Success Story: eBay Tackles Friction

Episode 5: Achieving Frictionless Commerce Podcast Series

Vertex Automation for Tax

Episode 5 Summary

Now’s your chance to hear a story about how an organization successfully reduced friction. In this episode Jacqui van der Leij, Head of Tax Technology at eBay walks listeners through the friction challenges her organization was facing, what solutions she considered and why, her biggest roadblocks within the process to finding a solution, and ultimately the outcomes that led to eBay’s success — plus, her advice for other retail IT professionals.

Achieving Frictionless Commerce: Helping IT professionals tackle friction within their organization.

Introduction: Welcome to Achieving Frictionless Commerce, a podcast series brought to you by Vertex, where tax and IT experts discuss how you can create a frictionless buying experience that will better support your customers and drive future growth for your organization—as well as the surprising way that tax plays a part.

Jeremiah Owyang (00:28): Developing a smooth customer buying experience is a necessity for today's businesses—it's no longer an option. For IT leaders that are challenged with making this happen, the good news is that there is help out there. 

Welcome to Achieving Frictionless Commerce. My name is Jeremiah Owyang, a technology analyst and entrepreneur based in the Silicon Valley and your MC. Join me and my guests, experts from Vertex, a leading global provider of indirect tax software and solutions in our six-part podcast series. Together, we will discuss what the term means right now, what you can do to minimize any barriers in your customer journey, and the best steps to take that will lead to long-term organizational growth. We will also take a deeper dive into the somewhat surprising ways that tax plays a major role in reducing friction and give you a powerful example of one retailer who used tax software to do just that.

Today I'm joined by Jacqui Van der Leij, the Global Head of Tax Technology at eBay to give us some insight on how she tackled and reduced friction at eBay with the help of Vertex. 

Welcome, Jacqui.

Jacqui van der Leij (01:44): Thank you, Jeremiah. 

Jeremiah Owyang (01:46): Before we jump in, I would love to hear a little bit about your background and about your role at eBay.

Jacqui van der Leij (01:53): Sure. Right now I'm, as you've said, the Global Head of Tax Technology at eBay. Our team really focuses on finding innovative technology solutions to ensure that eBay can keep up with legislative demands as they come in from tax authorities but also as we as a business progress and change to make sure that we do the right things at the end of the day.

I've been with eBay almost 13 years in a variety of different finance technology and data and automation roles which I've thoroughly enjoyed. I think it's been quite a journey with eBay and prior to that, I spent almost 10 years at PricewaterhouseCoopers working across the globe, finance background, auditing background, but also doing much more interdata modeling and technology as I progressed through my career. 

Jeremiah Owyang (02:41): Impressive background, Jacqui. First, can you tell us why reducing friction is so important to eBay?

Jacqui van der Leij (02:49): Sure. You know, ultimately, we are a marketplace and it's all about our users, our buyer and our sellers. And if there is friction for a buyer or there's friction for a seller, whether they are listing, whether it's the tax part, no matter which part of their transaction, we fail as a company. And for us, making sure that there is this frictionless effort from all our users on our platform, this will be the most critical for us in terms of measuring our success.

Jeremiah Owyang (03:22): As a user of eBay, that makes total sense. It needs to be frictionless and seamless, but I would love to hear. What were the problems that eBay was trying to solve for and what were your goals?

Jacqui van der Leij (03:33): Sure. So, a couple of years ago legislation has changed dramatically for marketplaces like eBay where we, as the platform is now responsible for the collection and remaining of sales taxes. Ultimately, the goal is that for eBay especially, you know, we want our sellers to be able to list almost anything, within reason of course on our site, in any way they want to list that on our site. However, with this new legislation that must almost coming to drive how our sellers list and how our buyers buy. And our goal was to become agnostic, tax should be agnostic, to our site experience. And how were we able to make sure that those categories in which our sellers are listing, they are not always aligned with what they call a tax category and that impacts the taxability. So, for us the whole goal was how do we make sure that they have this frictionless, easy experience that they're used to, but we are still compliant? 
And what I mean by compliant, there's three perspectives here, right? There's a perspective that eBay is financially liable but also my buyers should pay the right price. A buyer should not pay an inflated tax price because we got the tax category wrong. And my seller needs to stay competitive because if we charge the wrong tax, my seller is not competitive on other marketplaces. So, the whole goal was how do we get this right in the most effective and easy way for our buyers and our sellers.

Jeremiah Owyang (05:08): That's a very challenging proposition to think about all three side of the marketplace there, buyers, sellers, and of course your own company, and then doing it in real time as quickly as possible. Who was involved in trying to determine these solutions and why were they involved?

Jacqui van der Leij (05:25): Sure, so it almost brings me to what solutions we were looking at. My first step was I got Vertex involved and it's actually a, I call it my couch story-

Jeremiah Owyang (05:34): Ah, okay. 

Jacqui van der Leij (05:35): ... because I was obtaining a Vertex Exchange, it was my first Vertex Exchange a couple of years back and I was sitting on a couch and talking to a gentleman from Vertex who's in the development space at that point of time, David Deputy, and he did this presentation earlier in the keynote and I was like, "Listen. I have this problem and I want to go on this journey, but I want to go on this journey fast." And six weeks later, they were on campus at eBay and we started looking at utilizing artificial intelligence to actually solve this problem.

We of course then had Vertex involved, we had to involve my product and engineering folks because in the end of the day they need to implement these models in our site. And at that point of time our tax technology was pretty new, we were two or three people, so it also involved RSM who is a partner of Vertex, but they're also a tax technology consulting firm. And together we created this team of people who actively worked on finding the solution. 

Jeremiah Owyang (06:35): Love that. The couch story to action in just a few weeks, that a great story, thank you. What other solutions did you consider as you were navigating through this process?

Jacqui van der Leij (06:46): So, I mean, there's obviously the first solution, which is just manual mapping, which is just not sustainable, right? We also considered building hybrid where we said, "Hey, it's a manual mapping with a use interface," but all of those resulted in not the right level of accuracy. Operational efficiency was just at zero. The return on investment will just not be there, it will just still not give me the accuracy and I would just put bodies at it, which is not the solution either. A final solution was to indeed change the way that our sellers list but that was obviously for us as a company not even an option.

Jeremiah Owyang (07:28): Let's talk more about that, uh, solution that you chose and why.

Jacqui van der Leij (07:32): So, we ended up using artificial intelligence to actually read our listings on the site and determine the tax category with that. Now, to give you an idea of the challenges, we have over 34,000 categories on the site.

Jeremiah Owyang (07:45): Wow.

Jacqui van der Leij (07:46): And we maybe use 700 tax categories, right?

Jeremiah Owyang (07:49): That's a lot. Yeah.

Jacqui van der Leij (07:51): It's mind blowing, so, but one of the things we have is we have a lot of data, so we were able to use that data and with the help of Vertex, build an algorithm that will read that and predict the tax category for us. We can keep learning from our data as things change or as new sites or countries introduce this type of marketplace legislation, we will be able to easily adapt and then actually roll that out without having to go through manual mapping exercises.

We took it a step further where we said, "Okay, how about the ability to teach this engine to speak the local languages?" Now, with all respect to translating engines, it doesn't capture the gist of a true language. Speaking a couple of European languages myself, I know that what the translation engine says and what the real of way of speaking is different. So, we were able to really train these models to be able to really understand on the German site, when a German seller listed in a certain way, that we were able to read and determine what that tax category is.

Jeremiah Owyang (08:56): You're right, that is not sustainable to do that manually at all, so relying on AI, that's clearly the way to do that. Would love to hear, though, what were some of the biggest roadblocks you were facing?

Jacqui van der Leij (09:08): I think the biggest one and—I think you will get that in any innovation or any state where you start—you're not just using AIML to do a job, you're doing something that's going to touch your customers, right? Because we don't want to get it wrong. So, the biggest hurdle was just to bring people on this journey. Not everybody is in the world of AIML. This hasn't been done before, it was brand new, so bringing people onto that journey and convincing them that, "Listen, it's okay to invest in this world and it's not going to go rogue," those were the biggest roadblocks we had.
The second roadblock I think we had which wasn't per se a roadblock, it was more something that we had to get over or (or find solutions for) is your data. You can have a lot of data, but that data has to be usable. So, finding smart and innovative ways—because we were not going to sit there and humanly label these listings to get the data—so we had to think very smartly and we used multiple tools to actually able to clean our data and to train our models. So, that stumbling block was something that we had to address, and I can happily say that today we've addressed that. We are able to really through automation clean our data and train our models without issue within a couple of hours.

Jeremiah Owyang (10:32): It's interesting, Jacqui, that the first challenge you mentioned would be an internal one with stakeholders, but we can't forget how important it is to bring along the organization. Looking back in retrospect, was there anything that you wished that you did differently through this process?

Jacqui van der Leij (10:48): You know, it's exactly that. We, sort of bringing in or getting our stakeholders, sort of decision makers later in the game. My biggest change I would make is just to bring them in much earlier. My original thinking was, "Get a proof concept, show them the proof is in the pudding, and then take them on the journey." I think if I brought them in earlier and then that would've made my journey faster and easier because I would've been able to do that education. I think we sometimes forget that everybody is in a different place in the journey—and you want them all to reach that same destination and it might require investments of different levels to help to get folks there. So, I think I would've started earlier, and I would've really made sure that folks who's really in the beginning of the journey, that I would be able to get them there faster.

Jeremiah Owyang (11:40): Speaking of that journey, how do you measure and track success?

Jacqui van der Leij (11:45): So, there is quite a few angles. I think the most obvious is just the accuracy, making sure that the taxes that we charge at checkout, that that's accurate would be the most obvious one. But for me, there's other things that is bigger than just the accuracy. I think the number of calls that we get from our customers about tax accuracy, that was something that we started seeing quite quickly and quite early. You know, you asked me about when did we start seeing results. As soon as we were ramped close to 100%, we started seeing the friction—obviously from sellers reaching out—to be down to 30 to 40% reduction in those contact points around taxability. We still see that because that's one of our key metrics, is we see it… we look at what's coming in, what type of questions are coming in, and what is the level of impact?

Jacqui van der Leij (12:39): Another way we measure success is just looking at how our tax authorities are auditing us and what the outcome of that is and we've definitely seen some great, positive results from various tax authorities in terms of the use of this. Another thing that we look at is just an overall operational efficiency, right? The cost, the ROI. As I mentioned before, if I had to do manual mapping and the amount of people I had to do and if I had to maintain that versus what we're maintaining today, so even when we look at the enhancements of these tools or whatever, we look at what is that ROI in the overall picture? And then, you know, as I mentioned before it's also the liability of eBay… so what is our exposure looking like, what is our accuracy level looking like, and what is potential liability that could hit us?

Jeremiah Owyang (13:29): Have you also measured the speed at which the site performs when you're doing these transactions?

Jacqui van der Leij (13:36): Absolutely. For us, even a nanosecond-

Jeremiah Owyang (13:39): Right.

Jacqui van der Leij (13:39): ... is for us a, an issue. So that's why we have built it in such a way that we are able to maintain our checkout speeds at the same way, regardless of these tools sitting in between. So, checkout speed is something that we test in general, not just in tax, so for us, that's pretty critical.

Jeremiah Owyang (13:59): Those are amazing capabilities and metrics. Looking broader, what have been some of your results to date?

Jacqui van der Leij (14:07): I think we are still seeing a general innovation inside of our people because I know that we're talking about friction between our buyer and our sellers, but the market that we are in today, we have to constantly think of new ideas and new ways to make sure that there's this less friction. So, I'm seeing the teams themselves who's working on these models—the engineers, the analysts, the data scientists, and even Vertex—working together to constantly improve what we're doing, how we're getting there, and how are we not just accurate on the side, but also operationally efficient. I'm definitely still seeing us being much more agile to legislation changes because we sometimes see these changes going really quickly and I think if I look back at what has been happening in the market, the speed at which we can respond is just significantly faster. 

Jacqui van der Leij (15:00): I think the biggest difference between what we've built in tax or in general what you build in tax and other products is… that the way it goes is we build something, you put in production, and it goes almost like in a maintenance mode. Tax is a constantly changing environment (nonstop) especially for a global company like ourselves, which is exposed to legislation everywhere. We keep seeing those changes and for us it doesn't ever really go in maintenance mode. It's about making sure that we have a product that is able to adjust quickly to whatever changes may come. So, for us I think we keep in staying in a development and production zone and we have to need to have the technology and the ability to adjust to that. 

Jeremiah Owyang (15:48): I think that's a great point, the agility needed in this ever-changing future. Certainly, there's many folks listening who are envious of the journey that you've taken, Jacqui. You mentioned starting this journey at the couch of Vertex Exchange. From that moment, how long did the process take and what was it like working with the team at Vertex?

Jacqui van der Leij (16:09): I think it was six weeks later the Vertex team was on site. We obviously had to get some NDAs and some legal stuff sorted out, otherwise I think it would've been much faster. The team was there… we became like a family [who] to this day we even still send messages on our birthdays. 

Jeremiah Owyang (16:27): Wow.

Jacqui van der Leij (16:27): We really became a true innovation team, and nobody was holding back. Vertex was not afraid to being in different experts. If they didn't have an expert, they might have had a partner who had an expert to come in and to look at it. They didn't just say, "Oh, here's your algorithm, you're on your own." It was really helping us through the whole process, and I think from the start [since] they came on site until we have our first predictions was about three months.

Jeremiah Owyang (16:57): Jacqui, looking to the future, who do you imagine that tax and tax technology will change?

Jacqui van der Leij (17:03): First of all, we are seeing changes in the marketplace legislation changing globally. Whether we're seeing a rise in marketplace legislation in APAC… we're seeing more of the states introducing new types of legislation, so I'm definitely seeing us having the ability to either understand the content or be able to identify those transactions—and for us that all involve, of course, AIML. eBay as a company—as we have seen our company growing and changing with our focus categories—introducing authentication options for our buyers. We are seeing that there are new types of fees coming, there's new type of taxabilities coming that we in a space that was new into which means had bought impact two sides, Vertex itself (as a tax engine for us to be able to make sure we calculate it right), but then also our artificial intelligence solutions (to be able to determine the right category).

Jacqui van der Leij (18:00): On top of that what we shouldn't forget is that there's a lot of data around tax, right? So, when you asked me how tax technology is changing, the ability to wrangle that data, the ability to get the right data out there and the right reporting to our tax attorneys, and also the insights of what we can gain from that data.

Jeremiah Owyang (18:20): Do you have any advice for retail organizations in similar situations to you?

Jacqui van der Leij (18:24): Sure. The number one thing is that you have to think outside of the box. We are not in the situation anymore where, "Oh, we need to do x, we do x”. You're going to have to consider is y or z is an option or maybe a combination of all three. And I think you need to stay back and not be afraid to think outside of the box and not be afraid to go on the journey, because I think we sometimes are so afraid of the unknown that we don't step into it and actually try. I think talking to your partners, whether it's your vendors, whether it's folks in other part of the industry, I believe that that is absolutely critical. I cannot imagine going on this journey without having a tax technology partner in RSM or Vertex who is really, really open to trying something new because this was definitely a new space for them. Finally, speaking internally to other folks and hearing their opinions, but also really listening, for us, listening to our customers. What is it that they need and how do I make sure that all of these things come together? 
Jacqui van der Leij (19:34): But I think that the most important thing is to really open up innovation. Open up innovation for yourself, for your team members. And then making sure that you talk about innovation openly and realize that innovation doesn't mean (in my case we didn't have a data scientist on our team, we didn't have a whole modeling set up, we didn’t exist) but by speaking to other people in the industry or folks who had experience, we were able to build the solution and successfully launch it. And Jeremiah, just one of the things we didn't touch here was is that from that six weeks later when Vertex came on site, three months later our proof of concept was there. 

Jeremiah Owyang (20:19): That's fast.

Jacqui van der Leij (20:20): I mean, we're predicting our first transactions and we're talking about multiple companies and teams having to come together without ever having done that. You're forcing a group of different people into a room and say, "Come up with a solution." And sometimes that scares folks. And I can tell you based on our experience and based with a group of people we had around the room—and there were lots of discussions and lots of whiteboarding—that all of us came together and were able to put it out there at that speed just shows if you just have that open mind and that open communication, I do believe that it's possible. And if it didn't work, you would've learned from it still so don't be afraid to try and take it out there. 

Jeremiah Owyang (21:13): Thanks so much to Jacqui Van der Leij for coming on the show today to share the success story of how eBay has managed to thoughtfully reduce friction. Don't forget to join us on our next episode where we discuss what your team can do to ensure your organization is set for the future and continues to scale. Subscribe now so you don't miss it. I'm Jeremiah Owyang, your MC and thank you for listening. 

About the Podcast Series

Are you a retail IT professional who is looking to learn more about frictionless commerce and how you can achieve that within your organization? If so, check out the Achieving Frictionless Commerce podcast series! Hosted by technology analyst and entrepreneur Jeremiah Owyang, experts from Vertex discuss what frictionless commerce means today, how you can discover and overcome roadblocks, and what you should do to prepare for future retail growth.

Sales Tax Compliance Solutions for Retail & E-Commerce

Explore cloud-based solutions built to automate commerce for today’s global e-commerce businesses.

Person purchasing eRetail product on phone.