Scaling a Resale Marketplace — And an Entire Industry — with Ryan Moser
An industrial engineer by trade, Ryan is a maker at heart. This is apropos, given that he is now the VP of Revenue at ThredUp, a recommerce marketplace that’s been around for 15 years. The company was built on the backs of sustainability-oriented Silicon Valley makers on a mission to solve the retail-waste problem.
When Ryan joined ThredUP in 2013, the company had just transitioned to a managed marketplace from a peer-to-peer marketplace model, where it facilitated used-clothing sales between consumers. Under the managed marketplace model, consumers now send used clothing items to ThredUP and receive a payout when eligible items are sold. Unlike many of its recommerce marketplace peers, ThredUP is responsible for processing items, warehousing inventory, managing product listings, and fulfilling orders.
Today, ThredUP has established itself as a preeminent player in the recommerce market. Ryan and his team have redistributed nearly 100 million unique garments and enticed millions of consumers to sell previously loved clothes on ThredUP’s online marketplace.
The rise of sustainable commerce
Back in 2013, fast fashion was getting faster, propelled by the growth of ecommerce. Even at this early juncture, Ryan saw that the pace of retail was becoming unsustainable — for the environment and consumers alike.
“Clothing is such a polluting industry. Recommerce just seemed like the direction society had to go in. In 2013, the question was, is this the right time or is it too early? Our bet wasn't just that society would change. It was that ThredUp would help create the wave; that companies doing recommerce would help build awareness.”
The bet paid off. Today, recommerce is surging. ThredUp’s 2023 Resale Report predicts that by 2027, the global secondhand market will nearly double, reaching $350 billion. While factors like inflation, environmental impact, and a distaste for disposable fashion have contributed to the demand for secondhand goods, Ryan believes that consumers have latched on to recommerce because innovative secondhand companies like ThredUp have earned their trust.
“The growth of recommerce isn't just that attitudes have shifted passively. Companies in the market have proved that buyers can trust recommerce.”
Authenticating second-hand supply
In the early days of recommerce, consumers were understandably suspicious of buying secondhand garments online. The prospect of purchasing used clothing sight unseen, smell unsmelled came with too much risk. ThredUp knew it had to combat consumer hesitations by establishing marketplace trust with a highly managed product catalog that delivered on its promises. Mismatched product-consumer expectations would not only compromise a buyer’s trust but deride the company’s reputation and chip away at its bigger vision: to get the whole world thinking secondhand first.
ThredUp’s managed marketplace approach was unique: most recommerce competitors at this time used a peer-to-peer approach to thrifting, which put the heavy lifting of sales on the sellers and raised questions of trust for buyers.
Still to this day, a large part of ThredUp’s value to buyers is that it vets every incoming garment and accurately markets it, ensuring buyers get what they pay for. The logistical challenge of managing ThredUp’s inventory cannot be understated: ThredUp has around 20 million pieces of clothing live at any given time—and it processes far more secondhand items than it actually sells.
To build consumer trust, the vital question for Ryan was: “How can we make sure that no matter where a garment is processed, no matter which shift, or who's doing it, that it’s as close to consistent as possible?” The answer came as a combination of hands-on inspections and automation at the logistics level.
Hands-on inspections
When garments enter a ThredUp warehouse — which Ryan describes as a football field-sized dry cleaner — each item is hand inspected. To eliminate variance and subjectivity, ThredUp’s warehousing team conducts a 12-point inspection and uses a standardized system to grade all incoming pieces of clothing.
“Our goal is to drive out as much variance from the system as possible. Whether it’s the variance between shifts, between items graded top quality or lower quality, or what is considered a critical flaw, that knowledge is spread out through all distribution centers.”
Automation
For product exploration and fulfillment reasons, each item must be logged as a single SKU. No pre-built ERP could manage ThredUp’s volume, so the company built out its ERP to manage tens of millions of SKUs and accommodate its attribute requirements.
It also implemented storage automation at its warehouses. After manually inspecting items, clothing is automatically organized based on an assigned grade. On the outbound side, ThredUp uses automation to ensure the correct items are being picked, packed, and shipped.
Striking a balance between supply and demand
Supply constraint, where stock is less than demand, is a common challenge for many marketplaces. Multi-vendor marketplaces must source a large body of quality suppliers to keep a healthy inventory that meets consumer demand. Initially, ThredUp worked around supply constraints by acting as a clean-out service. “We took the load off you. As a supplier, you just had to throw your stuff in a box. It’s out of sight, out of mind, and some money shows up later. That’s our value prop.”
When ThredUp adopted the managed marketplace model and began accepting everything that came in, the tides of its economics shifted, and supply surplus became its primary challenge. Ryan acknowledges, “You can't necessarily find a home for every single thing coming in because there isn't a perfect balance between the people who want the supply and target customers whom we are selling to.”
There might never be a perfect balance, but there are ways to mitigate the supply on hand. Ryan laid out the three-tiered funnel ThredUp uses to manage its inventory.
Top of the funnel: Supplier
ThredUp uses supplier incentives and disincentives to control supply before garments enter warehouses.
“To balance the marketplace, we offer incentives for some suppliers to send us more. Other suppliers, we charge fees, so they send us less.”
Mid funnel: Products
ThredUp has boiled down selling criteria to the smallest amount of data needed to determine whether they should keep an item or discard it.
Bottom of the funnel: Pricing
ThredUp revs up demand by meticulously pricing products in a way that gets surplus items out the distribution center door.
Managing a massive master catalog on a resale marketplace
Building a marketplace product catalog is part art, part science. Products must be easily searched, precisely categorized, and accurately attributed. The goal is to connect buyers to products of interest as efficiently as possible. In ThredUp’s case, this means pinpointing a small clothing segment from a universe of 20 million SKUs.
“A lot of the art of cataloging is creating narrow groups. If you're grouping things, size, brand, and style will be most meaningful to your customer.”
The ideal state for ThredUp is an intelligent search function that can connect buyers to items with a simple search bar inquiry. The company is working towards this advanced level of search functionality in the next year. As Ryan says,
“Anything you’re thinking of, we probably have it. The question is, can we use machine learning to surface it?”
Finding success through a mission
When asked what makes ThredUp a successful marketplace, Ryan credits many of the company’s achievements to its mission: to inspire the world to think secondhand first.
While it’s not necessarily tangible and you can’t measure it, it matters. Having the right people on board and bought in plays a huge role in the success of a marketplace.
Ryan explains:
“A lot of people have been there a long time and care deeply. And I think without having a clear sense of purpose, it’s really hard to do that [...] Do you have talented people who are bought in? We have a ton of those people. So I also think that’s a major part of it.”
Episode links:
Learn more about ThredUp:
Transcript
Ryan Moser: [00:00:00] ThredUp is a business that has a mission. And I think that's also part of a success. It's like, it's not, it's not tangible. It's not something you're working on, but I also credit a lot of threat of success to the mission. A lot of people have been there a long time and care deeply. And I think without having a clear sense of purpose, I think it's really hard to do that.
You know, as far as like recruiting talent, like a lot of people who want to work on threat of submission. So, you know, a lot of part of success is like, do you have the right people? Do you have talented people who are bought in? We have a ton of those people. So I also think that's like a major, major part of it.
Niklas Halusa: Welcome to Operation Marketplace. A series where we talk to the people behind marketplaces. We dive into their mission, how they found success and talk about the hurdles that they faced along the way. I'm your host, Niklas Salusa, co founder and CEO of Nautical Commerce.
Welcome to Operation Market. I am your host, Niklas Halusa. Today we have on the pod, uh, Ryan Moser, who has seen quite the development in the marketplace world in his career. He's been at ThredUP [00:01:00] since 2013, if I remember correctly, and, uh, has seen ThredUP go from, you know, an unknown little company in the Bay Area.
To a, you know, monster public company that is today through all the different iterations of that business. He's seen all different parts of the business in a strategy side. He's been on the inventory side. He has been on the software engineering side. So as an incredibly comprehensive view and understanding of, you know, what the business both entails, but also how the business got to where it is today and what the decision points were to make it as successful as a business.
As it is, and also, of course, understands, uh, a little bit of how to scale businesses in general, which is hopefully something that we also can spend some time on. He is a Bay Area or Oakland native. Uh, deeply embedded in the Silicon Valley world that said, Ryan would love for you to introduce yourself and tell us a little bit about your background, your story.
And then, of course, also what [00:02:00] ThredUp does.
Ryan Moser: So, you know, I grew up in Pittsburgh, Pennsylvania, went to Georgia Tech, was a mechanical engineering undergrad and went to grad school for industrial engineering. I worked in a satellite command and control. Out of school, later worked on a launch that was too slow paced.
So I knew that was not the direction I wanted to take in my career. I went into, uh, went into consulting after that incredibly value experience, breaking down some, you know, just being able to break down problems. Um, and I, I taught myself how to code while I was consulting. And through that experience, like I realized I was a maker and that consulting was interesting, but that wasn't really where I wanted to be.
So 2012, uh, you know, my wife wasn't happy with her job as a teacher. I wasn't happy with my job as a consultant. So, you know, I said, uh, you know, Hey, how about we both quit, move to Silicon Valley? It's a place for makers. Uh, she said yes, and we did. And, uh, ThredUP was where I landed in 2013 after we made that move.
So ThredUP actually started as a peer to peer business, but it transitioned to, uh, transitioned to, uh, a managed marketplace, uh, just prior to my, to my, uh, joining. And what I mean by [00:03:00] like a managed marketplace is, you know, there are a lot of like peer to peer marketplaces, think like Poshmark or even eBay, where, um, you buy it from me and I ship it directly to you.
Whereas, uh, if you're a managed marketplace, you're sitting in the middle and, uh, and you can add a lot of value by sitting in the middle. Whereas we hold the inventory, we take all the photos, we add all the data. So all that work is what ThredUP does. So ThredUP is one of the largest. Online secondhand clothing, uh, marketplaces in the world.
Uh, and you know, at the time it was smaller, but, but, but fundamentally it was still a marketplace. How big
is large, the largest, what, what, what kind of, how much do you move?
We have processed over 170 million, uh, unique SKUs to this point. So it's something we come out and that was last year's number. So this year I suspect to be closer to 200 million.
I don't have the exact number yet, but it's, it's a large, it's a large number of, uh, unique SKUs.
Niklas Halusa: Give us a feeling of where it started, right? Cause some of the growth story is, is, is wild here, right? You know, how do you go from. Okay. We just became a managed marketplace in 2013. I don't know how much you guys were moving at the time to moving that many [00:04:00] SKUs.
You know, what did you have, what did you guys have to do and put in place to get to that kind of a scale?
Ryan Moser: This is a great thing for me to talk to you because I actually joined the opposite side of the business, uh, in 2013 when I, uh, when I first started ThredUP and that was the, that was our biggest challenge at the time.
Like how do we scale operations when everything is single SKU? So, you know, we had a single warehouse in the East Bay, and it was really about trying to figure out how to scale these things up. And, you know, there are a number of problems that we had to solve for. Let's even say you can get the data, how do you store it?
So, like, you can't buy like an ERP system off the shelf to store millions of SKUs. You talk to a vendor, maybe you can store tens of thousands of SKUs. If you want to tell them you want to store tens of millions, they have no idea what you're talking about But it's use case just does not exist or or at least at the time Uh that you know There was no such use case So part of it was building out all of our own erp to even store the data and then within the dcs Which is where I worked, you know, it's worked at the first the first dc in the east bay here in california It was How do we take, how do we take [00:05:00] photographs at scale?
You know, how do you take 10, 000 photographs a day? Uh, you know, 100, 000 photographs a day. Processing the backgrounds are consistent. And then what sort of systems at that point? You know, there's very little, you know, like machine learning was not, uh, was not at the point where you could really use it. So it was, how do we enter all that data?
What sort of systems do people need? What are the processes people need to grade clothing to enter that data? So like, so those were all the sorts of problems we were trying to solve in 2013. Like, how did we scale this thing? Like when I first joined ThredUp, like actually the thing that drew me to the business was I went into that warehouse and there was something called a Z Rack.
You know, it's literally shaped like a Z. It's basically, you can put a bunch of hangers on, like wheel it around. And it was literally just floor space full of Z Racks, you know, picking, using like a grease pencil and a sheet of paper. And I was like, Oh, like this, there's a lot of work to do here. So I was like, we're going to have some really interesting challenges.
Like, I cannot wait to get started. Uh, on this, which is actually the reason I chose, one of the reasons I chose thredUP. Not, not the only reason, but a reason.
Niklas Halusa: You know, you were at z racks, where are you now? What does a fulfillment photo [00:06:00] shoot, you name it, location, look like for ThredUP now?
Ryan Moser: If you were to see it now, it would be almost impossible to draw a line, even in your mind, between how did we get from this thing to where we are today.
So today we have some of the largest, maybe the largest garment on hanger storage in the world. Uh, today. So, you know, you've seen like a dry cleaner where they have garment on hanger storage, like imagine like two football pitches side by side, like four stories tall, that's the sort of scale we're doing storage.
And it's just, and the amount of automation is quite amazing from the start of the DC all the way through where we have people, um, uh, who. Are physically like inspecting the closing, right? That's not something you can automate. You've got people inspecting the clothing, looking at the tags. It goes on a conveyor, it goes to a studio where, you know, it grades.
We separate out the clothing we can keep from the clothing that we cannot keep. It goes to the correct type of studio. We've got many types of studios in the DCs. Mannequin studios, studios for shoes, accessories, flat studios. Goes to the [00:07:00] right studio, and we actually have conveyors, uh, you know, conveyors taking, um, hanger conveyors taking, uh, garments from one point to another.
Uh, once they finish the inbound process, they're inducted into this, uh, massive carousel that we have. And then there's a lot of automation on the outbound side around, uh, picking, packing, shipping. So, you know, you wanna pick by the time you walk to conveyor, the, the conveyor swung around. The right garment is right there, put it on a conveyor, taken to outbound, and then, uh, it all, it all goes from there.
Niklas Halusa: And so when you talk about grading and. This is a part that I'm super interested in because this is a skill set that thread up among others, but I'm sure you guys have built this skill set in house that few companies have to build and few companies have to build at that scale. And I'm willing to bet has a huge amount of hard learnings on the way.
about where to draw the line of what's okay and what's not okay. And [00:08:00] there's like a deep, uh, body of tribal knowledge that you guys have built around that. I would expect. Can you talk a little bit more about, you know, what is, what is a human doing? What is this grading? You know, how do you know if something is, you know, makes it through the grading, makes it through the, through the studio correctly, you know, there's got to be a whole process around that.
Ryan Moser: Uh, there, there is. And that process is, those processes and those systems have changed greatly over the course of my time at ThredUP because the, you know, our goal is to drive as much variance out of the system as possible. Right. Because if you think about the experience as a buyer. Like one of the problems we're trying to overcome here is trust.
Like with, with secondhand, you know, there are some people who, who thrift, but we're, we're trying to, you know, we're trying to inspire, you know, trying to inspire the world to think secondhand first. A lot of people do not think secondhand first. So how do you get to that point? You have to build trust. And that starts with quality.
Like can the customers trust like the way that you're grading and what shows up at the door when you place a place in order, like the big [00:09:00] challenge comes back to what you said is. How do you create a set of criteria and a set of processes where we have, uh, four places where we process clothing right now, right?
We've got, um, Arizona, we've got one in Texas, a place in Georgia, and a place in, uh, and a D. C. in Pennsylvania. How do you make sure that no matter where it's processed, no matter which shift, no matter who's doing it, it is as close to consistent as possible? So, like, um, it isn't just averages, it's like variance is something that we think about a lot, where it's the variance of, variance between shifts of, Um, is the, you know, the same percent of items being graded top quality, like lower quality and like what sort of, what sort of a flaws can people spot and like, and also from the customer perspective, like what are the critical flaws?
to spots. That's something where we've done a ton of learning over the years. Like you said, tribal knowledge and like you never want tribal knowledge to be localized to a D. C. Where they figured something out and they're doing something different, even if it's, you know, even if it's better, like you want that knowledge to spread throughout all of your, uh, deceased.
Niklas Halusa: When it comes to building that system, [00:10:00] then to your point, you need to standardize it. You need to, you know, make this a company practice. One of the things that we talk a lot about is Is especially when we, when we onboard marketplaces is that it's the trifecta of people process and technology that makes a marketplace successful.
And many people may be just where the technology hat or just where the process hat or just where the people have, depending on what predispositions that they. You know, come, come with and, and, and the industries that they come from, when you look at a problem like that, especially given that you are a software engineer, how much, how would you break it out?
How much of it is people, how much process and how much can you realistically automate?
Ryan Moser: When you do something like that, one of the critical parts of automating is that you figured out you figured out the right thing to do before you automate. Like, once you automate, you've kind of locked into what you're doing.
You have to have a very clear picture of what the right thing to be doing is, and it's right for the long [00:11:00] term because it costs a lot of money to automate. You know, part of the reason for that race so much capital is, you know, building out these, uh, you know, hangar conveyors and, um, you know, I call these systems is you.
Tons of CapEx, right? And it's just critical to make sure that you're spending the CapEx in ways that make sense that it's going to pay off for years and years and years. And then going back to like, what is that balance? No one person has the expertise to do all of those things. We're getting the process persons, right?
Like the, uh, the interface that people are using, building, building the software systems on the, uh, the backend. So, you know, a lot of that work in the early days was just working, working across all those functions in the DCs, just understanding feedback from the operators. Feedback from the trainers and, you know, even thinking about onboarding people, like it isn't just how well does the system perform terminally?
Like how fast can you onboard? How fast can people learn? All right. Like, like even thinking about it in like those terms as you're scaling.
Niklas Halusa: That makes sense. Speaking of also building out the systems, [00:12:00] how much of, of thread up at this point is things that are off the shelf. You mentioned earlier that you had to build your own RRP.
It's a hell of an exercise. And so how much of what thread up is what runs thread up today is, is off the shelf versus built.
Ryan Moser: In the DCs, a lot of it is custom. I'd actually say that there's some things that are off the shelf. Like we're experimenting a bit with like garment folding, like there are people who make machines to do garment folding.
Like a lot of the Conveyance and storage is still custom and I would even say it's getting more custom because the systems to operate with our sort of duty cycles and scale didn't exist, you know, like a lot of those just like weren't out there to be to be purchased. So a lot of a lot of those are still custom and we're still even building like custom hardware in the software side.
I actually think the opportunities to do things that are, I'd say, you're combining things that are custom for our business and things that are off the shelf. You know, obviously, most e commerce stores, you don't have to figure out how to wade through millions of items. So there is some thinking [00:13:00] around what can we do there and like, like in the state of like AI is getting so much better that like, I think that's going to have a heavy influence on, you know, how we think about search and discovery in the shop going forward.
And I think that, like, the state of off the shelf tools is getting so much better that we will be able to use more like off the shelf models, but we'll still have to blend it with our own systems. Cause again, we still have this problem of millions and millions of SKUs live at like, even like right now, they're like SKUs live on the site right now.
And that's just like not a common problem to try to solve. Yeah.
Niklas Halusa: Especially not that scale. I think one of the stories that I once heard from, um, GAP, who I believe you guys work with, right. Um, is, uh, I think it was their CIO who said that, you know, GAP may only be doing, I think five or 6 billion in revenue, which may not sound a lot until you realize that they do five or 6 billion in revenue.
You know, 10 at a time, which means that the order volume is just spectacular. And that a lot of systems simply fall apart on [00:14:00] that level of scale, that number of orders. Um, and then one of the things that, that you've come back to over and over again is not only the amount of SKUs live right now, you said three and a half million, uh, and then also the number of SKUs you're doing a year. I think you said somewhere near 200.
Ryan Moser: It's closer to 20. So it's like, so it's like, you know, you look in 200, you're looking at the past 10, you know. Roughly 200 past 10 years. So like, it's like this year, the number we'll list will be 20 number. We'll see, we'll be much more because we don't process everything.
You know, we can't list everything that we see. So we process a lot more than we actually list.
Niklas Halusa: What's the variation of those 20 million unique items.
Ryan Moser: That's part of the trick, is figuring out which ones are unique and which are not. That's actually, that's actually part of my job. Yes, they're all processed as unique SKUs because we know nothing when those items come in.
Like, you ship us a bag, you ship us a box. We just tear open the bag and it's like, oh, pair of jeans, three, three tops. It's like, we know nothing when those items show up. So yeah, it's like, so we treat them all as single SKU. Then the back end, we are doing work to try to figure out, as you can imagine, this is [00:15:00] very useful in terms of.
How you present, present it to customers, how you price things, how you mark things down, how you sell them. Like, it's really interesting to know if we've seen, seen those things, uh, prior. But like, once you have some data and you have a photo, you can start to match items where, okay, like based on photos, like, and based on some of the item data we have, we know we've seen this exactly before, but like, that isn't something, you know, when the item shows up, you only, that's something you only know later.
Niklas Halusa: So actually, I didn't know that. So I've never, I have to admit, I've never used thread up. Is that really what you guys get in terms of an inbound? People will just throw everything in a box. And what do you, what do you demand of your sellers? Of the people shipping the things to identify the stuff.
Ryan Moser: So selling is like maybe the most critical part of our value prop.
Because if you think about marketplaces, what's the constraint in most marketplaces? They are supply constrained. How did ThredUP initially work around that supply constraint and get, get all the supply? It was through the clean out service. So taking all the load off of you as a supplier to just throw your stuff in a box and you just [00:16:00] don't think about it, it's just out of sight, out of mind, and like some like money shows up later, that's quarter our value problems.
Like nobody else does that. If you want to list on eBay or list on Poshmark, you've got to do the work. For us, you don't have to do anything. But then it creates these problems, as you can imagine, where we don't know what's going to show up and you've got to balance supply and demand in a marketplace.
You've got to find a buyer for every item that comes in, right? Otherwise, you're wasting money processing items that you can't sell. So like, how do you solve that problem? And, you know, that's something that we've really thought a lot more about solving over the past year. Where, in the early days, it was, send us your stuff, we will find a home for everything.
And we've realized that you can't There is a limit to that. You can't necessarily, uh, find a home for every single, single thing coming in because there isn't a perfect balance between the people who want the supply and target customers whom we are selling to.
Niklas Halusa: And nobody's really buying Shein.
Ryan Moser:
Ding, ding, ding, yeah. So, and just, and just think about like how much, how much latent value is there in a she in anything? Kind of like nothing, right? You, it's worn twice, it's falling apart, right? So there isn't a whole lot of like latent value there. Yeah. You know, if [00:17:00] we think about this as a funnel, right, top of funnel is who are your suppliers?
And like, what are they sending you? So initially, it was anybody could send anything, whatever they want. Ideally, we'd like to get back to that point, but we do have some levers to encourage, like we have a segments of suppliers and something to think about. The top of the funnel is who is sending us what?
So some suppliers, we can offer incentives for us to send more other suppliers. We can charge fees. So they send us. Less. So that's, that's one mechanism to try to balance the marketplace. Very top of funnel. Next level of funnel down is what actually shows up in the DC because you will never get a perfect, you know, in one of our, one of our warehouses because you, you will never get a perfect balance, right?
Because we can't, we literally can't order what we want and we have to carry the inventories we're taking on risk. How do we do risk it once things shows up? And a lot of the thinking there is like, what is the least data you need to know if you want to keep this item or not? There's this constant tension between every piece of data costs you something and you learn something.
So it's just like, how quickly can you learn? Is this something that makes sense for a marketplace that we want to list? Or is it something that we do not want to list? So that's the [00:18:00] sort of funnel of things coming in where we can try to balance the marketplace. And the final, the final stage of course is like, Using pricing to balance supply and demand.
Niklas Halusa:
Got it. That makes sense. And so, there's a decent chunk there of having a database of things that you know about. Because I expect you then don't, you can't say, okay, GAP sweaters sell well, if you don't know that it's a GAP sweater where every single thing is done ad hoc. What does that look like in terms of?
You know, how did you build a master catalog or how do you build the master catalog that all of these new clothes fit into and what do you have to know about that master catalog to make it efficient, to collect the right data and make the right analysis on, on stuff that you guys get in the warehouse.
Ryan Moser:
Yeah, that's a great, that's a great question. So like a lot of, this is where like a lot of the art comes into the business where like, what, like what are the critical pieces of data, what are the most critical pieces of data? [00:19:00] And you have to somehow cluster. The inventory coming in, because nobody can make sense of 200 million unique SKUs, right?
Like you can't even think of like, it's not even possible to think about it. So a lot of the, a lot of the art and frankly, a lot of the tension in the system is you want narrow groups because narrow groups are more meaningful to customers, right? If you're grouping things by style, you know, size and brand and style, that's going to be the most meaningful to your, to your customer.
But if the groups are too narrow, you can't make sense out of what you basically can't make sense of the data, right? There's too much data for a person to make sense of.
Niklas Halusa: So when you say narrow, just to confirm that you're saying You know, you can't make a, make a group of, you know, show color, three roll, two cardigans, both the warehouse workers and the customer just don't understand maybe the technical terms of what exactly you're selling.
So you have to make some ambiguous decisions about what the big categories are. Is that the right way of understanding it?
Ryan Moser: Yeah. And it's also like. If you scope it, if you, if you scope your SKUs too narrowly, like let's, let's do the really extreme version here. We treat everything as its [00:20:00] own unique snowflake.
It's truly single SKU. How can you understand if something is going to be, you know, um, understand if something, like supply and demand, when you have one of something, you'll see it once ever. Like you can't, that's not a, that's actually not a solvable problem. So you have to somehow group things in a way that's meaningful, where you see enough of those things, where you can predict.
How many of those things are going to come in and all those bags and boxes showing up in your DCs, right? If you group things, even though you don't know what's going to come in at any given bag, if, if you tell me 10, 000 bags, I can come pretty close with how many pairs of jeans will be in those 10, 000 bags, how many crop tops, how many skirts, right?
So it's sort of like thinking at that level and like, what is the right way to cluster with those things like brand brand category size tend to be like the key things. As far as like building those clusters, but, but like, but that, but again, like, I think a lot of the art of like building those clusters at the right level,
Niklas Halusa: just thinking this through, you guys have a wild data set now on consumer preferences, you know, the [00:21:00] structured stuff.
Okay. But you can get that off the internet, right? Size, color, brand. Right. But then you also know, you know, what does a customer care about in terms of level of quality, you name it. But you also know, you know, what at any one point, what are people getting rid of and what are people trying to get their hands on?
You know, the supply and demand effectively of fashion trends of brand trends, all these kinds of things. Uh, first of all, did I understand that correctly? Second of all, how much of this is now an, a machine learning problem?
Ryan Moser: It is increasingly like a machine learning problem. So if you think about like just the search problem or like discovery on our site, like the ideal state of nature that I think we're working to this year is like, imagine you can dream up anything.
Describing clothing and just type it into a search bar. If we can get that to work, like we probably have it, right? Like it does, it does like anything you're [00:22:00] thinking of, we probably have it. The question is, can we use machine learning to surface it? Right? And even in terms of like, even in terms of clustering, a lot of that is like machine learning based where, you know, you have all this data you can store in a database, but, um, all it's really hard to describe in words, the style of a garment, but you can get that from the photo, right?
Which is purely, which is a very much like a machine learning problem where you're trying to match. You know, it isn't just customers preferences for brand, it's like, it's actually their preference for style. And, and, and, you know, that's actually like an amazing area for like, uh, for ML, for, for, for us.
Niklas Halusa: I remember a friend of mine back when I was in college tried to write a machine learning algorithm to recognize melody, I believe, in music.
Because melody, harmony, and rhythm are all different things. And I see the guitar in your background here. Most music that will recommendation engines For the [00:23:00] Spotify's of the world, from my understanding is largely, you know, what's the BPM and what's the genre, right? And then there's some aspect of people who listen to this also listen to that.
Right. But you're really trying to get into more abstract understanding of styles and objects and wants that mean that two things from two completely different brands that in principle should not be, you know, are not usually categorized together, might be in the buying habits of the same person. Now, for me, I buy, you know, on the one hand, I buy Uniqlo and on the, on the other hand, I buy like custom shirts and in principle, no obvious algorithm would put custom shirts and Uniqlo in the same, in the same bucket.
But actually as a consumer, I'm more faulty facet in that. And so if I understood that correctly for what you were saying, this is kind of where you're trying to go is start to understand the psychographic profile of the buyer, uh, in a much less predefined or pre structured way than, you know, typical sorting categorization methods might do it.
Ryan Moser: [00:24:00]
It's a critical problem for us to solve. Like it's kind of unique to running a big marketplace where, you know, if you're the, if you're the gap, people are coming to you because you fit their style already. Where if you come to our marketplace, we've got millions of items making great use of your time is a really important thing for us.
So like actually being able to neck down to it. What are the styles and brands and price points you care about is a critical thing. Just be able to navigate the sheer volume of items for sale.
Niklas Halusa: Now, I think we've covered a decent amount on the operating side and on the, on the center side. On the buyer side, what are the, what are the channels that thread up sales through?
Ryan Moser: I mean, all of, all of our revenue is a DTC. It's almost all like direct, you know, selling online, all direct.
Niklas Halusa: I read a little about retailer partnerships. Uh, that thread up is starting to ink is that then only on the supply side, or are there also retail partnerships on the demand side?
Ryan Moser: Oh, no. They're also, they're also, they're also partnerships on the demand side.
So the [00:25:00] bulk of our revenue is still what we call the CMP, the core marketplace, which is again, this two sided marketplace, but then a growing, a growing part of our business is this RAS, uh, retail as a service. Um, is a growing part of our business and that can range, you know, and we have a range of offerings there.
Everything from a supply side, if you want to, you know, help your customers clean out their closets, we can come in there. Everything through like a, everything through a white label shop, where if you want to have your own secondhand shop, we can also do that for you.
Niklas Halusa: Got it. And how, how new is that model for ThredUp?
Ryan Moser: Uh, we've been doing this for at least a few years now. Secondhand is the fastest growing segment in apparel. And it's just grown globally across many different, uh, you know, it isn't, it isn't grown just with an apparel. First, it's just a growing segment. And, um, if you look at younger customers, they have a great interest in this segment.
So if you're a brand and you want to get in on a rapid, you know, the most rapidly growing segments, uh, uh, within apparel, how do you do it? You can go it alone, but it turns out that doing a recommerce is really, really hard or you can come to somebody who is. [00:26:00] Who, who, who does it? And it, you know, and it isn't just figuring out on your own, if you want to, if you want to second hand products to sell for your brand, chances are we already have thousands or tens of thousands or more of those garments to sell.
So you can just. Get up and running very quickly.
Niklas Halusa: And, and why is it growing so quickly? What's the, what's the driver here?
Ryan Moser: I'm not entirely sure. I mean, frankly, like people's attitudes are changing. So if you look like generational shifts, younger generations tend to have a much, much more favorable view of, of secondhand.
I also think that the company is getting into the space. This is sort of like a chicken and egg thing, right? Where like more companies are getting involved. Customers become more aware. You start to build more trust. You know, so if I go back to like the early, you know, like what I was thinking when I joined ThredUP.
Back in 2013 was 2013 was very, very early for this, right? Like very early for secondhand anything really. And like, what was the fundamental bet? I was thinking like, this is where society is going just from the sheer waste, right? Just in terms of like taking care of the planet. Attitudes are shifting there.
And clothing is such a polluting industry. That just [00:27:00] seems like this is direction society has to go in. And the question in 2013 was, is this the right time? Or is it just, is it too early for that? But like part of the bet wasn't just that society would change. Threat up would ride the wave. Part of the bet, even at the time was straight up would actually help create the way that like that these companies coming on doing e commerce would actually help build awareness, build trust.
So I actually think that's part of it isn't just that attitudes are shifting passively. I think some of that would have happened. I think it's actually companies in the marketplace, proving that you can do it, proving that you can trust. Secondhand is also a driver.
Niklas Halusa: Where do you think it's going from here?
So it's growing like crazy. I know, I know that one of the stats out there is that it's going to go to, I think, 350 billion e commerce by, by, by 2027. But you've seen enough here and you have a background also, um, in the strategy side of things, you know, what else do you see down the line for, for ThredUP or ecommerce?
Ryan Moser:
There's still a chasm to cross in [00:28:00] terms of, in terms of e commerce going truly, truly mainstream. So, you know, like what is, what is our goal as a company to like, we want the world to think. We, we, we want to inspire the world to think secondhand first. And we aren't there yet, but that's, that's a place that we want to get to.
I'm doing the right thing for the planet. I'm being stylish, right? I'm doing great things for my budget. So we're still not at a point where like the first thing people think of is thread up. That's a point where that we would eventually like to get to where like it hasn't gone where like reconverse is not going to fully mainstream, but I think over the coming decade, I think that it will.
Niklas Halusa: Got it. That makes sense. And what are you tasked with? Now with thredUP to go and, you know, make some of these things a reality.
Ryan Moser: As a VP of revenue strategy, I think a lot about the inventory, you know, getting pricing promotions, the inventory right on the site, and then working with partners on like the marketing side of like, how do we, we're going through a big revamp of how do we onboard customers, right?
And it's who is the right core customer to attract, how do we onboard those [00:29:00] customers, how do we make sure that we're building trust along that journey? Um, and you know, we're improving the experience of using thredUPs. Website, but like really like my role in that is getting the pricing inventory side, right?
So customers feel like they're getting great value and then again, working on the overhaul of the onboarding process. So customers have a great experience being onboarded to thread up.
Niklas Halusa: And how did you end up with that span? Are you a particularly organized person?
Ryan Moser: You know, over 11 years of a start, you know, going from a startup to a public company and continuing to grow a ton of things change.
And my, my mindset is like, where is the greatest value to be generated in the business? And what can I do to help make that possible? So I, you know, I've always had this attitude of like looking out, like, where are the business, where's the business going? What are the gaps that need to be filled? And how, how am I going to be the person to fill those gaps?
That's how I've ended up in so many different roles within ThredUP and. Much to ThredUp's credit, they're totally, they're totally great with that, right? Some companies want you to come in and do a certain thing. Like you're an engineer, you're a salesperson, you're whatever. [00:30:00] And one of the amazing things about ThredUp is if you, if there's an area where you think you can add value to the business, you can prove it out.
They will, they will, they are happy for you to do it. And you know, it isn't just me. There are people from other parts of the company who have, you know, people from the Infra team. But they're interested in, uh, you know, uh, one of their hobbies is taking photographs and somebody comes and says, Hey, I've got a different idea for the way we can set up our cameras.
But we've had a lot of those things where if you, if you have an idea, uh, that can, uh, improve the business, you will have a chance to do it. So, you know, it's like, that's the reason I'm still here having joined in 2013 and that's how I've ended up in, um. You know, that's why I'm on my, his sixth role, having done all this, all those things.
Niklas Halusa: And when you started, did you know, first of all, that ThredUP was going to be such a success? And did you expect that you'd be here for a decade now?
Ryan Moser: I don't think you can plan things out that far. Like I would have never planned this. Like if I were to be talking to myself a decade ago and tell him, Hey, your VP of revenue [00:31:00] strategy, I would be asking 10 years older of Ryan.
Like how, how, how the heck did that happen? You never know it's going to be a success, but it just felt like an idea. It was like time had come. It's just so like, like that was the bet. When I joined, I think part of the bet you make when you're joining, joining early stage company, at least the mindset you want is.
This thing is going to be a success because I'm going to help make that happen, right? Like you've got the mindset of like there are a bunch of big crazy problems to solve and I'm going to need, I'm going to need the person to solve them. So that's just kind of like the mindset you need going in. It's like, did I know it would work?
Did I know it would work? Of course not. But um, I have always had the attitude that any problem that's out there to be solved, if it matters, I can, I can figure out how to solve it. And there are other people fed up who are like that. You know, you can't guarantee success, but at least all the pieces are there.
Niklas Halusa: You were a consultant before this. So you have a good understanding of the strategic aspect of it. What did ThredUP do right that made it such a success?
Ryan Moser: At least starting out like the, the [00:32:00] cleanup kit and the supply experience is, was so different, but like that, like getting supply in the early days was such an amazing.
You know, it's just such an amazing thing for like a marketplace to have. I think that's one part, but I'm, I'm sort of getting beyond like the mechanics of the, you know, getting down the mechanics. Like threat is a business that has a mission. And I think that's also part of the success. It's like, it's not, it's not tangible.
It's not something you're working on, but you know, if I, I also credit a lot of threat of success to the mission, a lot of people have been there a long time and care deeply. And I think without having a clear sense of purpose, I think it's really hard to do that. So I wrote this sort of like a more like abstract.
Concept that it's hard to measure, but it's something that just feels like it matters. Like the people who like, uh, you know, as far as like recruiting talent, like a lot of people who want to work on threat submission. So, you know, a lot of part of success is like, do you have the right people? Do you have talented people who are bought in?
We have a ton of those people. So I also think that's like a major, major part of it.
Niklas Halusa:And then finally, if you look back for [00:33:00] your, on your career in marketplaces, would you be able to pinpoint? What the two or three biggest challenges that you're proudest of, of, of, of tackling are
Ryan Moser:. So I can give you two, I mean, I've got a couple of moments burned into my memory.
The one is like when we first automated the warehouses where we didn't have, when we first added conveyors. In to the warehouses because it was people physically moving racks around and we eventually knew that that couldn't scale You can't go up. You know, you can't reach a third level mezzanine you've got to you have to have conveyors and you've got to figure out how to route things somehow and um, I worked on the first first conveying system with uh, some of my colleagues and that was Like, building software is, is great, but there's something that's on a completely another level when you go into a warehouse, you see this massive system, you turn it on, and then [00:34:00] close, you start routing around and moving, and it's something that you built, it's something that you did.
Like, that is a level of magic that, that you cannot capture just doing software. Like, when your software, when you see your software moving things in the real world, that is just a level of magic that's absolutely amazing. Um, so I'd say, like, that is As far as like a proud moment, a moment that's like burned in my memory, like, like literally the people hitting the on switch and then seeing it work, that was, that was incredible.
It's like something I will never, ever forget. And then like other challenges to overcome, like I've spent a lot of time working in the pricing space and just, and just some of the challenges with pricing a single skew and even how to think about those problems and frame those problems and just comparing across like.
And just figuring out ways like to give a really good price to an item you've never seen before. I'm also really proud of having done that work because like I just don't know if there's any precedent for having done that. Like there's no textbook you can read. There's nothing you can do. It's just a purely creative act.
And um, you know, there's, there's no guidance. It's just hugely daunting. You just [00:35:00] have to have the courage to just jump in and Jump in and do it. And yes, it's also something that I'm very proud of.
Niklas Halusa: Yeah. And something that we see a lot in our customers as well is this pricing question gets really, really hard, especially, you know, the constraints, not just what will people buy something for.
But also, you know, what is going to be sufficient pricing for the, for the suppliers to actually bother giving you all this stuff, right? And trying to tread a fine line there and, you know, there's always a mismatch. You always think that your thing is going to be worth more to the new buyer than it really is.
And the amount of forces and constraints on this problem have got to be a hugely interesting, effectively, data science problem.
Ryan Moser: Oh, yeah, yeah. Yes, it's absolutely, it's absolutely a data science problem. And it's a fascinating set of problems. Like, if you're running a marketplace, you're always thinking about this balance of, um, You're always dividing value between your buyers, your suppliers and yourself, the person running the business.
And ideally you're doing about, you're doing that and thinking about that in a way of like, how do I grow the pie of, how do I grow the pie? So it isn't just splitting value. How am I creating more value? So there's more value to [00:36:00] our buyers, more value to suppliers, and of course, more value to us throughout.
Yeah.
Niklas Halusa: And parting thoughts. I know that you're active in the marketplace community. What resources would you point somebody to who's now either looking to start a marketplace or looking to go work in a marketplace?
Ryan Moser: Um, I mean, for somebody trying to start a marketplace, so much progress has been made over the past decade plus that the thing I would, uh, exhort people to do is.
Take advantage of the fact that people have done it before. So like there are communities out there for people who have, you know, people who have built marketplaces, people have helped scale up marketplaces. There are people like me who are out there who just love marketplace things and love helping and love talking about marketplaces and helping people scale up marketplaces.
So I think the main piece of advice I'd have for somebody founding a marketplace, creating a new marketplaces, people have done it before. They'd probably be happy to help, help you don't go it alone, reach out. Um, I think that's the main, main, uh, main thing. Like there isn't some general advice. Like it's like every marketplace is like, uh, he, [00:37:00] uh, you know, has some pieces that are going to be unique to that marketplace, but there will be pieces that, that, that somehow rhyme or match things that people have done before.
Niklas Halusa: And would you be open to people reaching out to you to learn? Is there a best place to reach you?
Ryan Moser: People do it all the time. Uh, you can find me on LinkedIn or, um, yeah, that'd probably be the best place. Best place to find me.
Niklas Halusa: Awesome. Ryan, the level of nitty gritty details that you understand and know and have hit on is something that I can see comes from.
And look, your background shows it, uh, really doing the work and being in the weeds and seeing what you do and going through more than just one or two cycles doing, you know, 10 years, you really start to see what impact the stuff that you have has. And, uh, and I really appreciate you bringing the passion on this show.
And, uh, we wish you the best of luck for 2024 and beyond.
Ryan Moser: Thank you very much, Nicholas. It was a pleasure.
Niklas Halusa: Thanks for [00:38:00] tuning in to Operation Marketplace. This show was brought to you by Nautical Commerce, the end to end marketplace platform. If you have any questions about optimizing or starting a marketplace, hit the link in the description.
We hope you enjoyed the show. See you next time.