind of have to ask this question first to get into it. I'm just going back to the very early days. I think you might've mentioned you did a PhD in particle physics?
Sahill Poddar:
That's right.
Turner Novak:
How do you go from that to where we are today? How did that even come about?
Sahill Poddar:
Yeah, so in some ways I was born in an entrepreneurial family in India. I was almost like the black sheep trying to do academics. I took a lot of explaining to my family members as to what a PhD in particle physics really meant.
Turner Novak:
What is it actually? What does that mean?
Sahill Poddar:
Yeah. So in short, there are four fundamental forces in nature that physicists like to believe exists. There's electromagnetism which is responsible for light and x-ray a strong force, which is what keeps the stable together. There's the weak force which is responsible for PDK, and you can kind of get to the age of fossils using that. And the fourth one is gravity. And so, particle physics and research in particle physics is an attempt to experimentally verify theories that try to come up with one explanation for all of these forces existing. And the Large Hadron Collider at CERN, which is the largest particle collider where I did my PhD, is a pursuit. In that effort, there's a few thousand people from all around the world, including the best U.S. universities who are working there trying to discover the forces in nature and explain why they exist.
Turner Novak:
And so, what kind of stuff were you working on?
Sahill Poddar:
Yeah, so I worked on a few different things. In 2012 there was a particle called the Higgs Boson, which was the last missing piece in the standard model. And the standard model basically takes three of the four forces and is able to combine them into one. That was the missing experimental evidence that ties all of the standard model together-
Turner Novak:
To make it a fourth piece.
Sahill Poddar:
No, it's sort of one force that has three different ways in which it exists in the world, but it's really one force or one model that can hold it all. And by one model, I mean it's essentially a group theory, a Lie algebra model, which physicists like to consolidate models and have one giant model that is able to explain many different things. And so this is an attempt at that. It takes three out of the four and is able to consolidate into one.
There was one experimental evidence that was missing, that experimental evidence basically proves why objects have mass. Why can you and I not move at the speed of light? It's because we have mass. And why do we have mass? And why do some objects have more mass than others is explained by the Higgs boson. What's more important though is a PhD in particle physics is fundamentally a PhD in data science. And because, at the end of the day what you're doing is you're taking very large number of data sets, crunching them through algorithms and through code doing deep statistical analysis, machine learning, in order to separate signal from noise. To see these type of the experimental evidence for something like the Higgs boson physicists collide billions of particles at once and try to collect billions of those collisions happening through several years and try to look for signal versus noise. And in doing so you end up equipping yourself with the latest and greatest in the machine learning tools.
Turner Novak:
And so, what was it called again? The Higgs Boson Collider?
Sahill Poddar:
Higgs Boson is the name of the particle, Large Hadron Collider is, yeah.
Turner Novak:
And this is it in Switzerland?
Sahill Poddar:
That's right.
Turner Novak:
Is it underground, super deep underground or something?
Sahill Poddar:
That's right. It's roughly, if I remember correctly 100 meters underground, on the border of Switzerland and France just on the outskirts of Geneva.
Turner Novak:
Is it this super long tunnel or something too?
Sahill Poddar:
Yeah.
Turner Novak:
How does it work?
Sahill Poddar:
It's circular, more or less circular tunnel, and so you kind of have to take an elevator and go down in order to view it. But most of the times when the particles are colliding, in fact all the time when the particle is colliding, you're not allowed to go underground, because it's highly radioactive, and you probably get torched.
Turner Novak:
So, everyone's sitting up above.
Sahill Poddar:
Everyone's sitting up above in control rooms monitoring screens, and essentially the whole system through monitoring systems that we've built that feed us back data that can be looked at using code.
Turner Novak:
Does it get super hot in there, like thousands or tens of thousands of degrees?
Sahill Poddar:
Yeah, I would say there's certainly heat, but more than heat there's radiation that would be invisible to the eye.
Turner Novak:
It would be like a particle that's probably ripping your skin or whatever, or tearing organs because it's going so fast kind of thing. Okay, interesting. And how does that actually work? When you're running the test, do you put some material in there and you put something on it, or put an x-ray or some kind of beam? How does it all happen?
Sahill Poddar:
Yeah, it actually starts off, given the scale of the experiment, it's a 27 kilometer circumference tunnel.
Turner Novak:
Oh wow. That's huge.
Sahill Poddar:
It's massive.
Turner Novak:
It's a circle?
Sahill Poddar:
It's a circle. And there's particle detectors on either sides of the circle that are sort of concentric to the circle that are measuring collisions coming out of the circle. It all starts off with just a bottle of hydrogen. You have a bottle that has hydrogen in it, and you connect that, you let the hydrogen flow into the pipes and you start stripping out neutrons from the hydrogen to get a single beam of protons, which then go around the beam, around that circumference and start colliding at those concentric circles that have been established over the circumferences.
Turner Novak:
Does that speed it up? Is that what that does?
Sahill Poddar:
No, you want to get a steady beam of protons, because the collision is happening between proton proton particles. In colliding protons you can essentially learn about the fundamentals of what ... A proton comprises of these particles called quarks. So, a proton is up, up, down, and so you're basically getting those quarks to collide, and proton just happens to be a really good particle to circulate in the beam, because you can get it up to very high speed and you can use what are superconducting magnets in order to get it there.
Turner Novak:
Let's say you get this all up and running and it's going, what's the thing that's being studied? What do you get out of this from watching them collide?
Sahill Poddar:
Yeah, so fundamental, it's fundamental physics. So, you're trying to, we as humans are trying to improve our understanding of the very nuts and bolts of physics and physics theories. Fundamental physics is not ... The pursuit of it is not necessarily you have any immediate applications. But there are things that emerge out of it that are hard to predict.
One of the things that emerged out of experiments at CERN was the World Wide Web. The World Wide Web was invented at CERN because two researchers went back to their hometowns after having worked there. I believe it was England and France, and this person called Tim Berners-Lee, who was English, he basically wanted to send data over to his French colleague and therefore invented the World Wide Web to do that. So, that's kind of a spinoff effect.
The second kind of spinoff that we've seen at the large Hydron Collider is the ability to build large superconducting magnets and systems, in this case a 27 kilometer circumference system. And what that essentially does is it teaches us how to build these large-scale engineering systems and maintaining them. And what is a superconductor? A superconductor is an object through which particles can flow through without any resistance, which is kind of very weird that that exists.
Turner Novak:
Is that what space is like when you're in deep space with nothing around? Is that what the same environment you're in?
Sahill Poddar:
Not quite. There is very little resistance there, but in this case we're talking about electrical resistance. So electricity, there's always resistance, which in the way we think is power. If you try to move an electron or a proton through a circuit or any sort of space, you will start to feel some resistance. But if there's a superconductor around it, it can move without any resistance. And the reason for that is through this new sort of phenomena that was discovered by actually an American physicist, Douglas Barley, the only physicists to have won two Nobel Prizes, one for discovering the BCS theory of superconductivity and the second for inventing the transistor, which are both monumental.
Turner Novak:
Transistor. This is in the '50s, the thing that is the reason that computers exist, basically?
Sahill Poddar:
That's right.
Turner Novak:
Okay. Wow, that's crazy. Then what was the point of the whole CERN project? What were you guys trying to discover again?
Sahill Poddar:
Yeah, so part of it was discovering the Higgs. So, we can say, "Hey, we understand the standard model really well. And lets kind of put a stamp of approval and let's look for newer things." The part of the newer things was also discovering things like the existence of extra dimensions. And so, the simplest way to explain that is, in physics we see that the three forces, electromagnetism, strong force and weak force are all much stronger than gravity. One simple explanation is that the reason gravity is perceived weak to us is because it is escaping into extra dimensions. And as a result, when we try to measure it, we are only measuring its strength in the three plus one dimensions that we are aware of, which is three space and one-time dimension. The mathematics of it works really well.
However, not everything where the mathematics works is nature's secret, you still have to experimentally test that. The test we do at CERN is to essentially measure whether graviton or gravity particle is escaping into extra dimensions. And what we really discovered there and continues to be true is that because where the Higgs was found, the mass at which the Higgs was found meant that a lot of these new types of theories are not going to be discovered at CERN. And-
Turner Novak:
Wait, why is that?
Sahill Poddar:
The mass of the Higgs constrains a lot of new theories and says, "In order to mathematically have a lot of these new theories exist, you have to let Higgs be a little bit heavier than it was experimentally observed to be." By achieving the Higgs discovery, which was a phenomenal achievement and actually led to Peter Higgs winning the Nobel Prize in 2012, a lot of the future dreams of CERN were in some ways shattered, which led to folks like myself branching out of physics after my PhD and moving to industry.
Turner Novak:
So, you mentioned before though you were thinking about leaving, going out of academia more in the industry. How did that come about? Because I know you were at Robinhood, you didn't actually go right to Robinhood. What was kind of the journey there?
Sahill Poddar:
Yeah, so once I finished my PhD, it was pretty clear to me that the opportunities that were presented at CERN would be fairly limiting in terms of how exciting it would be for what it meant for our understanding of the universe. I decided to look around and see where could my skills be applied, and data science was picking up as a pretty hot topic in tech. This is like early 2010s. And so, I was living in Germany back then doing my PhD in Europe, and so I decided to move to industry and use my skills in the tech sector. And it just happened so that within technology one of the largest data sets that existed, I think until today, is on advertising. And how to monetize ads data for businesses. And so, I worked for a few months at a startup based out of Berlin before moving to Meta, then Facebook, and being part of their ads team in Europe and grow that business, which at the time was the fastest growing GEO for Meta and Facebook.
Turner Novak:
Then what kind of stuff are you doing there when you're running, working with these large data sets? For somebody who is not aware, just the scope or the scale of that, what does that entail inside Meta?
Sahill Poddar:
Wat that entails is essentially using large data sets from customers, interacting with ads, clicking on ads, buying behavior, and across many different advertisers. And you can imagine a company like Facebook has essentially billions is the floor on any kind of data set they have with this. Using that to predict what future performance for advertisers could look like. And coming up with better targeting mechanisms. And so, again, not only is Facebook collecting data on what ads you're clicking on, but also what pages you're interacting with. Facebook has this thing called the Facebook Pixel, which is on pretty much most websites across the world-
Turner Novak:
To see how you're interacting with other websites?
Sahill Poddar:
That's right. And so even if you have not told Facebook that you like Nike shoes as an example, but on Nike Shoe website Facebook knows. And Facebook knows who you're dating and by virtue of the two people living together in the same house, or who your partner is, et cetera. And so, they can leverage that type of information to create better and better targeting. If you went to a Nike shoe website, they'll probably show your wife the ad as well, because, "Hey, as a household, they're probably interested in Nike shoes." And-
Turner Novak:
Deal with partners in the household. That's interesting, yeah.
Sahill Poddar:
Oh yeah. Partners in a household. I mean, you have to kind of disambiguate between roommates and partners, but they're able to do that pretty easily, their models. And they can even predict if two people are going to start dating but haven't started dating yet.
Turner Novak:
Interesting.
Sahill Poddar:
Just based on interactions and models that they have. And what helps them get a true data set is very often people will post their relationship status on Facebook back then and say, "Hey, I'm dating X, Y, Z, or I started dating X, Y, Z. I got married to this person." And so they have a full picture of when that happens and how their interaction started and how that evolved over time. And so, you can use those type of signals to target people better to make sure you're showing them more contextual ads, which is better than showing a ad for a random thing. It's better for the user, because they're seeing something more relevant to them. It's better for the advertiser, because reaching these users at a much more cost-efficient manner, and it's just overall better for the ecosystem.
And it allows a company like Facebook to scale and spend huge amounts of money on CapEx in building things like Llama. Facebook in particular has been exceptional in giving back to the community, having built some of the best open source projects. But also, now with AI being open source. So, data science is an excellent tool to help companies like Facebook, Google, et cetera, monetize their ads business and ads network by getting advertisers more efficient and by getting the ads that we put in front of customers a lot more contextual.
Turner Novak:
Are there any lessons from the Facebook days that have been helpful today with Parafin, specifically on a lot of data science across all these different platforms, you probably have more data than just a Facebook platform at this point?
Sahill Poddar:
It's obviously an exceptionally well-run company, both Facebook and Parafin, but the thing that Facebook does really well is emphasize the done is better than perfect. And the speed of getting things done. I mean, today there's other companies as well that maybe have even taken the baton from them, but just like early 2010 Facebook was moving extremely quickly, much faster than others in the space. And even for its own size and had a pretty low bureaucracy, fast-moving culture. Done is better than perfect was plastered across pretty much every Facebook office on their walls. And so, that's one cultural trait that we value a lot. We call it the ability to compress time and get things done. Since I think it's largely true that speed is the only real advantage that a venture-backed company has against an incumbent. And if you lose sight of that it can be pretty dangerous. So, keeping that as part of our culture and DNA is extremely important.
Turner Novak:
And so, then talking about Robinhood, I know, took you a while to convince them to give you those. How did that kind of all come about? Because it wasn't like immediate, bang interview, you're hired. How did that process go?
Sahill Poddar:
Yeah, so I happened to be in the U.S. working for Facebook, and I was on a U.S trip. And I really wanted to move to the Bay Area and be closer to the metal in terms of being in Silicon Valley working for a tech company. And through some stroke of luck I met a venture investor who had invested in Robinhood a seed fund, which now is defunct.
Turner Novak:
Oh, wow.
Sahill Poddar:
And that person introduced me to Robinhood. And even that was extremely serendipitous. Because I went to this person's office, the fund was called Rothenberg Ventures, and I went to their office and as I was talking to one of their partners he was showing me around his virtual reality lab and introducing, talking about all these other companies and none of them are really interesting or already relevant to someone who could do data science.
And then as I was about to walk out of his office, he just said, "Hey, I have one more thing for you. There's this seed company called Robin ... This early stage company called Robinhood that I've invested in. Maybe you should talk to them. The founders are also mathematicians and physicists." And so I was like, "I'd love to, I just read about them having raised around-
Turner Novak:
Had they launched yet?
Sahill Poddar:
I think then probably on the cusp of launching, I think it was maybe a little bit pre-launch. And so, I reached out and I got connected to who then became my manager there. He was the head of data science and data engineering and was one of the first five employees of the firm. And as I got to know him over time I was borderline harassing him with each trip that I was making out of the Bay Area to work for Facebook and saying, "Hey, can you meet me again? Can you meet me again?" And at some point he just turned around and said, "Hey dude, do you want to just come into an interview?"
And I was like, "Yes, that's exactly what I was hoping for." I met them and roughly it was a few dozen people, interviewed, did not hear back from them for several months. And then one fine day I get a response from the HR saying that, "Hey, they've accepted me and they have an offer, et cetera."
Turner Novak:
Just out of the blue?
Sahill Poddar:
And they'd like to work on a work permit and a work visa, which by the way is also not trivial to get for the U.S., as I'm sure you're aware. And so, in the end it all worked out, they helped me through the process. But I think the end-to-end process took about a year of me first contacting them versus me starting working there. And so yeah, at the end of the day it, persistence and perseverance paid off. They probably saw something in me that they liked, and I was very fortunate to be part of the early team there.
Turner Novak:
We'll, what was probably the single greatest lesson you learned at Robinhood?
Sahill Poddar:
Yeah. Well, Robinhood taught me many lessons. And so, before I started working for Robinhood I knew nothing about financial services.
Turner Novak:
Really? Okay.
Sahill Poddar:
In fact, the first time I bought a stock was after I met the Robinhood team, I went back to Germany, opened a German brokerage account and started buying stocks to even see what the whole experience was like. And by the way, I was roughly 30 years old then. So, it seems, today when you look at 21, 22-year-old-
Turner Novak:
We've got high school, middle schoolers in Robinhood and buying meme stocks.
Sahill Poddar:
Exactly. And so, I was pretty much, much older than one would expect someone dabbling in a stock trading in the U.S., but in Europe most people do not trade stocks. It's much more different mentality. It's more of cash, preserving capital versus investing and trying to grow capital. And so culturally it's a very different place. So yeah, many lessons learned along the way. I think everything about the U.S. equity markets that I knew until I left the company was due to Robinhood. Everything I knew about option trading was due to Robinhood. Everything I knew about building a fast-growing venture-backed company was due to Robinhood. Learned how Robinhood thinks about building products, achieving product-market fit, measuring product-market fit.
Turner Novak:
How does Robinhood think about building products?
Sahill Poddar:
Yeah, I think first and foremost, you have to ask yourself the question, what is the status quo? Where do you have a right to play? And most importantly, where do you have a right to win? And so, extending equity trading into option trading in hindsight seems pretty obvious. But option trading, as it turns out, is a much, much more complex thing to enable, which is why we haven't really seen many other Robinhood competitors enable that. And you can easily lose your shirt if you get it wrong. And so, how do you go from a working system, which is already complex to building something that is 10X more complex, has many different risk considerations, CapEx considerations, responsiveness considerations, entering new product categories like Robinhood until very late until very recently was treated like the stock trading app with fund money in it. But more recently, if we've been tracking the company, they've been growing AUM very steadily and launching all these cool new products.
There's a Robinhood credit card, there's a retirement account, there's passive investing, there's crypto, et cetera. And so, how do you think about entering adjacent markets? By the time I had joined that company they had pretty strong product market fit on the core product, even though it was on tens of thousands of users, you could see that there was some magic happening there where each user was coming back pretty much every day due to track their portfolio.
Turner Novak:
Interesting.
Sahill Poddar:
And I think the thing that Robinhood got really right in hindsight with their first product category compared to competitors. If you remember the time there was Wealthfront, and taking off at the same time there's Robinhood and there's few other sort of wealth management passive investing.
Turner Novak:
Betterment was maybe one.
Sahill Poddar:
Betterment is another one, yes. And both those products got to a pretty, have pretty decent AUM over time.
Turner Novak:
Yeah, I think Wealthfront just tried to go public, didn't they?
Sahill Poddar:
That's right. I think 80 billion, roughly 350 million of revenue, so-
Turner Novak:
Yeah, I think pretty profitable, like 40% margins, if I'm remembering right. Good business, good profitable company, probably worth a decent amount in the public markets.
Sahill Poddar:
But not growing as fast as Robinhood, for example, not entering as many adjacent product categories. And if you were to say who replaces, who is the Charles Schwab of the new era? I think very quickly you would deduce their answer is Robinhood and not something like Wealthfront. I think what Robinhood got really right is the team there early understood fundamentally that this dream and this hope of unlimited wealth generation, Lie, and if you can capture that, which is kind of ingrained in American culture. If you do the right things, there is unlimited upside potential that exists in the United States.
And kind of productizing that and tying that to U.S. equity markets led to great things. In Wealthfront, by definition there's only sort of 10% return you can get max on an annual basis, because they're tracking the SB Bio or whatever your favorite index is. But in Robinhood, you are kind of in charge of your own destiny by investing in the right categories or the right stocks, and doing so at the right time. You can beat the market like 10 or 100 x over. And capturing that zeitgeist in that American part zeitgeist and tying that to U.S. equity markets and productizing that is what they got absolutely right from the get go.
Turner Novak:
Yeah, that's a very American thing of individualist ability to go and pick a la carte, the freedom of choice, but then also unlimited upside optimism, et cetera. I feel like Wealthfront's maybe more of the European approach or something like that where it's like ... Neither is wrong, but.
Sahill Poddar:
Yeah, I would say Wealthfront is pretty good on the capital preservation side and Robinhood is better on the capital. It's Robinhood is good to make money and Wealthfront is good to kind of ... Robinhood is good to get rich. Wealthfront is good to stay rich. Or the Wealthfront approach is good to stay rich. But fundamentally, if you're going after the millennial user base that are more likely to adopt a mobile product, you want to attract how to get rich cohort and not the how to stay rich cohort.
Turner Novak:
Yeah, reminds me, I think of when I'm talking to, I don't know, Gen Z, people like 19 leaving college, whatever, I feel like the mindset right now that young people have is you kind of have to go to ... It seems like the philosophy is you kind of need to shoot for lottery tickets. That's really the only way to get rich is winning the jackpot. It's hard to do the traditional American dream and the Wealthfront approach of slowly grind it up. You just don't get anywhere. So you have to just keep looking for those spikes where you win the lottery, you're like, you buy NVIDIA call options and you get a 1,000X and that's how you get rich. It was just very different from the way, growing up mine was like, buying my first stock through Scottrade, right? It's like do all this research and you got to contribute to your retirement account every month and make really smart decisions and go slow.
Sahill Poddar:
Yeah, I would say yes. But there's sort of even on the lottery ticket side of things, there's measured ways of doing things that are more repeatable that to someone on the outside might look like a lottery ticket.
Turner Novak:
Yeah, venture capital right there.
Sahill Poddar:
Yeah, I mean, yeah, that's right. There's funds that have vastly exceeded market returns and done so for several decades and remain as the best in business. And that is a systematic sort of formulaic way of doing things that's repeatable, versus buying an NVIDIA call option before earnings and hoping the stock pops. That is more in the extreme lottery ticket category versus a measured way of doing things. So, since the latter is not repeatable you never know if you're going to be able to create that same video option call and make those same returns again.
Turner Novak:
Did you learn any interesting of growth related? I know you were on the growth team at Robinhood. Did you learn any sort of growth frameworks or things that you've carried, being able to carry through throughout your career from the time at Robinhood?
Sahill Poddar:
Yeah, absolutely. My time at Facebook actually taught me ... And there was some great material which Facebook published, internally then, but I think it's external since then, around measuring product market fit, focusing on the aha moment of why is a customer coming and what makes them stay on in an app. And so, we kind of ran those similar analysis of Robinhood. We were seeing extremely retentive user base, which was a clear sign of a product market fit in a large market. But we kind of take one step further and say, "Okay, why are these users being retained? Let's try and understand what is the action that is leading to retention." And if you do the right set of data science analyses you can back out into the fact or the insight that when a user comes and when they buy their first stock, and in fact when they sell their first stock is what leads to the maximum retention.
Turner Novak:
Really because I would think sell. You're withdrawing your money.
Sahill Poddar:
Yeah, well the sell matters because that's when you realize gains. And you can kind of track actions and be like, okay, buy leads to good retention, but actually sell leads to even better retention. And there's not that much slippage between the buy and sell. Because most people buying a stock eventually want to sell it. So, let's try to make sure that we can make people do that, buy, sell action as quickly as possible. So, how do we back this into something that can help us grow the user base? And so, we took that insight and one other sort of experiment that we ran at Robinhood, which vastly successful, was our referral program. And so, initially the Robinhood referral was some flavor of, "If you refer a friend, you can get $10, which you can do whatever you like with," most likely you'll buy a stock.
Turner Novak:
That seems like pretty straightforward. That's pretty common growth strategy. A lot of people give out free money.
Sahill Poddar:
It is a common growth strategy, but I want to say in 2015, 2016 it was not as widely adopted by the market as you would expect it to be. But that alone doesn't work as well giving $10, because you could be doing whatever you want with $10. We haven't really given you a quantum of the experience that we want you to have from the product. And so, how can we change that into a quantum of experience? Then you can kind of go one step further and say, okay, you can only buy a stock with this $10. Turns out that doesn't work as well either. What the team did then is we essentially took ideas from behavioral economics that humans love unknown variable rewards. And people love scratching a card from a lottery ticket and seeing what they got and things like that.
Turner Novak:
That's great white elephant gift by the way. Is you just buy $25 of lottery tickets and put it out, and people go crazy over that. If you ever do a white elephant party, bring lottery tickets, always a hit.
Sahill Poddar:
That's a great tip. That's a great tip. Or maybe an unknown stock, right?
Turner Novak:
Or unknown stock. Yeah. So, that's what you did at Robinhood.
Sahill Poddar:
With Robinhood we went with the unknown variable reward, and the reward was it could be a stock of one-off end companies, the expected value of which is call it $10. But there was a small probability that you got a stock of Facebook or an Apple, which I think around the time was like 100 plus or 200 plus dollars, but most likely you would get a stock that was like $10 or slightly under $10.
Turner Novak:
So, you would get one share of stock and the price was around $10. It might be GE worth 9,50 or like-
Sahill Poddar:
That's right.
Turner Novak:
AT&T trading at 17 or something. You just get one share.
Sahill Poddar:
You just get one share.
Turner Novak:
Interesting.
Sahill Poddar:
And that way, what we've done is we've given you a quantum of the smallest possible experience of the product. Because now you have a stock, you're tracking it, you likely sell it to cash out-
Turner Novak:
Because you're like, "What is this AT&T?" Like, "I don't want this."
Sahill Poddar:
And by the way, we know that once you buy and sell a stock you're highly retentive. These users will turn out to be users that eventually generate revenue and high LTV. We took that insight and that particular experiment, just blew every other experiment out of the water in terms of performance, CAC and LTV performance. And we just started pouring the fuel on the fire. And I think it's obviously difficult attribute how many users Robinhood has today from this, but I would say a big amount of a user base was captured using this technique. That was a mix of learnings from some learnings I had from Facebook, behavioral economics, the Robinhood teams referral program, and tying that all together. And a few different iterations of that that led to the winning formula and the winning experiment. A second one was gaming the Google Ads network, and maybe gaming is not the right word, because it sounds like we did something-
Turner Novak:
You cheated or something.
Sahill Poddar:
Something fishy, but it was mostly being highly scientific around how do we leverage Google's ad network to send us the best users that Google has for our product. And using same, similar techniques, we basically built a LTV prediction model very early on. And based on the user's one or two day action, you can tell Google that you really like this user, send me more of this. There's a trade off that if you wait for too long to send Google the signal, Google's ability weakens to send you similar users. But if you send it too early, your ability to identify good users is not good enough-
Turner Novak:
Isn't tested enough or refined enough.
Sahill Poddar:
Isn't tested enough. You have to build a model which relies on little amount of data and sends what we call a post back fast enough to the Google network.
Turner Novak:
So, this is within two days is what you found?
Sahill Poddar:
Yeah. We found that actually within one to two days there's enough traces that a high LTV user leaves in the product, and we can basically use that to inform Google's ad network to send us more similar users.
Turner Novak:
Is this like lookalikes? Is that what you'd call it?
Sahill Poddar:
These are lookalikes, but these are custom-built lookalikes. These are post-packed lookalikes. So, we are posting back ourselves what kind of lookalike we want, versus Google telling us what kind of lookalike we should get. So, highly customized version of a lookalike.
Turner Novak:
What was the leading indicator of a really good Robinhood user or profitable user?
Sahill Poddar:
Yeah, so it was around the number of pages they were looking at, what kind of pages they were looking at within the product. Were they interacting with the bank linking flow? Were they initiating a transfer? Even what instruments they were buying initially led to intel on their long-term LTV.
Turner Novak:
So, you could tell how active they would be, how much money they might have, and what kind of dollar amount might be flowing through. And then go back to Google and say, "Find people that have maybe this income threshold or searching for these specific types of finance or investing-related topics?"
Sahill Poddar:
That's right.
Turner Novak:
Okay. Interesting. And then it's interesting too when you talk about the reward-based referral program. It sounds like basically you figure out what are the profitable, what are the most valuable users that you want, and figure out how to incentivize getting more of those. I mean, it sounds super simple. I'm really kind of reflecting back on it, but do people not do that usually, or?
Sahill Poddar:
No. So, it does sound simple, and I think that is, it's a good sign that it sounds simple. Any sort of insight that-
Turner Novak:
It's too complicated, you can't replicate it. Is it real?
Sahill Poddar:
Post-hoc should sound simple, but what actually it entails in building and experimenting and testing. It is a highly scientific endeavor. But we should be able to explain it in the simplest way as possible. And as simple as it sounds, it does take quite a lot of effort from multiple different folks on the team to get there.
Turner Novak:
Can you predict churn?
Sahill Poddar:
That's another problem we worked at Robinhood and obviously is a big part of our modeling at Parafin is to predict the churn of a small business. Either they completely go out of business or they lead the platform that we are working that we require them through. At Robinhood we were also predicting churn and basically measuring user activity. What pages are they looking at, how much time they're spending on the app? Did they just do a terrible trade and they lost a bunch of money, so they're unhappy, et cetera in order to predict churn? And we were pretty good at it.
Turner Novak:
Then you tie that back in also to the acquisition of figuring out. Can you predict what types of behaviors they might do in the future that those behaviors lead to churn?
Sahill Poddar:
Yeah, I think the actionable on churn is there's few different things. You can fix things in the app that are broken, which are leading to customer churn, but that typically you'll also get signal from your customer support teams that will hear that, "Hey, users X, Y, Z are complaining about this." The second category is if users are losing a lot of money and churning, you can be better around educating them around what they're about to buy and sell. "Hey, this is a volatile stock. These are the metrics of the company, so be aware of what you're getting into." And doing so in a way that is in the American spirit, that you are doing your best to educate but not making the decision for them. And then the third category is a little bit harder when folks are, they're just saying, "Hey, I'm kind of done investing now, because I have some other personal life financial goal that I want to meet." Maybe it's buying a house, maybe-
Turner Novak:
So, you'll sell and withdraw?
Sahill Poddar:
Yeah, and that's a little bit harder based on the data you have at Robinhood to predict. And you can still predict it, but get an insight into why they're doing that is a bit harder. And I think the best way to address that is to build financial tools that can help capture those transactions as well. So, we've seen Robinhood get into mortgages recently.
Turner Novak:
Oh, do they? I was just going to ask, do they have that yet?
Sahill Poddar:
Yeah, they do. I mean, I was playing around with my app a few days ago and I saw that you can apply for a home loan through Robinhood. They have a partnership with Sage Loans, I believe. And so, capturing more surface areas through products such that you can meet the user as they're doing more financial transactions is the way to stop that churn.
Turner Novak:
Okay. Then how did Parafin come about? I know you're at Robinhood for a while, how did this whole ... It seems like from what you described, you had an entrepreneurial family, so maybe it was like the indications were always there, but it doesn't seem like you were like, "I want to start a company." So, how did this all come about?
Sahill Poddar:
Yeah, so entrepreneurial families are certainly, I think there's always been a push from the family, "Hey, when are you going to start your own business?"
Turner Novak:
Cut out this academia thing. Like your mess around with stock trading, come on.
Sahill Poddar:
Exactly. And they were kind of disappointed in me when I was leaving meta to join Robinhood. Or not disappointed, but rather questioning the decision, "Hey, why are you leaving well-established company like Facebook to join"-
Turner Novak:
A startup, it's too risky.
Sahill Poddar:
A startup and taking a pay cut, et cetera, et cetera. So, there was some educating that has to be done, no fault of theirs. It's such a fast-evolving world that I probably have similar concerns if my son was making these decisions at some point. So, there's obviously that push from the family, "Hey, when are you going to start your own business?" And then when we were Robinhood, what we saw was we were undergoing through the mobile era a consumer finance Renaissance of sorts. There was Coinbase, Chime, Robinhood, Venmo, and maybe 50 other apps trying to for the U.S. consumer for financial services.
However, small businesses that have been closed to both our hearts and my heart and mind was not being addressed the same way with financial technology. Yes, there was Square and Stripe to some extent doing things where they were making it easier for businesses to process payments and solving for funding through earnings problem. But no one was really solving funding through debt problem for small businesses. With the exception of Square Capital and to some extent Shopify Capital, Square Capital was doing a decent amount of embedded lending, and Shopify had just started on that journey doing embedded lending as well-
Turner Novak:
Those certain types of businesses, like Shopify is just E-commerce sellers.
Sahill Poddar:
That's right.
Turner Novak:
A couple of percentages of all business in the U.S., like Square, it's a physical store or location. And again, it's not like, it's a pretty large swath that's not being served by those products.
Sahill Poddar:
That's right. And they do what is called a first-party product in that they only serve the merchants that are transacting using Square or transacting using Shopify. But there is a vast sea of opportunity outside of those platforms with platforms like Walmart, DoorDash, Amazon, and all the different payment processes that are out there that are not Stripe and Square. And so, we saw this opportunity of essentially bringing a similar product to life what Square had done in the brick and mortar space and Shopify had done in the E-commerce space, but doing it in a way that is more robust than what they do, and that is being vertically agnostic.
When you're vertically agnostic, you are inherently less risky to your capital supplier, because you're not indexed on any one particular vertical and therefore macro movements in that vertical. For example, 2021, 2020-2021, we saw a big boost in E-commerce driven by COVID, but '22-'23 we saw mean reversion in E-commerce. And so if you were doing E-commerce landing in '21, you could be fooled to think that this party will continue and you should be giving bigger and bigger loans to these E-commerce businesses.
Turner Novak:
I mean, if you're looking at the spreadsheet, AOV going up, CAC going down, velocity, growth rates increasing, and if you're just looking at data and numbers, they can fool you.
Sahill Poddar:
Exactly. And if you gave out loans to E-commerce business in 2021 with those biases, come 2022 you were having difficulty getting that money back.
Turner Novak:
I mean, probably on average E-commerce seller probably shrunk in 2022.
Sahill Poddar:
That's right.
Turner Novak:
And that's not good if you're lending them money.
Sahill Poddar:
That's right. That's right. And by being vertically agnostic, we can essentially absorb macro movements in any one particular vertical. So, today we work with pretty much every major U.S. vertical that is out there. There's about, in the small business category about 60-ish million Americans work for small, medium businesses. The largest single largest category in that is restaurants, which employs somewhere around 12-ish million Americans. And so, naturally the distribution of businesses we work with is indexed on their contribution to the economic activity, but we cover pretty much every major U.S. particle.
Turner Novak:
What about McDonald's? Is that a certain, or is this excluding the big chains?
Sahill Poddar:
McDonald's locations have taken loans on Parafin through DoorDash.
Turner Novak:
Oh, because they're franchise-owned obviously?
Sahill Poddar:
That's right. We've had McDonald's owners take cash advances. While DoorDash we've had Papa John's. We've had both large national franchises, but also regional chains that have 10, 20 locations use our product.
Turner Novak:
I know that you did quite a bit of research before you got into this. Did you know, "I'm going to do small business lending," or was it a process of doing some research? I know user research, there's a really strong culture at Robinhood. Did you learn anything there that related to how you approached this?
Sahill Poddar:
Yeah, for sure. One of the things which was at the time we really had to fight the urge to was write code at the beginning of starting a Parafin. We were three technical founders. When we started the business, the natural hunch was, "Let's get started. Let's start writing code. Let's build something that we can sell." But instead, we were very intentional about just going out there and spending the first three months talking to the market and talking to platforms. And I still distinctly remember, we would show up in the office and the three of us would look at one another and be like, "Hey, what are we doing?" I said, "Well, we have these 10 calls lined up for the day."
Turner Novak:
Customer conversations?
Sahill Poddar:
Potential customer conversations. Many of them were not the right ICP, but we were just going one after the other, talking to these platforms and even D2C businesses, trying to understand where are the pain points, where they lie.
Turner Novak:
So, you knew SMB lending or business lending in some capacity?
Sahill Poddar:
Actually, even one step before that, we knew SMB Financial Services, what we were ... One hypothesis that we had, which was falsified, was that we could build a zero-cost payment network on the back of ACH rails. Because we were doing a lot of that at Robinhood, moving a lot of money from consumers to the brokerage, and being like, "Hey, if we could bring this experience backed by something like Plaid to the SMB space, we can make it cheaper for small businesses to earn money."
Turner Novak:
And if you look at Target has an ACH, do you know Target's RedCard?
Sahill Poddar:
Yes.
Turner Novak:
Yeah. So, if people are not familiar with, Target has this thing called the Target RedCard, it's probably like 10% of all Target's revenue, it just run on basically this internal card program. It's like the Target credit card or the Target debit card. And when people pay with the Target RedCard, it's all on ACH, so they they on about 10% of their, I don't know, $40 billion business, I actually don't know how big Target is today. But if you just think about 10%, you're not paying any payment process fees, so that two to 3%. I mean, I don't know how big they are, but that's probably like $100 million in costs that you cut by just building this thing. And it's probably mostly free cash flow. And if you capitalize that, that's like a billion dollars of market cap or maybe more than that, that Target has saved by building this internal program. It's actually pretty powerful.
Sahill Poddar:
And it's incredible. And many countries out there, like India, Brazil, have naturally real-time payments. There's some version of that here in the U.S. as well, but the system that we all rely on more often is ACH. And so, we were asking ourselves a question as, "What can you do in this space?" It turns out small businesses actually don't care about the 3%, or rather it's not the top priority. They would rather not compromise a conversion in order to save that 3%. And anytime you put a non-card related product in front of the American consumer, they may shy away from that or might start looking for, "How do I use a card to make this transaction happen?" I think over time someone like Apple or some version of a company like that as well positioned to break mold there. But I don't think this is something that we wanted to pursue doing, because we very quickly falsified that this was wrong.
And what in fact mattered more to small businesses is not to make that funding through earning bucket more efficient, but being able to fund their business through debt, which was a completely untapped opportunity. That was a much solving for a much bigger need, was much more of a green field. And the emergence and the scaling of marketplaces, vertical SaaS point of sale solution meant that we could now tap into data sources that would allow us to build a product and sell it, which is responsible, which is sustainable, and which can help essentially lead to a venture-backed company in the sense that we can raise capital and generate good returns on it for investors. Because we've identified a large market where we can scale into using code and data designs.
Turner Novak:
I mean, it seems like that's been a big thing you've harped on a couple of times is the differentiation is like helping them fund with debt, versus funding with equity.
Sahill Poddar:
Funding with earnings.
Turner Novak:
With earnings, yeah.
Sahill Poddar:
Yeah, funding through sales. Which is what payment solves. Payment solves for the funding through earnings and then know Parafin solves for funding through debt.
Turner Novak:
And I know a lot of times people ask you for advice on fundraising. Did you raise a little bit of money around this time to start building stuff, or did you wait until you already had some customers? How did that go?
Sahill Poddar:
Yeah, so I want to say our seed financing was probably one that I would never advise a founder to follow. It was very unique. My co-founders and I did not have work permits to start our own company. We had work permits to continue working at Robinhood but not to start our own company. And so we went up to Robinhood investor, Rivet Capital, who's our largest shareholder. And we said we wanted to start our own company and we had sort of half big ideas around small business financial services. And they mentioned that we should go and have the conversation with Vlad and come out in the open and let him know. And once we did that, they'd be more than happy to fund the business. And so, we went ahead and did that. And in doing so they got what was the best reference check from Vlad, which is when they asked him what he thought about us, he said, "He was disappointed that we were leaving and he didn't want us to leave." And when venture investor hears that they obviously want to fund you even more than they did a few minutes ago-
Turner Novak:
That's always the best signal. It's like prior boss doesn't want you to leave, trying to keep you, but he's investing. They know that you're going to be good.
Sahill Poddar:
Yeah, that's right. Rivet ended up leading our seed round. They've been great partners ever since. We did not run a conventional seed funding process. We did not talk to too many investors. It was just a handful and literally a couple. And Rivet was one of them and the one that we liked the most.
Turner Novak:
And this was because you needed some money to pay for visas?
Sahill Poddar:
That as well. So, there was a chicken egg problem in that we had to fund the company in order to get visas and in order to start working, and obviously that is through a venture capitalist who you don't know well, they'll be like, "Well, I'm not going to really put money in your bank account if there's no guarantee that-
Turner Novak:
"Show me that you've built some product, talked customers, et cetera."
Sahill Poddar:
Exactly. At this point we had no product. We barely had a bank account to put the money in. And so we got that set up, Rivet funded it, we ended up getting visas that led to us starting Parafin. So, it was very unconventional fundraising round, the seed round, but it started to get more conventional from the series A onwards where we started to go to more investors. But in the early days, even for series A, most of our Rolodex was existing Robinhood investors, since we deeply trusted them, we knew Robinhood had had excellent experience in working with them. And some of us had also interacted with these investors as part of Robinhood's fundraising round. For example, I would pitch for investors when Robinhood would be fundraising.
Turner Novak:
And then you mentioned before that DoorDash is a customer that you work with. They're a pretty early customer that you had. How did that go? I don't know if they were the first, but getting the first people to agree to work with you, because the approach that you took, it's like, "Hey," obviously going direct to the business might be easier as a startup of like, "Hey, random business, we're trying this thing, try our product." But when you go to DoorDash, "Okay, we're a startup, doesn't exist." How did you land that?
Sahill Poddar:
Yeah, honestly, there was a lot of uncertainty. In hindsight there was a lot of uncertainty, and I would say huge kudos to the DoorDash team for taking a bet on an unproven company. I distinctly remember there was one conversation where we had with them and they were like, "What's the size of the company?" And I turned around and said, "We're all on the call right now." And it was just five of us on a call with three of them.
Turner Novak:
Okay, that's pretty very big, big-ish. You outnumbered them at least.
Sahill Poddar:
We outnumbered them. But we had pitched to them early. They were very intrigued by the product. What helped us was the DoorDash team had some folks who had worked at Square, so they understood what the impact of having such a product is for their customer base. They were extremely scrappy, so they didn't care that we didn't have a built-out product. They were willing to do things on a very manual basis to get it up and running and testing product market fit, which typically large companies don't like to do. And after we made the pitch and presented them, they kind of disappeared for a few months and they said, "Hey, we are busy, we'll be back." And we thought we'd lost them forever and they disappeared into the ether.
But what turned out to be true is that they went public at that time, and so they were not doing any net new vendor deals, specifically one around financial services. And then as soon as they went public, one month later I saw them on one of my documents. I got a notification, which I shared with them that someone at DoorDash is on this. And then a few minutes later we got an email saying, "Hey, we would love to jump on a call and push this."
Turner Novak:
Like a Google Docs?
Sahill Poddar:
Exactly, exactly. I think it was DocSend or one of these other products. But I saw it in the evening and I distinctly remember messaging my co-founders, and we were all giving us ourselves the rocket emoji in Slack. And they moved pretty quickly. So, we got an agreement signed. I believe they also ran an RFP with Stripe, which was a competitor. Stripe Capital has a competing product in the market, and DoorDash is one of their largest customers. And what we understood in our conversations with these large platforms is that they actually do not want partner with Stripe for lending, because lending is a way for Stripe to entrench platforms deeper into their payment ecosystem.
And as you're growing, you actually want to negotiate your payments rate to be cheaper and cheaper. And you lose that leverage if you are signaling that you're willing to be entrenched in their payment ecosystem. And so, by counter positioning and saying, "Hey, we can actually build this in a payment process diagnostic manner, we have enough ACH payments experience from Robinhood that we believe we can manage the risk here." We were able to offer a product that allowed platforms like DoorDash to break away from Stripe. And since then we've actually gotten many other Stripe customers. Mike Bode, which is Stripe Connect customers does payments with Stripe, lending with Parafin. Jobber, which is Stripe Payments customer does payments with Stripe, lending with Parafin. So, we've been able to amass both non-Stripe customers where the larger opportunity lies, but also Stripe customers.
Turner Novak:
And you've never really done sales before?
Sahill Poddar:
I have never done sales.
Turner Novak:
How did you get good at sales? It seems like you like it now. It seems like you kind of like this chasing the deal, closing it, getting the rush. What did you learn to appreciate about it?
Sahill Poddar:
Yeah, I think I've certainly become a deal junkie, for lack of a better term. My co-founder jokes that I can sell a bald guy a comb, even prior to starting Parafin, because I would keep pitching him on various different things. I think it was just some part of the DNA, it just kind of got to flourish once I started Parafin. And what I've seen work is, in our product category buyers are typically extremely smart. They are very well-informed. And with the age of AI, you have to assume that knowledge is free-flowing and all the time. So, I really try to get as analytical as possible in a sales process and down to the details and getting all the details right, we've seen that really helps. Aligning incentives really helps, and showing hunger for that business really helps. Our pitch to DoorDash in the early days was, the way I would end one of the calls was, or tell our DoorDash, POC that you have in front of you on the call is the DoorDash capital team. We will work-
Turner Novak:
We're like your dedicated support team.
Sahill Poddar:
Exactly. We will work as hard for this product as an internal team would, or perhaps even more. And so feel free to push us as much as possible at any time of the day, at any day of the week, and we'll make this happen. And just going with that commitment to every single customer is important. B2B infrastructure or SaaS for that matter, the way I like to frame it is the art of selling mass-made suits, but making the buyer feel that they're buying a custom-made suit, right? So, as a B2B buyer you want to know that this vendor is dropping everything else and just paying attention-
Turner Novak:
You're the only customer.
Sahill Poddar:
But as a B2B vendor, you want to be able to build a product that you can sell to many people. In some ways behind the scenes you're building a mass-market product, but to each customer you have to make them feel like it's personalized for them. So, you're building a mass manufacturing suit, but you're selling it to each customer like it's a custom-made suit.
Turner Novak:
Is there anything else that you kind of found secret sauce in these partnerships go-to market?
Sahill Poddar:
There's a few tricks that we've used in the past that have helped us in some ways. For example, when platforms, one of the questions that platforms would really come up with is, "We don't think our customers need this product."
Turner Novak:
Interesting.
Sahill Poddar:
I was like, "All right, we can actually debunk that pretty quickly. How about I run a survey with my own money and go and reach out to your customers, because I can do that in the open web, and ask them questions about, 'Do you need capital? How likely are you if platform X were to offer it?'" And so on and so forth. And it turns out you will discover the same thing over and over again, which is businesses do value getting easy access to capital. They're more likely to take it from people who they know versus people who they do not know. And so, we would take these surveys to them and say, "Hey, look, your own customers are saying that they need this."
A second aha moment for us was, one day I was with my then-girlfriend-now-wife walking out of lunch from a friend's place on a weekend. And we walked into an auto shop and I told my wife that, "Hey, I need to speak to the auto shop-
Turner Novak:
About your car?
Sahill Poddar:
Well, no, it wasn't about the car. I went in there and she came along with me and I asked him, "Hey, how long you've been running a business? I see you only have one lift to lift cars up so you can fix them. Have you ever thought about getting a second lift?" And he says, "Yeah, I would love to have a second lift." I said, "Have you ever considered taking a loan?" And he's saying, "No, but I keep getting fliers from all these different people. I don't trust them." I said, "Okay, which payment processor do you use? What's the software you use to run your business?" And he said, "Well, I use this payment processor called 360 Payments. And I'm like, "What if they were to give you a loan, would you then take that?" And he was like, "Yeah, I think if they were to give me a loan, I'm much more likely to take it, because I trust them. I've been running my business on them for years."
And the next thing was I went back to my desk and I found a friend who knew someone from 360 Payments, harassed him until he made the intro, and they became one of the customers as well. And so, there was a aha moment that went off. We had a hypothesis around it, but we're seeing it time and again, and this is yet another anecdote which proved the hypothesis without even having to build the product, that small businesses are much, much more likely to engage in a product, especially one which is related to financing their business, when it comes from a trusted source. And what source is more trustworthy than a tool they already use everyday to run their business?
Turner Novak:
Interesting. And it sounds like for both of those examples you just gave, it's the thing you mentioned earlier, is both of those was collecting data points to bring back to the customer to convince them that this is something they should consider.
Sahill Poddar:
That's right. That's right. And so, I would get on a call, when we got on a call with 360, we basically told them, "Hey, we've spoken to your customer and this is what they told us." And they're much more likely to buy the story, not push back, and want the next part of the conversation. And that remains true until today. A friend of mine was at a med spa recently and he mentioned, "Hey, I'm at this med spa and I think you should talk to the owner here." And so, I got on a call and very quickly got a hang of which vertical SaaS she was using, and then would she take a loan? And then you contact the vertical SaaS and be like, "Hey, I've spoken to this business of yours and they want our product. What's stopping you from launching this?"
I think you have to keep, I wish there was a formulaic way. The reality is, you have to push on 100 different things and keep doing that on a regular basis. And only then can these materialize in meaningful relationships and meaningful partnerships and meaningful revenue for us.
Turner Novak:
Do you have a favorite founder or CEO or company that you've learned from throughout history?
Sahill Poddar:
Yeah, I would say it's probably Vlad at Robinhood.
Turner Novak:
Really? Okay. What's the biggest takeaway from that?
Sahill Poddar:
I think the biggest takeaway is that Robinhood has gone through a very difficult time and having IPO at 38 bucks, stock popped at 80 and then felt like six or $7, I forget what. There was GME, there was a lot of negative media, negative press about it, lack of trust. And they've been able to turn that around into massive amounts of revenue, a huge product service area, re-earning trust, and doing so in a very short amount of time in financial services, I think is practically unprecedented. And what I've learned from him is just perseverance. He's not someone who gives up easily. He just keeps pushing and pushing. And I think doing that with strong first principles and having hope and enthusiasm about the future is the way to do it. And I think realizing that, hey, while trust, once you lose it, it's pretty bad. If you put one foot in front of the other and you do that on a daily basis, you can actually win it back if you're doing it right.
Turner Novak:
Yeah, it's probably an incredible way to end it. That was a good last talking point. This was a lot of fun.
Sahill Poddar:
Likewise.
Stream the full episode on YouTube, Spotify, or Apple.
Find transcripts of all other episodes here.
Loved this!