Evan: Hello and welcome to the Mining Your Own Business podcast. I’m your host, Evan Wimpey, and today I’m super excited to introduce Eric Sims, who’s joining us on the show. Eric is a senior data analyst at LendingTree. Eric, thanks so much for coming on the show. Excited to chat with you today.
Eric: Yeah, glad to be here.
Evan: Fantastic. Hey, to get started, can you just give us a little bit about your personal background and how you got into data analytics?
Eric: Sure. So like a lot of people I have, you know, I have a background that didn’t start in data. It’s pretty common. I have worked in things ranging from academia to local government, manufacturing, management, all sorts of different stuff, and I like to say that I find data or data finds me kind of wherever, wherever I ended up going. And one day a few years ago, I was having kind of an informal job interview for a marketing role, and I was talking to a guy who. We quickly realized that the role wasn’t a good fit and it wasn’t a big deal.
But then for some reason, he decided to take an extra few minutes of his time and just asked me like, well, what is it that you like? And I was going on about data stuff, this, that and the other. And he said, you know, data’s a thing. You can do that, like marketing operations or sales operations. And that was a huge – that was a career totally impacted, impacting moment for me.
And I am so grateful to that guy that he took that time to let me know, because otherwise, it wouldn’t have put me on the path that I’m on today. And so, yeah, so that’s, and how I ended up getting, finding out data was a thing. And then deciding, okay, well I’m gonna learn some, start teaching myself Python on the weekends, take an online stats course, see if I actually like it, you know, Coursera, before I decided, okay, I like it enough, I’m gonna do it. I’m gonna invest in doing a master’s degree and. So that was a very worthwhile investment. I don’t think that everybody needs to do it, but it worked really well for me. And then I got my job at LendingTree and here we are.
Evan: All right. Wow. That’s fantastic. It’s funny what just a small conversation, you don’t realize the impact that it can have. I think that’s probably the case with a lot of folks. I’m thinking of Korri Jones, who’s at Chick-fil-A, does data engineering work there and he had a similar story just sort of happened into a conversation that, that landed in where he is today.
So you are at LendingTree, big financial institution. It’s pretty big. You’re probably not just the data analyst at LendingTree. Can you talk about where you sit in that organization?
Eric: Yeah, so strategy and in and analytics is kind of what encompasses Yeah. Pretty much all the analysts, all the data analysts, and then also our small data science team, and that currently sits underneath the CFO, but I’m about as far from a finance and accounting person as I could be.
And so we kind of have, as far as analysts go, every analyst usually supports one or maybe two different verticals. And so LendingTree, we do mortgages, we do small business loans, auto loans, student loans, all sorts of stuff. I support small business and I love it. And so I have on my side, I have a counterpart who also helps support small business. She works with the sales team and is more outside-focused and I am more inwardly-focused. I work with the product team and the marketing team, the GM and who knows what else stuff comes up. And so that’s kind of how we divide between analysts the work and, you know, there’s definitely still some overlap which is nice when you need somebody with a little bit of a similar domain expertise but we’re not like stepping on each other’s toes, but it’s good for spit balling.
Evan: Sure. Okay. Yeah, that paints a good picture. Maybe, maybe we could dive a little deeper and can you give us a flavor of the type of project that you would work on maybe with that marketing team, with that product team?
Eric: Yeah, for sure. So, let’s see here. Some of the product projects that I’ve worked on slash I’m working on that are most interesting.
Routing is a big part of my role. So when somebody comes to our page and they click, I Want a business loan, they start filling out the form. They can go either to experience A or experience B, and deciding which of those routes is going to be better is really important because experience A can hold large capacity experience. B only has a very small capacity, so we really need to spend it really wisely. And then, so there’s that piece. And then understanding also working with like the marketing team to understand when we bring traffic through from different channels, who’s getting sent to experience B, cuz that one is more valuable but also limited. So are we, are we driving, are we maximizing that? And then on experience A side, once somebody goes to experience A, we have kind of A1 and a A2 as well that they could go to. And those are governed by different mechanisms models, mechanisms, whatever you wanna say to decide what that final routing will look like.
So, that actually encompasses marketing cuz they’re the what’s flowing and product because they’re the pipe for the whole thing. And so that’s been cool to implement that over the past year. We’ve made a lot of really good progress with how we route optimally.
Evan: Awesome. That’s super exciting. And I like that you have an example that touches on product, on marketing, and you seem a bit like a bridge there – at least you’re connected both ways. I don’t know if this is a typical project, but how would something like this come to fruition? How does it get legs in the first place? Is marketing coming to you and say, we need this? Are you using your creative ingenuity to say, Hey, you guys need this?
Eric: So this, that particular project came from product. A couple weeks after I started at the company. The product manager said, Hey, we have this, this system that we developed, that it’s there and it’s running in the background, but we’ve never actually checked to see if it works. So we’re not using it. And so, You know, could you look into that? Cuz we think it has some potential. And so I started looking into it and found, yes, it does have some potential. And so from there it became, Well, it became a really big project because at first it wasn’t working correctly and so there was a lot of work – I got to work – it was a great introduction because I got to work with different developer teams in different parts of the company that normally I wouldn’t really, I don’t normally work with them, but on these bigger projects I had a reason to. And so it was a great way to like get some good relationships and at least a very basic understanding of what these different teams do and be able to bring that to the GM to say, this is the profitability impact that this is going to have. Do I have your blessing? And the answer was, heck yeah. And so we, and so that’s how we ran with it. So the idea originally came from product.
And it’s funny, like sometimes I feel, I feel…I don’t know if “guilty” is the right word, but sometimes I feel kind of guilty, like, oh, like there was this great project, but like, I mean, I didn’t have the idea to do the project. Does it still count or it’s like, does it still count as if we’re like keeping score of who did good enough work or whatever? I don’t know. I can’t be the only one who ever feels that way. But yeah, so I do feel really proud of it though because I think about, think back on it, it’s like, well, if I wasn’t doing it, it wouldn’t have gotten done. And so it’s okay because that’s why we’re a team. Because one person doesn’t have to do the whole thing from you know, soup to nuts. And so anyway, that’s kind of a side point on it. But I was really glad that the product team brought the idea up and that I could run with it.
Evan: Awesome. Yeah, and I don’t know if I would characterize, yeah, I’m slipping around on the word guilt too, but thinking of my own experience maybe “regret” is the thing I – “why I should have been able, why didn’t I think of that?” Of course, now that I’ve done some work here, I should have been able to realize this was a good place to apply some analysis.
Eric: Yeah, definitely.
Evan: So you sort of got an idea from product. It sounds like you did a lot of the legwork or at least part of a lot of the legwork and then you go to a GM and say, Hey, here’s some expected return on this type of work that that’s probably not just to snap your fingers and you’ve got that. And I think this is a challenge probably for a lot of organizations. So I don’t know if you can put any flavor to that or really just what are your biggest challenges when you’ve got some ideation and you want to turn it into a project and actually kick it?
Eric: Yeah, so LendingTree is probably the most complex, it’s the most complex organization that I’ve ever been a part of. I’ve been, and I, you know, there are more complex organizations for sure. But generally, I mean, my last company. We, there were like 30 people and so, you know, I could do anything. I was the Salesforce administrator, I was God. And so it was great.
Whereas at Lending Tree I have, you know, like a very small set of permissions and there are like different little things to think about for everything. And sometimes it’s frustrating because it’s like, oh, look at this. Let’s just say it’s a hundred thousand dollar opportunity. And then by the time you’ve weed-whacked off all of the little caveats, it’s like, how about a $23,000 opportunity? You know what I mean? And so when it comes to figuring out how are we gonna get this thing off the ground? Yeah, there was all the legwork of saying, can we get this process to where it’s working and validated? And then from there, collect data to have enough past data that’s just kind of been running in the background to then present and say, can I now put some resources behind this to, it’s kind of already deployed, but integrated I guess into our workflow. And this in this particular case, one, because it’s dealing with routing and two because pretty significant financial thing, and I was talking to my boss and then my boss’s boss, and then like the head of sales and the GM, just like making sure that everybody’s like, okay, yeah, we think this is worth it. And so it was really good because I have a deck that I made that has some slides in the appendix that are very detailed, but otherwise, we’re just starting at the top with this is the big number that we’re shooting for here and here’s some important details that you might be interested in, but otherwise you know, ask me questions. And so just kind of have like this one go-to way to be able to talk about it and, and make sure I was getting buy-in from everybody so that yeah, we’re all on the same page. We think this is valuable, and we’re happy to dedicate the resources to it. And so I’m grateful to back now and see that it did work. There wasn’t any time, I think when it was touch and go as to whether we were even gonna implement it because fortunately we were able to prove early on that it would be worthwhile. But that doesn’t mean it all happened fast. That’s the other thing I’ve learned is, stuff just takes way more sprints than you think it’s gonna take.
Evan: Yeah. I think that will resonate with anybody who’s listening that has worked on an analytics project, not just the technical data build, data model deployment, but really that sort of, that front-end effort that you just talked about on generating the buy-in. Sort of trying to ready the organization to absorb some change or some new implementation.
Eric: Yes. So, so true.
Evan: You’ve talked to, you know, you’re getting, I love hearing you’re getting stakeholder buy-in and you’re getting it from a lot of different people. You work in data. We, data folks, have a particular vocabulary that we use. And it sounds like it’s great. You’ve got a bunch of appendix slides with maybe a lot more detail. Can you talk about any challenges in trying to convey your data terminology into the LendingTree world? And I guess conversely it’s probably like this in any industry, but in my head it’s a lot harder in finance the terms, the acronyms, how do you keep up to speed with all of the fancy things that your finance counterparts talk about.
Eric: Yeah. So I don’t think I struggle too much to get analytics terms into business terms, you should ask my stakeholders. They would have a better, more accurate, valuable opinion on that. But I can definitely say we have a lot of acronyms at Lending Tree a lot, and we even have a periodic table of acronyms, which I love. It’s kind of tongue-in-cheek. It’s way fun.
And it’s funny because there are certain acronyms. CPL stands for cost per lead. CPL means a different thing if you’re talking to a marketer than if you’re talking to a salesperson. And in fact, they mean opposite things. And so, from a marketer, we’re saying, oh, our cost per lead, how much did we pay to get the lead on the sales?
They’re like, cost per lead. How much did they pay us for us to sell them the lead? And I’m like, that’s revenue. But if you call it cost, you call, you know, whatever. And so that’s definitely a learning curve. And so also when I’m talking to my counterpart who works in sales analytics, we have to make sure we are talking about the same acronym, acronym even, and we’re both analysts.
So that’s, that’s definitely a big thing as far as getting analytics to be accessible, whether it’s, I think it’s less of a terminology thing in, in my case at least, and more of a – like I can throw a table of a bunch of numbers and decimals and dollar signs at Tableau and just say like, can’t you see?
There’s, there’s brilliant insight here. But it’s really a good, and I think really worthwhile challenge to lean into making it as clear and simple as possible. You know, like fewer labels. more meaningful colors, you know, fewer shapes if possible, you know, whatever it is, so that the meaning can be just like so clear, even just like making that the title of the chart. And so I think that’s for me the biggest thing for making analytics accessible is if I’m presenting it, how do I just spell out exactly what I’m trying to say on the slide in, in pictures instead of in too many words.
And then the other piece is self-service. Like I have been working with the product and marketing leads that are closest to me to show them like, here’s how you use Tableau. And now when you have a question about something like what was our close rate last month for one of these experiences, you can make that up in three minutes, whereas you’d have to ask me and like, I’ll get to it when I have some time. And so I really like teaching and showing and sharing. Here’s how you can do it. Ask me when you forget later. And we’ll just keep practicing through it. And I’ve seen some good, really good progress and results.
Evan: Awesome. Yeah, that’s great. Teach a person to fish there. I’m curious, and I think you’ve, you have some background in teaching as well?
We do some teaching work, and I think there’s nothing, you know, you build a great model. You show a great insight. But I don’t know that there’s anything more rewarding than seeing sort of a light bulb come on and somebody like really understands. I, you know, not to put you on the spot with this, but do you have any light bulb moments? Do you have a favorite, you see somebody that understands the analytics or the insights in a way that they maybe didn’t before?
Eric: Oh yeah, for sure. A few months ago, September, we rolled out a new feature. For that, you know, a one to a two experience piece, routing piece. And it uses in this case it uses a predictive model, a lead score model.
And we used to do well. Before that, we basically had something that said like, did you come from this marketing channel? Go there. Otherwise, go to the other experience. So something really, really basic business logic. And then suddenly now we have this model, which is something that people haven’t seen and are less familiar with and has 11 or 12 different parameters to it. And trying to explain that some are more influential than others, and it was just, it was just challenging to get, to help people see what it was accomplishing and to trust it. And so I came up with an analogy using M&Ms because we can all relate to M&Ms course and so I was essentially just saying we have, let’s assume that green M&Ms are more valuable than all the other M&Ms, all the other colors of M&Ms, but you can’t see inside of a package of M&Ms. And so explaining how this was going to. Oh, and different channels or different M&M suppliers sometimes have more green M&Ms than other ones.
And so if we just cut it one by on a single variable like channel, like that may or may not work or may not hold accurate over time. And so how can we use this model? And so anyway, long story short. Using M&Ms made it super clear. Somebody like was happened to be snacking on M&Ms that afternoon and like posted about it in Slack that they were doing, you know, company research and people have talked about it since, you know, even weeks later.
And so even if it didn’t, even if people weren’t like, oh yeah, XG Boost now I totally understand. It’s like, oh, this lead score model. Like I feel just more comfortable with it because now it feels a little bit more accessible and so, I’m trying to get that lesson put into Confluence so that one, I can remember it later and two other people can reference it. You know, anytime. Anytime there might be questions or perhaps they can repurpose it for their own vertical.
Evan: Ah, also, that is a great story. Yeah. I think that that makes it very accessible. And hey, now you, you’ve got an artifact here. You can put this, put this podcast recording there in Confluence to come back live on forever. Yeah. I like that story a lot. I’ll try to remember to reference you when I continue to tell it.
Eric: Get as much mileage as you can.
Evan: I wanna shift gears a little bit away from Lending Tree. Eric, you’re quite active on LinkedIn, which is great, and I just, well, I was about to speak for the data community. I don’t have the authority to speak for the data community, but I’ll speak for myself. I really appreciate all the content that you share, that you post. Super insightful, super thoughtful stuff. I don’t think I’ve seen the M&M story on there but a lot of ways to, tips for conveying data, tips for thinking about data, tips for communicating data.
And one of the things that I really like to follow along is you chat a lot about some side projects that you’re working on. Some of that is teaching other folks. Some of that is asking for help, which I think is, is great. I’m curious how you, how you decide what kind of, you know, a lot of people trying to break into the field or trying to learn and grow in the field, you know, that that’s something that we hear a lot. You know, do a side project and then you’ll learn. Right. How do you choose a side project and what do you try to learn from it?
Eric: Yeah, so first off, thank you for reading my LinkedIn posts cuz you never know you’re putting it on a little paper boat on the ocean. So side projects are, I love side projects. So there are two pieces to it. One is some people – one is coming up with ideas for the project, and then the other one is choosing a project to do and actually following through and getting it done right. So I’m an ideator. I love thinking of stuff, but I also really love trying to think of things that are original, a little different, and a little different in like, kind of a surprisingly fun way. And so, other people are not like that. Other people want to think of something that’s like very serious and very real world applicable. Totally awesome. Just doesn’t get me excited. Right?
And so when I’m trying to think of projects, I try to think like, okay, like what’s something that would be valuable to me? Like I’m interested in, I’m interested in recommender systems. Okay, well I like recommender systems. And then, oh, I see this job posting. It’s about, it seems to. Like, I seem to have all these different qualifications for it, but it’s talking about N L P. I wonder what I could do that would get me some N L P stuff. Okay, well, I’m not really that interested in predicting stock prices, so that’s not gonna be on my, on my radar. And so it’s kind of helpful to like whittle down some topics and then I just try and keep things on my mind slash I have a Google Keep note where I write ideas. And if I’m really interested in it, I’ll start to like drop in links from Stack Overflow or media articles and really things where I’m like, I have no idea how to do that thing, but I have a rough idea that it’s probably, you know I’m gonna need spacey and maybe this transformer model, blah, blah, blah, blah, blah. And I put all those things together and then at that point I’ll say, do I have a way that I can do it? That would be fun.
So a couple of examples, about a year ago I read a paper, this is kind of related to LendingTree. I read a paper about how LinkedIn uses naive Bayes to catch spam accounts based on just their name. And I thought, hmm. We have people that just basically give us their name at LendingTree. I wonder if I could make a little prototype model of this and cuz I’ve, never done anything with naive Bayes. I didn’t know how it worked. I wanted to try it. That’s actually what started the whole thing. And so I built a little version of it and shared it with the GM and it was cool and all, but it just wasn’t a priority right then. And so it kind of got just put on the shelf for a while, no big deal. But I still wanted to share it cuz I wanted to get mileage out of it. And so I thought, how could I repurpose this? I need some way to classify you as something or not as something, and it needs to be something preferably interesting. Ah, St. Patrick’s Day is coming up. Why don’t I take people’s names and see if they’re Irish or not? And so I just like took names of like Irish people in Parliament, something I could just scrape from Wikipedia really easily. And then I found surnames from all these different countries around the world. And I created this little model and built a little stream lit app that now people could go to put your name in, it’ll give you a score and tell you if you how much beer you should drink for St. Patrick’s Day or whatever, right? And so that was because I wanted to learn Naive Bayes.
Several months later I was like, I wanna do some N O P stuff. I have this idea. I think it, I saw a guy… So I got inspiration from another project. This guy said, I’m gonna train a model on 50,000 fortune cookies and I’m gonna make a model that will create fortune cookie messages. And I thought, what if I could turn something into a Fortune cookie message, specifically a movie plot? Because if I could say, you know like take Toy Story, like a toy cowboy is threatened when new space toy shows up. It shows up in their toy box or whatever. Well, what if I could just have it say, “you will feel threatened when a new space toy shows up in your toy box” as a fortune cookie. And so I got to figure out how to do that. I’d never seen anybody who had converted – I couldn’t find anything online of like a clear project of taking something from the third person, converting it to the second person and doing it in a way that was easy or something. And so it was like, this will be a fun little project to work on. And so that’s why I did, and I always like to try and build out if I can some kind of a little application so that you can go and you can play with it, you know, and test, you know, like I set up a quiz with like the movies to say like, can you guess which movie this is? Or whatever.
And so I like to, and so my current project now is I like tea. And so I found a social network based all about tea. And so I’m trying to one, get enough data to work with and then two, create some sort of a recommender that will allow you to choose, I would really like to have like a little dial of like, do you want stuff that’s like, keep it familiar, do you want it a little bit interesting? You wanna walk on the wild side and get some reviews that are gonna be like really way out there and test your tastes and stuff like that? So I think the most important thing when it comes to you have a list of projects you wanna do, pick one you like, just pick one that you’re interested in because otherwise it’s gonna like even if you are working on something you like. Sometimes it’s a slog. And so you gotta want it.
Evan: Yeah. That’s great. Those are great examples and I think that’s a great sort of framework for how to go about it. And it sounds like even if you’re explicitly steering towards interesting versus applicable, it sounds like you’ve learned, like you’ve gained, like you have some new skill sets that you wouldn’t have had had you not worked on those side projects.
Eric: Oh yeah, a hundred percent. I could at least have now a heading if I wanted to get into applying it in a business setting. Like definitely. But I enjoyed the process more than I probably would have otherwise.
Evan: Awesome. Awesome. Eric, I wanna ask you one last question. We’ll bring it back to Lending Tree, but thinking about this sort of ideation, creative, what are interesting problems to work on? I wanna press the magic button here. All the Lending Tree stakeholders are on team Eric Sims. They’re aligned to your vision. They want to follow whatever creative idea that you have come up with. What’s something that you would want to dedicate analytics resources towards to work on at LendingTree?
Eric: Hmm. This is such a hard question.
Evan: You could recommend tea to Lending Tree?
Eric: I could. Yeah, absolutely. Lending Tea. I wonder, to me, I think the most valuable thing to me, and I think to any company, and that’s a pretty big statement, but I’m pretty confident in it, is knowing your customer – and knowing not just your customer’s data, but like knowing your customer’s voices and like their problems and all of that. And so I’ve been thinking about it recently and I messaged somebody last week and just said, Hey, can I like either listen in to some calls or like from our call center. Cause we have a lot of people in call centers or can I listen to recorded calls or whatever. And, so I’ve actually done some of both, just recently and it has been really nice and really, really interesting.
So, this is not a great analytics answer, but maybe that’s because I’m not a super huge strategic analytics thinker or analytics strategy thinker. I really think that if people know our customers and we all know our customers the same way, then we’ll all have the ideas that we need to grow our respective areas of the businesses. Because otherwise, my ideas for analytics would be really great for small business and they’d probably be really fun for me. But they definitely wouldn’t help our stock price or, you know necessarily make our customers’ lives better. So I would say something about like, anything to like, really everybody take a week, doesn’t have to be the same week, take a week, listen to customer phone calls. You know, I think that’s what – you know the company Automatic, they make WordPress?
Evan: I’m not familiar.
Eric: No. Okay. Automatic. They make WordPress. Everybody from the CEO on down spends a week every single year working the call center. Not just listening to calls, but actually working in the call center and answering questions. So you know, the customer, I love that. I think it’s so cool and I think it would be so hard but also really like helpful in staying connected to the mission of the. So that’s what I would do if I had all of the control.
Evan: Awesome. That’s great. That’s a really thoughtful answer, Eric. I appreciate that. And I think there’s certain, there are teams, not I didn’t know about Automatic, but there are certainly teams that try to do that. I don’t know, to the scale of every single year everybody does it. But yeah, I think that’s a really great and thoughtful idea. Eric, thanks so much for coming on the show. You’ve had a lot of thoughtful and great ideas. Very happy that you’re ableto chat through them today. Folks, please follow Eric on LinkedIn. And there, he’s an idea machine. The interesting ideas just keep coming up. Eric, is there anything else if folks wanna follow along on your side projects or anything else, is there anywhere else they should follow you or look for your work?
Eric: LinkedIn is the best place. I mean, I’m also on GitHub, but you can find my GitHub on my LinkedIn. It’s Eric postmaster, capital E, capital P, capital M on GitHub, but otherwise LinkedIn. Oh, and you know, if people have or are looking for or wanna chat through project ideas, message me. I love to chat about it.
Evan: All right, perfect. I hope your inbox explodes with interesting ideas here, and if you think the types of things he’s doing at Lending Tree are cool, Lending Tree is hiring. We will post some links in the show notes here. If you’re interested. you can check those out. Our guest today, Eric Sims, senior Data Analyst at LendingTree. Eric, thanks so much for coming on the show.
Eric: Thanks, Evan.
Today we welcome to the show Eric Sims, Senior Analyst of Strategy and Analytics at LendingTree, an online lending marketplace.
