This week we are joined by Romain Vermeulen, Head of Data Science at Wisepops. With experience working for tech giants like Facebook and Amazon, Romain shares his insights on the current state of online marketing and website visitor engagement. He explains how Wisepops is helping brands focus on their own websites and engage visitors with on-site channels like notifications and pop-ups, all while being mindful of what data they need to collect to deliver.
Tune in to learn more about Wisepops' mission and how they are revolutionizing on-site marketing.
Henry - 00:00:00 Welcome to Ethical Data, Explained. Join us as we discuss data-related obstacles and opportunities with entrepreneurs, cyber security specialists, lawmakers, and even hackers to get a better understanding of how to handle data ethically and legally. Here to keep you informed in this data-saturated world is your host, Henry NG. Good morning. Good afternoon. Welcome back to Ethical Data, Explained. I'm your host today, Henry NG, and today we have a very special guest. Data scientist. Worked for Facebook. Currently works for Wisepops. Worked for Amazon. He's got a long list on his CV that makes him a very impressive individual, from analytics to intelligence. All the way down to data science itself. We have Romain Vermeulen. And, yeah, Romain. Probably I said, mispronounce your surname, but feel free to correct me and tell us a little bit about yourself, your background, and your experience in the tech industry.
Romain - 00:01:00 Yeah. Well, first of all, thanks a lot for having me. My honor to be invited. I must say, it's my first podcast, so I'm not exactly used to this, so a good experience anyways. Thanks for the introduction. So, as you mentioned, I'm head of data science at the moment at Wisepops. I've been here for about two years. So, helping to shape the tech for the first marketing platform that helps to engage visitors directly on our customer’s website. The approach is that it was about eight years at Amazon and Facebook helping to scale the e-commerce platforms with a focus on large-scale data and operations. So keep it simple. And over my career, I've moved between France and China, and I'm currently based in Singapore for about the past seven years.
Henry - 00:02:14 Brilliant. Brilliant. I've recently come back from Singapore. Absolutely. I love the food there. I love the hawker markets. I love the weather. Great place to visit.
Romain - 00:02:23 Yeah, a great place to visit. We are loving it, too.
Henry - 00:02:25 Exactly. Exactly. So, Wisepops, you're there at the moment. Tell us a little bit about it. What's the kind of product that you offer? What's the company's mission and who do you target as clients on a regular basis?
Romain - 00:02:39 Yeah. I'll start with a big number. Essentially, if you look at the global online ad spend, it's about $600 billion. So, a huge market. And if you look at where it spent, it's mostly brands. Paying Facebook and Google, TikTok, and other providers to acquire traffic. So they are using those platforms to target customers and get them to come to their website. But once they reach their website, actually little happens purely in terms of this budget. Less than 5% of it goes to the brand's website in terms of marketing and engagement, and reach. And it kind of results in on-site marketing being pretty far behind the marketing that happens on the big platforms, something that I've obviously seen in my career. The large platforms are very impressive at giving you tools to run your business and target your audience and communicate messages on the large platforms. But when it comes to your own website, there are actually very few levers for you to play with. And so the outcome is you have 100 visitors coming to your site, less than 2.5 end up converting. And so we think that brands should focus a lot more on what happens on their site. And that's where Wisepops come in. So we offer them on-site channels, typically notifications or pop-ups, or any other ways that we can engage with the visitor while they are on the site. With the purpose to provide a delightful experience, understanding what they are looking for, trying to share the messages that the brands want to share with them, and ultimately trying to convert them.
Romain - 00:05:11 And the timing is actually quite relevant with the recent privacy laws. They make targeting a lot less efficient on large platforms. And so that means that brands end up paying a lot more for fewer results. It used to be the case that you could have your website mostly up and running. You have decent product pages, you have a smooth checkout process. You know, your site is pretty well set up. And if you want more sales, you would acquire more traffic and they would simply... You would push more people through your conversion funnel. And that's basically how things have been. But now that the targeting on the platforms is less accurate and more expensive, we think it is time that brands take... pay more pay, more attention to kind of the right side of the funnel. So what happens once the visitors land on their site and how do they interact with the brand? So, yeah, that's kind of where Wisepops come in.
Henry - 00:06:21 Yeah.
Romain - 00:06:21 Online advertisement space.
Henry - 00:06:24 Perfect. Perfect. So you've talked a lot about brands. Do you only kind of work within that e-commerce realm or do you work with people outside of the world of e-commerce as well to cement their brands or websites?
Romain - 00:06:36 We are not only e-commerce. E-commerce is still like 70% of our customer base. But we work with different industries, being NGOs or gaming or we don't really have a strong focus apart from e-commerce. Anyone with a website I think should benefit from the ability to engage with the audience on their website. And so it's not specific to e-commerce.
Henry - 00:07:12 Brilliant.
Romain - 00:07:12 And we have small and large customers. Small, it's usually some 80/20 share. We have a majority of small sites, but we do get larger ones as well. Like L'Oreal or Unilever do use Wisepops on some of their sites.
Henry - 00:07:33 Okay.
Woman: - 00:07:34 This podcast is brought to you by SOAX, a leading proxy provider enabling your business to unlock the world of publicly available data. Get data at scale, SOAX Data.
Henry - 00:07:51 Cool. And what attracted you to Wisepops? I know that both you and the CEO, Benjamin Kane, you both worked at Amazon. Was there any link there? Do you feel like there's any kind of influence and experiences you've taken from Amazon into Wisepops as well?
Romain - 00:08:08 For sure. So, Ben and I are friends since our days at Amazon. Quite a few years ago. And it's been quite the journey I would say. So Ben started with Wisepops back then, and I've been kind of on the side admiring the shaping. He was able to bootstrap Wisepops from scratch. From basically $0 to 1 million of recurring revenue to 2 million of recurring revenue. And we are on our way to $2-3 million dollars of ARR. So I've always been in awe at the feat that this represents. I have a lot of admiration for him. But as I said, I was kind of watching it on the side for a while until the opportunity and the timing finally were right. I was particularly attracted by the decentralization nature of the mission of Wisepops, which is to empower brands to engage with their audience on their own website. They are very dependent on the large platforms, and there is a huge gap between what is offered and available on the platforms and what they can do on their own.
Romain - 00:09:38 So kind of this mission to empower individual stores was appealing to me. And Wisepops was also a cornerstone in their journey. It was born with a much smaller scope and it was initially a more simple pop-up tool. But now the mission has gotten pretty, pretty big. And we want to be a full-fledged marketing platform. And so there was a lot of need to be more data-driven, to have intelligent products and multiple channels, and provide easy-to-use and automated tools and campaigns. And being a much smaller structure than big companies also meant that I would have a much more direct impact on the business. And that's something that was appealing to me.
Henry - 00:10:39 Okay.
Romain - 00:10:39 So in the end, yeah, kind of a planet alignment between the inspiring mission and the ability to have a direct impact, plus the fact that Ben and I have good trust and we are good friends. All of that made it possible. And I'm still enjoying it.
Henry - 00:11:05 Brilliant.
Henry - 00:11:06 So, Ben, if you end up listening to this podcast, Romain is very, very happy to work with you by the sounds of it. And for Wisepops. But so, Romain, you came to Wisepops to kind of work pretty much as the head of data science. So data science is a very broad topic in general. What do you focus on a day to day? What does your role actually entail?
Romain - 00:11:30 Yeah. Well, first of all, pay little attention to job titles, be it now in a small company where it's, as you say, it's pretty broad. And if the company is not that big, you end up doing a lot of different roles. But the same is in a bigger company. I even tried it myself for a while at Facebook. We have carried this same job title for years that referenced an organization that was there when I joined but had disappeared for years. And I kind of like to keep this old job title because it didn't mean much. It didn't reflect anyways what I was doing. So with that being said, I'm still happy to talk about what I focus on a day to day. I can maybe split it into three, three areas. One is the core of my role, which is around data. And I'm helping to build the foundations for Wisepops to be a data-driven product company, and that goes from all the way to collecting the data and storing it, to processing it, and then delivering it into our products. So that has many facets to it, very technical and engineering ones, more scientific ones. When we build models as well as some operational and even legal sides of it. So, it's a bit daunting. But essentially I'm kind of looking at this angle primarily.
Romain - 00:13:18 And yeah. Do a lot of. I have it on my own leverage and also the team to integrate and share responsibilities and share the fun. The second angle would be something like innovation. I often try to shape the work I do in the form of a technical demo. It may span a bit beyond data, but for instance, last year, I built a proof of concept for automated on-site notifications that we could build on Shopify. And so now we're trying to turn that into an actual product. Or more recently I've built a demo for an analytics dashboard. And we are working to actually ship it to our clients. So I'm trying to turn... yeah, to showcase what is possible with different technologies and hoping that we can turn these into a product as a team. And the last angle is a bit more of a vague umbrella of technical leadership. I would say, I'm also involved in the early stages of conversations when we are shaping ideas and potential features that we want to work on. I'm trying to bridge, what's possible and what are some technical implications to the technical side of things. So between the product strategy and the technical capabilities. It's also a role that I'm enjoying.
Henry - 00:15:20 Man who wears many hats, having the whole lot at Wisepops. So out of obviously those three sections and your roles in general, what would you say is the most rewarding thing about your work and what would you say is the most challenging thing about your work?
Romain - 00:15:38 Good question. I would start with the most challenging one. Something that was less exposed to in big companies. Typically in a big company, you are not exposed to the end-to-end of what you're working on. That means... So, usually have different teams in charge of different aspects, from data engineering may be to web distribution or to legal aspects. And when I came to Wisepops I was responsible for shipping what I was building into production. And that was quite a challenge. The reason is that Wisepops is obviously not operating at Facebook's scale. We are much smaller, but still, we process some 10 billion requests per month covering the millions of visitors that thousands of our customers have. And so that means that we also cannot mess up with the services that we put together. Our infrastructure needs to be sufficiently well-designed so that it doesn't break our client's website or our bank accounts. We need to come up with a good enough architecture and scalability from the start. And so it's a pretty high bar.
Romain - 00:17:15 You would think if you're a startup, or if you work for your own company, you only need to address the needs of your own company. But as a SaaS, there's this multiplier effect where we need to address the needs of all of our companies. And so instead of scaling for one, we need to scale for a thousand since Wisepops was already a pretty mature business in a way. And so there were tons of good learnings and challenges. So I learned a lot from it. We have very short iteration cycles, so between the moment we ideate on some new concept to this phase where we experiment on it and then we're going to go about implementation and then we're going to go about getting customer feedback. All of that happens pretty quickly. And it's quite rewarding to be able to oversee, what you're working on from end-to-end and get direct feedback on it.
Henry - 00:18:23 I think that rings true. So on your website, you have a quote that says "Wisepops breaks the technological wall that separates brands and people". And I feel like even on small projects that kind of just showed how that works and I know one of the key ways that Wisepops works, if I'm not mistaken, is that you look a lot into the clients' on-site audience and work on customization for things like their pop-ups. And based on audience behaviors and profiles, how do you collect that data and do you feel like collecting that data helps improve the customer experience overall? And obviously, adhering to a slew of GDPR and CCPA goes behind that. But just on the basis is how... What's the process of collecting that data and using that data from your side?
Romain - 00:19:13 Yeah, sure. Maybe, first of all, we are privacy-centric. And so anything we do, we need to comply with the GDPR and the different privacy laws. And so we are quite careful to ask for consent before we collect any visitor data. But that being said, our whole product is about engaging with a visitor. So at some point, we also need to deliver a message to the customer's visitor.
Romain - 00:19:58 And so we do collect data. So once we got the approval to collect data, we load a piece of code on the website so our customers install Wisepops on their website and Wisepops will monitor simple usage activity to personalize the experience. We are really focused on first-party data. And so in order to personalize the experience during the browsing session, we really focus on that particular browsing session. We are not there to harvest whatever personal information we can get from third parties. We don't touch that at all. We simply stick to what happens on the website when the visitor joins the website. And so if they consent to the cookie and the privacy policies, then we will do anonymous tracking on the browser. There are also some integrations that customers can bring, and so we can map, for instance, their segments on Klaviyo or on Shopify to better target the audience with particular campaigns. And so it's as simple as that. So we are merely responsible for collecting this data and storing it. The data belongs to the customer. We are not there to reset it, not our business.
Romain - 00:21:53 That's not what we want to do. And so we simply collect first-party customer data our customers have access to it. It is just anonymous data. And they can decide later... The compliance piece means they can access it. They can request the data be deleted. And at the end of that, it enables us to do some profiling on the visitor, what device, based on what device they use, what geographical location they are in, what kind of products they see on the page. We try to infer what their goal is, where they are at in the conversion funnel. Are they going to convert? Do they need a little nudge to complete the cart? Would they be interested in this other product? And so we try to really figure out the minimal amount of personalization we can take without much information because, at the end of the day, the cookie laws do impact us. We will obviously be a lot more capable of personalizing stuff by calling someone by their name or things like this.
Romain - 00:23:26 But if you take the parallel to like a real store you enter. A real physical store for the first time. And the salesperson there welcomes you by your name. And you're "What? How do you know me?" It would definitely not be a good experience for you. But on the other hand, as you walk the store, the salesperson will understand what you're looking for and will be able to provide you with better advice and give you a bit more personal touch during your actual walk of the store. So it's more like this lens that we are trying to adopt. Like you are on the website, you engage with the brand. We try to understand where you're at in your customer journey and map that to what the brand has in mind in terms of messaging that they want to deliver, or the experience they want their visitors to have. So we try to bridge this gap with a little bit of profiling that we're able to capture on the site. It's quite an interesting exercise, that combines a bit of data science, obviously, but as well as some psychology in a way.
Henry - 00:24:45 Sounds like a lot of detective work. Basically, it's not about the person. It's about what they're doing and the small actions of what they're looking at. That determines how you create that profile.
Romain - 00:24:57 Yeah.
Henry - 00:24:58 That must be quite a difficult job for you on a regular basis. I'm not gonna lie. What was, especially managing that data and having that information to hand and that limited information to hand? What are the challenges that you face, you know, in managing that data, and how do you normally address those challenges? Not being able to use due to the cookie laws, the in-depth information about a client? Has that been a major issue for you on a regular basis?
Romain - 00:25:31 I would not say it's making it a lot more challenging for us. We do need to be more careful to make sure that the data we collect is anonymous and sometimes that requires extra steps. For instance, if you at some point have a log with the IP address, the IP address is identifiable data, and we need to obfuscate it. So we do some processing on the IP address so they become anonymous. If you capture just some payload from a web page, it may include stuff like some personal information from the user. And you don't want that to end up in your database either. So, we need to pay this extra attention to what collect. But also, I think it's mostly industry standards. For example, we run encryption. From the moment the data is in transit to the moment it is in storage. We provide access and management to the client, to our clients. We secure the data internally as well. We are not a big team, but still, we limit who in the team can access production data.
Romain - 00:27:05 So things like this. Overall, the less we do and the more simple we keep it. I think, easier it is to keep it secure. And so we don't have that many processes where the data is moving everywhere. It's a pretty straightforward flow and so it's more secure by design. Kind of outside of the data, I think in the end we are obviously impacted by having a bit fewer signals than we could have from on-site visitors due to the privacy constraints. And I also think we are less exposed because, as I said, our model relies mostly on first-party data and unlike some other advertisers that rely extensively on the ability to track you amongst multiple websites, to understand that you've been visiting this type of product on this website and so they can promote you to that on another website. Our model is not that. We are really focused on a brand owner's own website and so I think that has also lessened our exposure to those privacy laws. And so we need to comply and we are compliant. But it has not affected us fundamentally, I would say.
Henry - 00:28:39 It does sound like Wisepops is more focused than a lot of the other providers and really focused on the right data, the right level of data, and also directly into the brand itself. And obviously with what you said about Ben starting Wisepops and the progress that you've gotten over the number of years. 2023 is a new year. Is there any kind of room for growth or innovation, in your opinion, for your service? And do you feel the rise of ML and AI-tech, is something that is going to be integrated into any products or any competition to your product, do you think?
Romain - 00:29:21 Yeah, lots of underlying questions. So we do have some machine learning already. We do have some plans to have more. In terms of opportunity, as said earlier, the purely market size is pretty vast and quite lagging behind where it should be. So there's really a lot of opportunity. The tools and the maturity of on-site marketing are really behind that of the larger e-commerce platforms. And so in a way, it's a bit unfortunate, but it's also fortunate for us because we don't need to look too far to have big pain points that we can help our customers to solve. So for instance, one of the recurring pain points in e-commerce is cart abandonment. So about 70% of visitors who start, who add something to their cart, who did would not end up purchasing it. Picture yourself in that hypothetical physical store we had earlier. You can picture hundred people with stuff in their cart and they leave it on the ground and they leave the store. And you want to understand why they did not go to the cashier and made the purchase. And so we want to start looking into this and equip a brand, equip site owners with smart segments that predict cart abandonment that they could use in their campaigns to identify those visitors that are likely to abandon their card, but also likely to convert if they are given the right nudge. And so we are going to experiment both on the prediction piece and the way that we can influence that.
Romain - 00:31:29 It is just one example. As I said, the problems are pretty well-known. We don't need to reinvent the wheel. And there's a big opportunity ahead. I think, in terms of timing, things are accelerating in on-site marketing. And you could say in a way it's reaching a certain tipping point. If you look back some six-plus years ago having a pop-up, to prompt users to subscribe to the newsletter was relatively a new use case. And right now it's become a really clear sales channel so there's an industry benchmark you can compare to in terms of how any conversions and how many dollars you can get from collecting this audience. And we started from there. And then we thought what's the next competitor that's going to come at us? And that's actually how Ben came up with the idea of on-site notifications. He was trying to picture what would be disruptive to Wisepops' business. And instead of letting it disrupt that, he went ahead and started building it. So we came up with the notifications. What we are trying to do now is that notifications is also trying to convert them into established sales channels. So figuring out what use cases that work well with this little bell and this little inbox that you can have in your inbox on the customer website. It's a much smaller footprint, but it's much deeper content that you can fit into it. So it works very well for mobile devices, for instance. And so we're trying to identify the right ways to interact with the visitors through these notifications and reach a place where it's also become very mature sales channels when you know you're going to install the bell on your website. And it's pretty clear what kind of returns you're going to get and what kind of dollar value you can associate with that. So I think it's kind of where we are heading.
Romain - 00:33:56 And it just feels really like the scratch of the surface of what we could be doing behind. But things, as I said, are kind of coming together. These past two years we've invested a lot into our company engine or tech infrastructure, or data stack. And those things are in place now. So we are pretty well equipped, to start building the use cases one after another. And the good news is that they are well-known. You ask an e-commerce store owner what are the pain points. So some of them are kind of out of our scope, but a good chunk is something that we can influence by providing a more intelligent campaign on their own websites.
Henry - 00:34:48 Sounds exciting.
Romain - 00:34:49 This is where... Yeah, it's quite exciting. Good market, good timing, and a good team. It's usually the framework that is used to evaluate an opportunity. And I do think this one is a good one.
Henry - 00:35:04 Brilliant. That was really good to hear and really excited to see how Wisepops develops in 2023 and onwards. Not just scratching the surface, but making real shifts in the market as well. So we come to the end of our kind of normal interview questions. We have three questions that we ask all of our guests before we wrap up. The first one is who in the world of data or technology would you most like to take out to lunch? You can say, Ben, if Ben's listening, I'm sure he'd love to.
Romain - 00:35:33 Yeah, well, it's not. It's achievable, let's say. I don't need to wish for it. Apart from a geographical distance. But we do get lunch whenever we are together. I was thinking of Greg Brockman actually. He's the founder and co-founder, and president of OpenAI. So, behind ChatGpt and I think before that he was CTO at Stripe and both companies are huge. And he has also grown them from the funding stage to how big they are now. So I'd be very curious to hear the stories and the learnings and learn how it's done, yeah.
Henry - 00:36:22 Yeah.
Romain - 00:36:23 Pretty amazing.
Henry - 00:36:24 If for any reason Greg is listening to this, we'll try and organize something and you can get in contact with Romain. I have a feeling he's a very busy man at the moment with everything going on with OpenAI. But maybe one day, maybe one day. Our second question is what piece of software could you not live without?
Romain - 00:36:47 So, firstly, I could probably live without software. So I will restrict the scope to "Not work without". And I think it's a bit tough to pick one, but I've been thinking of the core tools that are really tough to replace. There are a number of tools and services we use that I think we could find alternatives or easily swap with another. But I think the core ones probably boil down to your text editor and your cloud provider and something in between that brings the code from your text editor to your cloud provider. And roughly today I started using VS Code. Not so long ago I was on Vim. Still love it, but VS code is really packaging a huge amount of experience without much configuration. AWS. The same story packages. A huge amount of value. It's not the simplest to use, but you can go from very high-level services to very low levels as well. So it's pretty vast what you can achieve with it. And then GitHub in the middle. Same packaging. A tremendous amount of services. Between obviously the Git and the code versioning, but also the dev machines, and code deployments. It's pretty remarkable how they are fitting all of that into one package.
Henry - 00:38:36 I feel like the majority of our guests who have come, have always said GitHub is the main go-to that they couldn't live without. It's just an invaluable software and platform for anyone in the world of tech. The final question we have is "When have you as an individual used data to solve a real problem that you have had"? It doesn't have to be work-related, could be personal-related. But how have you utilized data to solve a real-world problem?
Romain - 00:39:04 So yeah, the real world is a big word. But on the personal side, I like to do my accounting myself. And so I've usually written scripts to download my bank statements from different accounts and you know, pass categorize the transactions, calculate unrealized profits or calculate expenses and income and things like this. Having lived in different countries, you have different currencies and it gets tricky to reconcile all of that. And so I've usually spent probably too much time doing this myself.
Romain - 00:39:51 And I'm yet to find a service actually that provides the same value out of the box.
Romain - 00:40:01 If there is one, I'm definitely happy to switch, but yeah, it's one area, one personal small project I've been doing on the side for quite a few years and I do enjoy it.
Henry - 00:40:17 Brilliant. Brilliant. I will say if you need any beta testers, I would love to have access to the tool, especially with the end of the tax year coming in the UK. It will make life a little bit easier than having to pull them manually. But that is the end of the podcast. That's all we have time for this week. Firstly, I want to thank all of the listeners for tuning in and listening to our podcast. And most importantly, want to thank Romain for being such an amazing host, sorry, an amazing guest, and answering all of our wonderful questions that we had, some of them very in-depth. And we hope to have you again at some point, Romain. Maybe when there are some Wisepops developments you want to share with us.
Romain - 00:41:04 Cool. Well, thanks again for the invitation and it was a good chat. I hope to come back at some point. Yeah.
Henry - 00:41:11 That's great. And again, thank you very much for tuning into Ethical Data, Explained I'm your host Henry NG, and we will catch you next time. Thank you.
Henry - 00:41:19 Ethical data, Explained, is brought to you by SOAX, a reputable provider of premium residential and mobile proxies. The gateway to data worldwide at scale. Make sure to search for Ethical Data, Explained in Apple Podcasts, Spotify, and Google Podcasts, or anywhere else podcasts are found. And hit Subscribe so you never miss an episode. On behalf of the team here at SOAX, thanks for listening.