The End of the Funnel With ThredUp's Dan DeMeyere

Fred Schonenberg
Hello everyone and welcome to the VentureFuel Visionaries podcast. I'm your host, Fred Schonenberg. I'm so excited today to have Dan DeMeyer. He is the Chief Product and Technology Officer at ThredUP. He's a longtime leader at the company. He's helped scale one of the largest online resale platforms that there is. 

I was fortunate enough to be on stage at Shop Talk a couple of months back with Dan. You all know when you get on stage, oftentimes someone's promoting something and they're representing the company. Immediately on stage in our prep calls, I realized Dan was a practitioner, that he had been in the weeds of building what makes ThredUP so special, as well as sort of on the cutting edge of some of the AI advancements that are modernizing it even further. 

So he has this sort of product and technology background and brought that perspective into that conversation. And it was just really dynamic. And so I begged him and pleaded with him to join us on the show today, because after our talk, I posted a few things and everybody is talking about transactional commerce, agentic commerce, AI commerce, brand visibility. It just felt like a really good time to take some of the conversation and bring it here. So without further ado, I know everyone will enjoy this. Dan, welcome to the show.

Dan DeMeyere
Thanks for having me.

Fred Schonenberg
For those that don't know about you, that don't know about ThredUP, could you give a little bit of a background and introduction to both?

Dan DeMeyere
Sure, yeah. So I'm from Michigan originally. I have a degree in computer science engineering and there's not much of a tech scene in Michigan. Going to my Calculus four final, had my Honda Civic Path for my entire life, took the class, took the final exam, drove off to California and I've been working at startups ever since. So I worked at a few startups in LA and then I joined ThredUP as one of the first employees back in 2010. I've been there ever since. 

At that time, we were a peer-to-peer swapping platform for clothes and then transitioned to be a managed marketplace, meaning that this is not just a technology play. We have distribution centers and technology. And if you haven't heard of ThredUP, we send you a bag, you fill up the clothes, we take those clothes, we process it on your behalf and then whatever sells, we give you a cut of that payout. So I spent my first eight years exclusively focused on the engineering side, just building a lot of the startup way, iterating, A-B testing, and big bets. 

And then I took on product management, data science, product design and as well as growth marketing for a while. I was able to expand to have a broader scope and portfolio across our business. And we're gonna sell over 20 million items this year. So we're definitely a high volume marketplace and that brings along a lot of very unique challenges.

Fred Schonenberg
Yeah, no doubt. It's cool to see how you guys have grown as a company. How do you think that your early engineering background and being one of the sort of builders early on, how has that perspective shaped the way you think about now as you're building products that have more of a scale remit to it?

Dan DeMeyere
I think having a technical background in product is really important right now. The rate at which technology is improving, it's just unbelievable. I think everyone knows you just like every day you're reading about like a new model, something, a new piece of tech that was released. And so be able to zoom out and think about, what is that through line? Like where was this technology 12 months ago? Where is it gonna be 12 months from now? I think it's really important to understand how the solution space will evolve.

And so I spent a lot of my time thinking about problems in the business or problems for our customers that historically have been limited in technology and figuring out what is the right time to go after them. And I was actually, someone who has like a really good way of framing this is Kevin Wheel, who was the former CPO at OpenAI. And the way that they thought about it was that if you're working on an important problem, but the way in which you're approaching it is kind of pushing the envelope of what the technology is capable of today, that you just need to keep going because at the rate of technology change, within a month or two, the technology will be there and then the product that you've been working on will sink.

So I think for me, it's kind of using the technology background, trying to understand where things are headed, understanding our customer or business well, and kind of trying to skate to where the puck is going to be as opposed to like where it is today. 

Fred Schonenberg
I think it's a really interesting idea and maybe plays into this question, which is on stage, we were talking and you kind of raised this idea of what if every customer essentially had their own agent or what would it look like if they had their own website, right? And people have talked for years about personalization in various forms, but I think the agentic piece that made it very real to me was this idea of, hey, how does that look if technology can actually enable that? And so I'm curious if you could maybe talk about that a little bit from maybe what sparked that idea and then what do you think that might look like in practice when the technology catches up with the idea? 

Dan DeMeyere
Yeah, about a year ago, I think when people started playing around with the digital technology and like productivity for themselves, it was like, okay, like an engineer has an agent now they have multiple agents and you can put yourself in the sort of the shoes of a product person and say like, okay, well, what if a user had an agent? Like, what would that look like? And I think that there are some people who are trying to decide whether that's like having an assistant to help them shop or a customer service agent to help them resolve an issue. I think that's like one way people think about it.

But the way that we've been thinking about it in addition to those use cases is like, what if you had an agent that was consuming event stream data across every single platform, every channel for a given customer? Like it knows who that customer is it knows like what emails and push notifications will tend to come to the website and when they're on the website, what do they do? It knows all of that and then how, if you could build intelligence around that, like how would it change the way in which you build product experiences? So this is something that we've been working on since the beginning of the years when we launched this. 

Yeah, it's a little bit difficult to wrap your head around, but I think it changes personalization when it comes to like, hey, if you built this website that works for basically like your best average customer and it can work for multiple different demographics. And for us, we have a really massive 5 million item catalog across 35,000 brands. So we've built this like a common denominator experience, but that's not for everybody, right? Like every individual customer probably needs a sliver of that. 

And so like, how do you take that technology and all the data and then like all the historical data and then think of product journeys that are just that sliver that customer needs to simplify everything and hopefully to make it feel much more contextualized based on what we know about them. 

Fred Schonenberg
Yeah, and I think related and maybe slightly adjacent is that you talked about that customer journey sort of collapsing and that in the past, we've sort of, if you think of going to any website, it's very much a decision tree, right? It's like male, female, shoes or shirts, right? You start to make these choices and it's certainly gotten better over time. 

And some of that's more dynamic, but that some people like are ready to start it third base on that journey. They already know I want this, this, then this, and I've always bought that thing, right? So how can it be more dynamic? Can you talk about sort of that idea of the traditional customer journey sort of collapsing and becoming more dynamic?

Dan DeMeyere
Yeah, I mean, I think that there's sort of like two different ways in which I think that the funnel's gonna change. I think one of them is that right now when someone comes, especially if like a newer customer you may not know like why they're there today and you then may have multiple features, multiple pathways and journeys that you put in front of the customer to kind of solve for like every possible like modality for us of shopping.

Do you wanna search? Do you wanna browse? Do you want some sort of discovery-based experience? I think what's gonna change is that, and we're working on this, is like how do you detect that the modality that they're in that day as fast as possible? So within a few interactions, like how do you know if they already know what they want and you just need to get out of their way and get them to that product and then get them through checkout and collapse the funnel just for that one journey versus like, hey, this person clearly doesn't like have one thing in mind. How do we sort of figure out like, are they shopping for an occasion? 

Are they trying to figure out how to evolve their style? And like, how do you then steer them towards the funnel that makes the most sense for them? Or I think in the future, it's not even gonna be steering towards the funnel. It's like, how do you dynamically construct the journey that's like fitted for them? Because I think even when you say like, okay, how do I find what funnel is right for them? It's like, how do you send them down the best average experience? But if you can de-average the experience and dynamically construct journeys for every customer, I think that's where the unlock is gonna be.

Fred Schonenberg
I love that idea of dynamically constructing for each person. What is the biggest barrier to making that real today?

Dan DeMeyere
Well, the tech is almost there. I mean, I think that agentic tech in general is not the best performance. And I think that your data science, the way that sort of traditionally data science has been leveraged is like, okay, at night, you take everything you've learned in the last 24 hours and you take everything you've learned in like your history and you crunch and the model has a new output or whatever the data science technology is. 

Then the next time that customer comes back to the website or the app, then you've leveraged all of this crunching to be able to help inform something that is more intelligent or personalized than what they had previously. And I think getting that to be real time, I think is like one of the bigger challenges. It's not streaming. For us, it's like streaming and embedding a customer. Like the tech is there, but how do you do it such that, you come to a site, you start to click around things and then the background, we're doing some of that work to start to personalize. And then for us, we also have an agent that's looking at all of your event streams and then all of that needs to be coalesced into a decision of like, hey, let's put them down this journey or like let's simplify it or whatever that is. 

So I think that one of the problems is that part of the tech isn't there. I think that the other thing is that from a product tactician standpoint, like being able to understand like the art of what's possible with this tech is it's a little bit mind melting. Like if you've been in a product and you're building journeys, it's all, it's like linear. If they click on this and they go to this page or this is how they progressed on this funnel and it can be recursive and they can come back up and they go back on the same funnel. Getting to the world where you have this like composable product experience and where they go next, you don't know in advance.

It's sort of… it's gonna be dynamic. It's difficult to wrap your head around that. And so I think that there's part of this technology and part of it is just that sort of changing the way that like a product person can sort of wrap their head around all the permutations of what you can build for the customer. It's… yeah, it's very different than traditional product development.

Fred Schonenberg
And this might be an impossible question to answer in a very brief question. How do you help those product and engineering teams make that shift? Because it seems almost philosophical.

Dan DeMeyere
I think, yeah, no, it is tough. I think it's easier for engineers to sort of make that shift because they can think, they're already thinking about that technology and sort of in that way. And like what can be composed and orchestrated dynamically and with AI. I think from a product standpoint, I think what we found is like constrain the sort of the use cases initially. 

So instead of it being like the whole journey can be dynamically instructed, it'd be like, okay, what is the real estate? What are the components? What are the interventions that we know can be hyper-personalized? And let's just start there. And maybe you start with just content. Maybe you just start with product recommendations as opposed to full journeys and features. But I think once you can sort of constrain that and then you can kind of expand it to like, okay, we're gonna put this in email and push and the site and the app. And it's gonna be like a neuro scope that can be dynamically composed.

And then it's like, you can wrap your head around that and then you just slowly start to expand your mind a bit and sort of the use case. But it's tough. It's not something that people just get after like a couple of days of working with this tech.

Fred Schonenberg
I wonder what advice you would have for, let's say either a large retailer or a large brand on a retailer site and trying to figure this out. They're all very both excited and overwhelmed by the idea of agentic commerce, which of course that phrase can mean a lot of different things. What advice would you have for them if they were kind of coming at this, right? 

Because big companies, the first thing they think about is transformational capital T, like bringing in the consultants. We're gonna do a three-year project here, which both excites half the group and horrifies the other half. Curious what you would advise this sort of being, you're very much in this as a way to start to move towards this maybe agentic commerce future.

Dan DeMeyere
When I talk to others that are sort of like other CTOs or CPOs and I hear what they're doing, I think right now, like the word transformational, I think it's being overused. I think that you're seeing the technology provide a lot of productivity boosting for makers. But when it comes to the actual commerce side of it, what I've seen is that there's a lot more like novelty driven experiences and use cases and a little bit less of that transformational part of it. 

And that is, are you using AI to dramatically change how you deliver value to the customer for like the core customer problems, for those like core jobs to be done. I think that like you can do so many cool things with AI that being driven to like the novelty features or experiences, it's sort of the easy leap to make, but sort of like rewiring your business and your tech stack to be AI first and actually being transformational, I think that is something that not many people are doing. 

For us though, we started by doing some hackathons and I was like, let's just get everyone really immersed in all the tech that's available, start to build a little bit of this like collective knowledge on like what are some of the problem spaces that we think could change with this tech? And then typically it's like, okay, build conviction on something, create like a tiger team, so kind of start small, have like a cross-functional team that has all of their needs. 

So there's like no dependencies, no handshakes, like they're fully autonomous, have them go after that, share their learnings, have them iterate, and then you can sort of zoom out and you can say like, okay, like what's working with them? Do we wanna replicate this, have two tiger teams? Do we wanna take some learnings and change how we do things across the entire function? But for us, it was kind of starting small, like building the collective knowledge and awareness and making everyone feel comfortable with technology, but like starting small, building momentum, snowballing it and then rolling down the hill. And then over time, I think everyone, you can kind of feel like the broader momentum starts to ignite and that's worked really well for us. 

Fred Schonenberg
I have a very random, unscripted question for you. One of the things I heard from some other CTOs that I've met with recently is the engineers want to use a lot of different AI tools that are focused on whatever that one thing they wanna snowball is. But in large enterprises, you try to control the number of tools and especially with AI from a cybersecurity standpoint. 

How are you all thinking about that? And you can even use big examples, right? Like half the team wants to use GPT because it feels really safe and it's tied into Microsoft or Copilot, right? And the other half is like, no, I need Claude, I need to be going deep or I need Replit. How do you navigate that in a larger organization?

Dan DeMeyere
A year ago, it was very different. I would say a year ago, the team was exploring a lot of different technologies and you had pockets of people that were using things like Cursor to Devon, to like some Claude users. I think ChatGPT was a little bit late to like the coding space and Codex wasn't what Claude Code was but I think Cloud has a little bit for us to like to eat in the world in the last six months. I would say like 80, 90% of the team is Claude. 

But our view on that is, we try to put a lot of effort upfront if someone wants to use a new tool, a new technology, a new company and just like to make sure that, okay, like what do we know about them? Do they have access to the right information? And obviously not like sensitive information, like what can they be trusted with? And so if we feel really good about the company, the technology, then we leave it up to the team to cook. 

We want the team to kind of have them push the envelope as opposed to the leaders trying to determine for everyone what they should be using. And I think that if leaders are deciding that or even worse than like a technology leader, if there's like someone else that is not like a developer that's making these decisions, you're putting so much trust that they sort of like, they know what the best options are and they're the people who probably aren't spending all day, every day with the technology. 

And it is changing fast, like the fact that Cloud Code a year ago, if there's a fraction of the team using it and now it's the majority of the team, there's nothing preventing Gemini from really breaking through, even though they haven't been super strong on the development side. So the way that we've approached that is, it might sound like a little bit of a wild, wild west, but I think we try to put a lot of intention with which tools and technologies that we like green light from a security compliance standpoint. 

Then from there, the team can go wild. And then every Tuesday, we bring all of the engineers together just for 30 minutes. It's like, let's talk about what people are using, what we've learned. It can even be like what new developments have been happening in the open source space, but like let's all learn together, experiment together. And then we kind of just let the sort of organic adoption kind of go from there and the trends just sort of surface themselves. And yeah, that's how we approach it.

Fred Schonenberg
It's very cool. One of the things I love about doing the podcast is also a challenge. At the end, I have to come up with the title of the podcast and let them cook. It is right now trending in my mind. A great one. So let me ask you this. I'm gonna kind of give a little bit of rapid fire questions just to get your gut instinct on a few different things.

And the first of it is like, if we fast forward out three to five years, what do you think will feel just completely obvious that the shift was gonna happen, but that maybe is being underestimated today?

Dan DeMeyere
I think that a lot of people look at AI as something that drives productivity or something that enables a feature. And it's almost like a singular use of like, okay, we're gonna use AI like this and AI like that. And it can do this and it can do that as opposed to kind of zooming out and be like, what does your operating system look like in an AI world? And part of that is like your tech stack and like what needs to be composable? Like what needs to be a venture? And like whatever we'll need to change to sort of be in like an AI first world from a technology and architecture standpoint. I think that's one point of it.

But the other part is organizationally, like the talent is already changing. The ratio of engineers to product managers, that's changing. The skills that we're looking for when we hire, I think that's changing. And so I think that when it comes to product development, organizational design, even the way we think about talents and like the tech stack, all of it is changing. So if you're not thinking about all of those things, I think that you will find yourself behind and you might have tech debt, but you might have talent debt or org debt in three years if you're not currently working on that.

Fred Schonenberg
Very interesting. Is there an AI trend startup, something brewing out of those Tuesday meetings that you guys have together that you're particularly excited about?

Dan DeMeyere
For me, it's the combination of agentic technology and reinforcement learning and its ability to sort of de-average the product experience. I think that's the thing that I'm most excited about right now.

Fred Schonenberg
I love that. Is there anything that you feel is overhyped currently?

Dan DeMeyere
Maybe it's counter to what I just said, but the mainstream appetite for agentic commerce. I think that some people are trying to automate shopping and maybe for utilitarian or logistical needs, that's fine. Like finding a flight or getting a rental car. But I think some people like there's joy in retail therapy and there's joy in expressing yourself through fashion and sort of other types of purchasing. And so I think just like having agents do all that for you, I'm not sure that the consumer is there right now or will be there in 12 months.

Fred Schonenberg
Is there a product on the market out there today that you think is really nailing personalization, whether it's one of your own or elsewhere?

Dan DeMeyere
It's really hard to top TikTok. I mean, TikTok is basically the Holy Grail of personalization. So I think that anyone who's in that space is to a certain extent just trying to play catch up with them.

Fred Schonenberg
Okay. One of the things we talked about on stage back in Vegas was this idea of buy, build, partner. I'm curious how you think about that and has it evolved as AI sort of comes to the forefront? 

Dan DeMeyere
It's definitely changed. We have our own framework in terms of what's strategic to us. We tend to buy more. Sorry, build more as opposed to buy. But now I would say that the sort of default posture is probably build. Like what you can do, especially if it's like internal tooling or things that are not customer facing, like it really feels like it's just like the right set of two to three engineers can like build a lot of the things that you would typically look at in like the SaaS world is like, oh, we need to buy that thing. So yeah, I think that has shifted a lot more towards build where it was probably indexing more towards buy five years ago, three years ago.

Fred Schonenberg
Is there a startup out there that you are watching closely that you're excited about?

Dan DeMeyere
Daydream is an interesting startup that's sort of in our space and they're really trying to take a, like a genetic personalization first approach to shopping. I don't think they've cracked it, but you know, closely watching them. 

Fred Schonenberg
I love it. If you had to summarize what you think the future of commerce is, how would you summarize that?

Dan DeMeyere
Circular? Yeah, I think that there's going to be so much of sort of the consumer journey is gonna be sort of interwoven into these circular ways where you can kind of do all of the things, whether it's buy, return, coordinate, communicate, share. I think it's all gonna be like woven together in these really great circular experiences.

Fred Schonenberg
I love it. And I have to say also, like, from my perspective, what you all are doing at ThredUp is such an important evolution of where all this is going, right? It's making sort of a product that gets created, works harder and has, find more love with more people. Like all that feels very ahead of its time. 

Like, I think it's becoming, obviously it's been around for a while and it's doing very well, but it feels like that is where things are gonna need to go and then where things are going. So it's cool, it's cool to see how you guys are innovating within that space.

Dan DeMeyere
Thank you. Yeah, we feel the same way.

Fred Schonenberg
Well, Dan, thank you so much for making the time. We know how busy you are and it's been an absolute pleasure and I know everyone must've enjoyed it. Where would you want listeners to go to learn more about you or your company?

Dan DeMeyere
Just, if you're curious, ThredUp.com. But yeah, no personal plugs. And I think like, yeah, I really enjoyed our time on the stage and the times that we've had to chat. So I'd also recommend following you.

Fred Schonenberg
Thank you, man. Well, I appreciate it. Well, thanks for everything and we'll catch up soon. 

Dan DeMeyere
Okay. Have a good one.

Fred Schonenberg
Thank you.