Exploring Data Science and Entrepreneurship with Mike Greenfield | Glasp Talk #9

Exploring Data Science and Entrepreneurship with Mike Greenfield | Glasp Talk #9

This is the ninth session of Glasp Talk!

Glasp Talk delves deep into intimate interviews with luminaries from various fields, unraveling their genuine emotions, experiences, and the stories behind them.

Today's guest is Mike Greenfield, the founder of Team Rankings, Change Research, a successful angel investor, and a former early employee at PayPal and LinkedIn.

In this interview, Mike shares his extensive experience in the data science and tech industry, discusses the evolution of data-driven decision-making, and offers valuable insights into building successful companies. He also emphasizes the importance of using data to solve complex problems and make the world a better place. Mike advises startups on leveraging data for innovation, understanding tech cycles, and overcoming challenges in various domains. Don't miss this inspiring and informative conversation!


Read the summary:

Exploring Data Science and Entrepreneurship with Mike Greenfield | Glasp Talk #9 | Video Summary and Q&A | Glasp
- Mike Greenfield has a strong background in data science and entrepreneurship, having worked at PayPal and LinkedIn. - He co-founded TeamRankings.com, a profitable sports analytics company, and has started a total of four different companies. - Greenfield believes in the power of data to drive bett


Transcripts

Glasp: Hi everyone, welcome back to Glasp Talk. Today we have an amazing guest, Mike Greenfield. Mike is a co-founder and chair of the board at Change Research. In his career, he has a bachelor's degree in mathematical and computational science from Stanford, back in 2000. He was an early employee at PayPal, where he built the first fraud prevention models, and then he moved on to LinkedIn, also as an early employee. If I remember correctly, Mike was one of the first 10 or 15 employees at the time, right?

Mike: Fifteen, yes.

Glasp: That's amazing. And also, he has invested in many successful companies like Pocket, Opendoor, and so on. Today, we'd like to ask him about his career experience and maybe AI trends, because I think he has seen many tech cycles from tech booms, mobile, AI, and so on. Thank you for joining us today, Mike.

Mike: Thank you for having me. It's great to be talking with you both.

Glasp: First of all, I shared about who you are and what you have done, but could you briefly introduce yourself to people who don't know you yet?

Mike: Sure, yeah. So, I mean, the way that I see my career, in terms of what I've worked on, is I was a data scientist early on, working on data problems in startups, at PayPal and then at LinkedIn. I started a company in college as a side project called teamrankings.com, which is still around and profitable after almost 25 years. After I left LinkedIn in 2007, I wanted to be an entrepreneur and make that my primary focus. I've started four different companies in various domains, from sports analytics to a website for moms, to a public opinion polling company focusing on politics. The unifying theme is the use of data to guide decision-making. I believe data can lead to better decisions and make the world a better place. I've devoted my career to creating data products and using data in better ways to build businesses and improve the world.

Glasp: Amazing, yes. Data is crucial nowadays, and that totally makes sense. When did you realize you were interested in data and data science? Maybe because you went to Stanford, and at the time you were already a data geek and an entrepreneur in your profile, right? Were you already a data geek at the time, or did your interest grow over time?

Mike: I mean, I was sort of a weird kid, and this goes pretty far back. When I was 10 or so, I was reading a book or set of books called Bill James' Baseball Abstract, which was looking at how data is used in baseball. This was like the mid to late 1980s. It was not super sophisticated use of data, but I was fascinated by that, and I got more and more into it. Sports became a significant lens of my exploration of data. It was baseball when I was a little kid and then a teenager. I got more into basketball, and ranking teams and statistics there. Over time, I realized that data is a useful way of solving lots of different kinds of problems. I did a little bit in politics when I was in college and was a political science minor, and across a whole bunch of other fields over the past 25 plus years.

Glasp: Oh, that led you to build TeamRankings at the time when you were at Stanford?

Mike: Yeah, so what I saw, and this was like the late 90s, I was a pretty big fan of college basketball at the time. Stanford had an amazing team, and they've had an amazing women's basketball team pretty much non-stop the past 30 plus years, but they had an amazing men's basketball team in the late 90s and early 2000s. I was pretty into that, and what I saw was the way that teams were ranked relative to one another struck me as sort of arbitrary. It didn't seem like a logical methodology. It's not an easy problem. You look at certain sports leagues, every team plays every other team a certain number of times, so it's pretty easy to tell which is the strongest team. In college basketball, you have 300 plus teams playing 30 games. Some are playing almost all those games against the weakest opponents, and some against the strongest. It's a tough problem to say this team has eight wins and 22 losses but played an incredibly tough schedule, and this other team has 24 wins and six losses but played mostly the weakest teams. Which one is better? I was looking at how to rank teams based on those results and predict games. That's been the story of TeamRankings over time, building more and more on that.

Glasp: That's very interesting. Did you come up with the model by yourself, or did you read some academic papers?

Mike: I mostly learned by doing. I'm not great at reading academic papers and applying them. I understood linear algebra at a certain level and thought maybe this linear algebra approach can apply to ranking sports teams. There were some other additions I brought in from other channels, but it was very much a new system and process to try to rank teams that way.

Glasp: Thank you. Then you moved on to work at PayPal. How did that happen, and at the time, you built the first fraud prevention model, right?

Mike: Right after I graduated from Stanford, I joined a small startup, which was not successful, and I was there for about eight months. While I was there, I found out that my cousin's wife was working at PayPal. I had heard about PayPal through different Silicon Valley channels, and I signed up because I got a $10 referral fee. I was still in college but signed up for this referral fee. I found out that my cousin's wife was working there. My cousin suggested I look at this company, and I decided to interview there when I was looking at jobs in a few different places. I was super impressed with the caliber of the team, and it seemed like an important and useful set of problems to be working on. When I actually started working at PayPal, it took me about six months before I was really effective. It was not a super structured environment, and it was sort of, here's some interesting problems, go figure out what to do. I was 22, about to turn 23, when I joined. I didn't necessarily have the skills to navigate that super well, so it took me a little while to ramp up.

Glasp: How many people were there at the time?

Mike: I think between 100 and 150 when I joined. Probably 20 or 25 engineers when I joined, and I was actually not an engineer. If someone like me who joined there today would be called a data scientist, but that term didn't exist back then.

Glasp: Makes sense.

Mike: There was some pretty solid and impressive fraud detection work going on when I joined. However, the way that different transactions and individuals were being flagged for fraud detection was pretty simplistic. It was very much like, if someone has done this and this and this, then we'll put them in the queue. That worked well for a lot of things, but as fraudsters got more sophisticated, those rules became less effective. I started building decision trees, looking at individual transactions to predict the likelihood of fraud. We had a good dataset of millions of transactions and thousands of confirmed frauds, so I ran my decision trees on those. It seemed promising, but the problem was a little bit unspecific. The breakthrough came around six months after I joined. There was an intern working on a specific type of fraud, Merchant fraud. He was looking at Merchants who had just received their $2,000, trying to determine if they were fraudulent. He had an Excel spreadsheet of 10 different things, manually figuring out if they were bad. We decided to apply my decision tree software to his Merchant fraud problem. We created a richer dataset and a better predictive model, which worked well. We then gave that to agents and scaled the process over the next three years. It involved making the decision tree software more accurate and robust and applying it to various types of fraud.

Glasp: Awesome. You spent about four years at PayPal. When you joined, there were 100 people, and by the time you left, it seems like PayPal was growing like crazy. How was the experience, and how did it grow?

Mike: It grew quickly. We went public 14 months after I joined, and eight months after that, we were acquired by eBay. That changed the trajectory completely. A little less than two years after I joined, we were acquired by eBay. It probably went from 150 to around 600-800 people by the time we were acquired. Being part of eBay changed the dynamic completely.

Glasp: Also, PayPal got merged with X.com, which is kind of Twitter now. They were in the same building, right?

Mike: That was actually before my time. PayPal and X were in the same building in 1999, about a year before I joined. They merged in early 2000. When I interviewed, the company was called X.com, and when I signed my offer letter, it was still called X.com, but it was in the process of rebranding to PayPal. I never worked with Elon Musk at PayPal. He had already left as CEO, so I missed that phase, which I gather was interesting and a little rocky.

Glasp: Interesting. Did you work with Max Levchin?

Mike: Yes, I worked directly with Max when I first started and as part of his broader team most of the time I was there.

Glasp: As K and I are building a startup, we wonder how the culture was at PayPal. What differentiated it from other startups?

Mike: PayPal was intense, no-nonsense, and focused on solving problems quickly. It was hard-driving, with no tolerance for unproductive meetings. The mentality was to get decisions made as quickly as possible and solve problems. PayPal was a tough business to build but succeeded through great execution.

Glasp: And then you moved on to LinkedIn in 2004, right? How did that happen?

Mike: Yes, in 2004. At PayPal, we had pickup soccer games once or twice a week. One of the people I played with was part of the founding team at LinkedIn, Lee Hower. I knew about LinkedIn through Lee. We had lunch one day, and I heard about LinkedIn, which had launched a couple of months prior. I started moonlighting at LinkedIn in early 2004 while still working at PayPal. LinkedIn had about 250,000 users and 15 people on the team. There were a ton of interesting data problems, and I liked the openness and ambiguity. After a few months, I decided to leave PayPal and join LinkedIn full-time.

Glasp: What kind of data problems were you solving at LinkedIn?

Mike: LinkedIn was the first social network dataset I worked with. I analyzed connections between people, usage data, and how to facilitate more connections. We looked at where people were dropping off during sign-up and how to improve the flow to increase user engagement.

Glasp: Do you remember any effective strategies for increasing user engagement?

Mike: Yes, we built the reconnect product, which suggested people who worked at the same company at the same time. Later, we launched "People You May Know," using various signals to suggest connections. Finding interesting emails to engage users was also effective.

Glasp: How was the culture at LinkedIn compared to PayPal?

Mike: LinkedIn was less intense and more product-driven. It was about building towards a visionary product, while PayPal was more about hard-driving execution. LinkedIn was the manifestation of Reid Hoffman's product vision, focusing on professional networking and strategic relationships.

Glasp: Why did you decide to leave LinkedIn and start your own company?

Mike: I enjoyed the speed of execution at TeamRankings, which I was running on the side. LinkedIn was slower-moving. I wanted to drive something forward as an entrepreneur. It took me about six months to figure out what company to start after leaving LinkedIn.

Glasp: During those six months, what were you doing?

Mike: I met with potential co-founders, consulted with startups, and explored various ideas. Eventually, I teamed up with Ephraim Love to start Circle of Moms.

Glasp: You have founded several companies. What advice do you have for aspiring founders on choosing co-founders and ideas?

Mike: Trust and shared vision are crucial. I've known most of my co-founders for years in different contexts. It's important to go through difficult challenges together before starting a company. Discuss values, priorities, and potential conflicts in advance. For ideas, prototyping and iterating can help refine concepts. Founders should balance learning and delegation based on the stage of their company.

Glasp: How about learning and delegation for founders?

Mike: Founders should focus on learning crucial aspects like sales and product in the early stages. Delegate tasks that don't add significant learning value, like ordering lunch. Over time, delegate more as others can handle tasks nearly as well as the founder.

Glasp: Do you have any advice for aspiring founders and professionals?

Mike: Discipline and steady improvement are often underrated. Creative brilliance is important, but consistent practice and skill-building make a huge difference. Be disciplined in identifying and developing necessary skills.

Glasp: Do you have a favorite quote or motto?

Mike: I don't have a specific quote, but I have a mission statement I review and revise every Friday morning. It helps ground me and keeps me focused on the kind of impact I want to have on the world.

Glasp: That's impressive. Finally, what kind of legacy or impact do you want to leave behind?

Mike: I hope to solve specific problems with data and make the world a better place. Data can make people happier, more productive, and informed. I want to inspire others to think bigger about using data to solve problems and foster positivity and possibility in making the world better.

Glasp: Thank you for joining us today, Mike. We really enjoyed your talk and experience.

Mike: Thank you, K and Kazuki. You guys are building a cool and exciting product with Glasp. It's been fun and inspiring for me. Thank you.


Follow Mike Greenfield:

Twitter

Share Your Excitement

Ready to highlight and find good content?

Glasp is a social web highlighter that people can highlight and organize quotes and thoughts from the web, and access other like-minded people’s learning.

Start Highlighting