the ConversAItion: Season 5 Episode 30

How Redfin Uses AI to Help Buyers Find Their Dream Home

Redfin is a real estate brokerage that combines AI-powered insights and agent savvy to make the home buying and selling process easier, faster and less stressful. In her ten years at Redfin, CTO Bridget Frey has played a major role in shaping the company’s powerful recommendation engine. This week, she joins us on The ConversAItion to unpack the role of AI in real estate, the impact of the pandemic on home buying and selling and why she tackles the issue of diversity in the workplace by thinking of it as an engineering problem. You can find Bridget on Twitter @svbridget.
Listen on Apple Podcasts badge
“You’re getting hundreds of pieces of data about everything from, is there a garage? To, how many bathrooms? But, you also have information about what it’s like to live in an area: what’s the neighborhood like? What’s nearby? Tax records, all sorts of things. And for someone who’s interested in AI, it’s a dream to have all this data to crawl over.”
Bridget Frey

About Bridget Frey

Bridget Frey is the Chief Technology Officer at Redfin, where she leads a team leveraging AI to power a stress-free home buying and selling experience for users of the full-service real estate brokerage. Bridget joined Redfin in 2011 after working as the Director of Engineering at Lithium Technologies. Prior to that, she held technical leadership roles at IntrinsiQ, IMLogic and Plumtree. Bridget earned a bachelor’s degree in computer science from Harvard University. You can find her on LinkedIn here and Twitter here.

Short on time? Here are 5 quick takeaways:

  1. In real estate, combining human insight & algorithms can drive better outcomes.

    Bridget and her engineering team got an early lesson on the importance of combining agent insight and recommendation AI. Initially, the team built an algorithm that suggested houses for customers based on their preferences, but it didn’t prove to be too popular when it came time for agents to use it. As it turns out, the algorithm didn’t allow for them to manually add or remove a home from that list based on personal insight—an oversight by the team. 

    In truth, agents were the ones who truly knew what their customers wanted—whether it was a specific type of build, a property by the water or another detail that an algorithm couldn’t suss out. Ultimately, human relationships are key in a high-touch field like real estate, and while algorithms can augment those relationships, they can’t replace them. 

  2. Data is only meaningful if it helps produce a tangible positive impact on people.

    Growing up, Bridget moved frequently, living in seven homes across five states before she turned 18. Each time, she witnessed firsthand how challenging the buying and selling process was. At the time, there were no online tools like Redfin—so her parents had to trust agents who they’d just met, or who weren’t giving them all the information, with one of the biggest decisions of their lives. As an adult, Bridget joined Redfin to make this process easier. 

    Today, as the number one real estate brokerage website in the US, Redfin has access to troves of data on both sides of the business—the customer side and the agent side—on a national scale. But for Bridget, given her own experience growing up, it’s critical that companies don’t just find correlations in a dataset, but actually use those correlations to ensure that customers have an enjoyable, stress-free home buying experience. Accordingly, by leveraging this data (which homes people are clicking on, scheduling tours for, making offers on and winning), Redfin’s algorithms are more effective than typical search terms (two bedroom, two bathroom, etc), helping home buyers find the right home, and sellers make more money, faster.

  3.  The rise of work-from-home caused an uptick in (virtual) home buying and selling.

    Redfin saw two key real estate trends as a result of COVID-19. At the outset of the pandemic, people pressed pause on their plans to buy and sell their homes—but as the months progressed, they began to see the value in working remotely long-term, and started reevaluating those paused plans. In fact, Redfin saw a significant uptick in the number of people searching for homes outside of where they lived—going from a quarter to a third of searches. 

    Similarly, there was a newfound interest in virtual home tours. These virtual tours were of interest to just 1% of customers pre-pandemic, but quickly became appealing to almost a third of customers. Redfin was tasked with rapidly adapting to these new trends and figuring out how best to help customers in the new normal. 

  4. Companies can—and should—improve diversity by treating it like an engineering project. 

    Bridget likes to think of diversity in the tech industry as an engineering project. This may seem like an unusual approach, but engineering teams are built to solve incredibly complex and challenging problems. If there are experts figuring out how people can live on Mars, Bridget believes they should also be able to improve diversity in the tech industry.

    While the issue may seem overwhelming, if we collectively think of it as an engineering problem to solve—with a mix of both large and small problems that simply need to be broken down—it becomes easier to work on. For instance, Redfin has started using the same bug-tracking software that’s used for identifying internal software problems to also identify changes in how they recruit, train, promote and pay—and it seems to be working. When Bridget started at Redfin ten years ago, she was the only woman on the Seattle engineering team. Today, 36% of the team are women (making it one of the most gender-diverse teams in the tech industry), and 10% are Black or Latinx. 

  5. There’s still ample opportunity for AI in real estate. 

    Going forward, Redfin is looking into leveraging new algorithms and data sets that help customers glean information about a property just from photos online (What’s the condition of the home? The price? What is it like to live there?). What’s more, additional AI may even help identify factors like pollution level and other things going on outside the home itself. 

    Bridget feels optimistic about the fresh innovation that’s bound to come in the next few years. Case in point: today, Redfin uses the “Redfin Estimate” algorithm, which helps customers determine the price their home will likely sell for. Just a few years ago, this technology would have been cost-prohibitive, given that there wasn’t the same cloud technology and algorithm speed we have today. 

Read the transcript



Jim Freeze Hi! And welcome. I’m Jim Freeze, and this is The ConversAItion, a podcast airing viewpoints on the impact of artificial intelligence on business and society. 


On today’s episode, we’re joined by Bridget Frey, CTO at Redfin, a real estate brokerage firm that combines AI powered recommendations with agent savvy to make the home buying and selling process faster, easier, and less stressful. Redfin analyzes buyer behavior from which homes people click on to which offers are successful, to drive a powerful recommendation engine. As CTO, Bridget has not only shaped this system, but also built Redfin’s big data platform from the ground up and launched new products like Redfin Estimate, and Redfin Mortgage. Along the way, She’s built one of the most diverse technical teams in the industry and become an advocate for how others can do the same.

Today she’ll walk us through it. From how AI helps buyers find the home of their dreams, to why she approaches diversity the same way she would solve an engineering problem. Bridget, welcome to the conversation.

Bridget Frey Thanks so much for having me.

Jim Freeze Yeah, we’re thrilled to have you. I read that you got a very early start on your technology career, getting all the way back to age five. Can you talk a little bit about your passion for tech, and how that’s guided your journey from your time at Harvard into the business world?

Bridget Frey Absolutely. So my dad is an appliance sales and repairman, and when I was five years old, he showed up in our house with this big box and inside was an Apple IIE, an early apple computer. And to him, it was like a dishwasher where you get the manual and you try to figure out how it works. And so those were my earliest experiences with technology. Just me and my dad figuring out how to write code. And so I became just a lifelong lover of technology, somebody who was looking for ways that technology could change the world. I studied computer science at Harvard and then have worked at growing companies ever since.

Jim Freeze Wow ,t’s interesting. I remember the first computer I got, and the really awful documentation for some of their software. And it was kind of like you had to figure out on your own because [crosstalk 00:02:21] so bad. So I know exactly what you’re talking about. It was kind of fun actually to try to figure out how to get it to work.

Bridget Frey Back when hacking was hacking, right?

Jim Freeze Yeah, exactly, exactly. So, as I understand it you worked with a Redfin agent to actually buy your home in Seattle before you officially joined the team. And obviously you eventually became the CTO. For our listeners who don’t know, can you talk a little bit about how Redfin works? What the process is like?

Bridget Frey Yeah. So we’re a real estate brokerage that was founded by a technologist. So we use technology to make everything about buying and selling a home just easier, less stressful, and more automated. And we have full-time real estate agents who work for Redfin and collaborate with our engineers to try to find the best balance of humans and computers in helping people get into their dream home.

Jim Freeze That’s great. I’m a bit of a real estate enthusiast, so I love all the online tools that exist. 20 years ago, I remember the old books you used to look through for listings of homes. So it’s amazing how the tech has advanced, and how the industry has advanced. What about the Redfin experience drew you to the company? What was the opportunity you saw, to grow the role of technology in the home buying and selling process?

Bridget Frey Yeah, so I actually moved a lot when I was a kid. I lived in seven homes in five states before I turned 18. And every time we moved, I got to see firsthand what my parents had to go through to try to find the best place for us to live. And as you mentioned, there were no online websites back then. It really was about which agent you knew, and developing those relationships. And you just always felt like you didn’t have the information you needed. You weren’t sure who to trust. And you’re making some of these really momentous decisions because where you live has such an impact on your life. So when I had this chance to join Redfin and combine my love of technology with this problem that I had experienced growing up, it just really connected for me.

Jim Freeze That’s terrific. It’s been, I believe, about 10 years since you’ve joined. What role does AI take in the Redfin experience? How’s it grown and evolved over time?

Bridget Frey Yeah, so I actually joined Redfin to found our Analytics Engineering and Big Data teams. So I was one of the first to bring Big Data and Machine Learning into the company. That’s a particular area of passion for me. And I think really it all starts with data. In real estate, there is a lot of data. You have information about homes for sale. Often you’re getting hundreds of pieces of data about everything from, is there a garage? To, how many bathrooms? All kinds of things. But you also have information about what it’s like to live in an area. So what’s the neighborhood like? What is nearby? Tax records, all sorts of things that go into this. And for someone who’s interested in AI, I mean, that’s a dream to have all of this data to crawl over. Redfin also has a really unique data set because not only do we have the number one real estate brokerage website in the United States. We have all these people who are coming and looking at homes and interacting with our site, but we also have a national real estate brokerage.

So we have agents who are actually helping you buy or sell a home. And we’re really the only company that has both sides of that business at a national scale, which means we have data, not just about what homes you’re looking at, but did you make an offer? How much was that offer? Was it competitive? Did you actually win the house? Who’s touring? And so all of that information can go together to feed into some really cutting edge machine learning algorithms.

Jim Freeze Yeah. I mean, I was a Math major as an undergrad and I have a Master’s in Math too. So I love AI, it’s Algebra and Big Data, right? And there’s a lot you can do from a predictive perspective.

Bridget Frey Absolutely. And you’re looking for meaningful ways to have an impact on your customers. It’s not just about finding correlation between two points of data. I think what’s really interesting about Machine Learning in a business like Redfin is that it only matters if it helps our customers achieve their goals. It has to help them get into a home faster and save them money and make sure that we’re really connecting them with their dream home. Otherwise, it’s just a correlation in a data set.

Jim Freeze Exactly. Exactly. So at Redfin, like our own company Interactions. We believe in balancing powerful AI with a human touch, and I know you do as well. Can you talk a little bit about this balance and why it’s important, specifically in real estate?

Bridget Frey Yeah, so, one of our first AI algorithms that we shipped was a recommendation algorithm. And the idea was that we were going to save our agents’ time because agents do spend a lot of time trying to find the right homes to recommend to their customers. So we said, well, what if we could replace all of that with Machine Learning? So we built this algorithm and it suggested for each customer what houses we thought they’d like, and we shared it with the real estate agent. And all they had to do was click a button and send it out to their customers. And we shipped this thing and we thought it was the greatest thing since sliced bread and the agents didn’t use it. And we’re thinking, “Why wouldn’t they use this such time-saver?” and so we started talking to them, and as it turns out, we had not put in a way for an agent to add or remove a home from that list.

We actually, I think as engineers had that hubris to think that we really knew what was best. But the agent knew that a particular house, maybe the customer really, they wanted it, the house to be on the waterfront, or there was some reason that this particular house now they wouldn’t recommend it because they really knew the customer in a way that the algorithm didn’t. And so we had to go back and add that way for a human agent to really interact with the recommendations’ algorithm. I think it was a very early lesson in how we needed to make sure we always had humans in the loop and that we weren’t just trying to have a machine replace this high touch real estate experience.

Jim Freeze Yeah humans need AI, but AI needs humans. So, it’s interesting after the first season of the podcast, and we’re going to be entering our fifth season. We had to start talking about the pandemic, and we’re still talking about it. In particular, the pandemic has transformed many aspects of all of our lives. But the real estate industry in particular has been so crazy for the last 18 months. I actually just sold what used to be my primary residence in Boston, and I’ve moved to South Florida. And I know that many other people because of work at home policies are reevaluating where they want to live. So can you talk about trends that you’ve seen and how Redfin has navigated them over the course of the past 18 months?

Bridget Frey Yeah, so, I mean, the first thing we saw, which many businesses saw when the pandemic hit was that all of a sudden, it seemed like all of our customers were thinking about other things. They weren’t looking to buy or sell homes. Everything was just kind of on pause. And so it was a very scary period in our business where we just didn’t know when were we going to get past this point. When more people are going to start moving again? But then it came pouring back where, as you mentioned, I think as people started to see that there could be opportunities to work remotely on a long-term basis, or people just reevaluating their lives. Because when something big happens in the world, it can lead people to ask questions about where they want to go with their lives. And so we just started seeing all kinds of things happen.

We started seeing a real uptick in the number of people who were searching for homes outside of the place where they lived. So, searching in other metros, it went from about a quarter to about a third of our searches, were people looking in other areas. So, that was one trend. Another was that all of a sudden people were really interested in a virtual real estate experience. So, as a technologist, I’ve always loved things like that. We offer 3D virtual tours where you can walk through a home on the site, almost like you’re in a video game without getting off your couch and going to tour. We had tried to also offer virtual tour experiences for people where an agent would go to the house with a webcam and kind of walk you around the house. But those features were starting to take off.

But then, when the pandemic hit, it went from 1% of customers wanting virtual tours to a third, almost overnight. It was just this huge change because people, all of a sudden didn’t want that in-person experience anymore. Or they only wanted it for homes that they really cared about. So, our business had to quickly adapt to those new trends and try to figure out how do we help people in this new world? How do we help them see homes and move in the midst of a pandemic?

Jim Freeze Yeah. Well, technology has really facilitated that, and it’s interesting. And I’m sure it’s going to evolve over the next year even more. One of the things too, that I think I’d love to touch on, is I know you’re passionate about diversity and inclusion in the Tech industry. Can you talk about why you think that’s important, both from a company culture and a business standpoint?

Bridget Frey Yeah. So it’s an area that I think as a woman in technology, I have one set of experiences of feeling like an outsider sometimes in my own industry. And I know that there are so many people for so many reasons who’ve been excluded from the massive growth and the impact that the Tech industry has for all sorts of reasons. So, when I think about what do I want my legacy to be after my career, it’s just part of it is to make the Tech industry just a little bit better and to make it a little more inclusive so that the problems that Tech is trying to solve include the problems of all people in the world, not just a certain set of people.

So, when I started at Redfin, I was the only woman on our Seattle engineering team. Today, 36% of our team are women, 10% are black and LatinX. And, while those numbers don’t yet reflect the diversity of America, they take up every year and it’s come through concerted effort to try to make our company an inclusive and welcoming place, and make sure that we’re going out and meeting people where they are and trying to bring them in.

Jim Freeze Well, being purposeful about it is a mantra and the way you think about it. And I recently saw in an interview you did your approach to diversity and inclusion as you would an engineering project, which I thought was a really interesting perspective. Could you elaborate a bit on that idea?

Bridget Frey Yeah, I mean, I think engineering teams are built up to solve some really tremendously exciting and difficult problems. There are engineering teams that are trying to figure out how people can live on Mars. There’s all sorts of things going on. And so, why is diversity the one problem that the Tech industry can’t seem to figure out how to solve? And so, I’ve looked at diversity as just, a set of some large and some very small problems, and you kind of just need to break everything down. And we actually use the same bug tracking software that we use for identifying problems in our software to identify bugs or things that we want to work on from a diversity perspective. So we’ve made a list and we’ve prioritized the types of projects we want to work on that we think will make Redfin a better place for people to work.

Jim Freeze That’s fantastic. That’s fantastic. One last question for you before I let you go. AI’s come a long way in real estate in just the 10 years that you’ve been at Redfin. But there’s likely a lot more innovation ahead. How do you see the role of AI in real estate growing and evolving over the next 5 to 10 years?

Bridget Frey Oh yeah. There’s so much still ahead. We have an algorithm called the Redfin Estimate that helps people figure out what their home would sell for. And honestly, the way that algorithm works would have been cost-prohibitive just a few years ago, before we had some of the innovation and cloud technology and algorithm speed that have come about.

So, there’s just, AI is continuing to rapidly evolve and it’s opening up new opportunities. So for example, one thing we’ve been looking at is just, if you look at images of homes and there’s so many pictures that are taken in the real estate industry, what information can you get from those images that can help you understand what it would be like to live in that house? Help a customer understand what the price might be of the house. Can you figure out the condition of the home in a meaningful way? So there’s all kinds of things that are going on there, but we’re also looking at how AI can help you figure out what it’s like to live in the area outside your home. So can we start to determine things like climate change, pollution, other things that might be going on outside the home. Just bringing in a lot more data sets and trying to assess that using a machine learning and data techniques.

Jim Freeze Wow, that’s fascinating. Bridget, thank you so much for joining us today on The ConversAltion. That’s been such a crazy year, 18 months, actually, for the real estate world. And it’s really fascinating to hear how Redfin is helping buyers and sellers navigate kind of the new real estate normal. We very much appreciate your time today. Thank you very much.

Bridget Frey Thanks so much for having me. This has been a great conversation.


That’s all for this episode of The ConversAItion. On our next episode, we’ll hear from Jennifer Flashman, Director of Analytics at Tinder, on how AI can help play matchmaker, and facilitate safer dating.

This episode of The ConversAItion podcast was produced by Interactions, a Boston-area conversational AI company. I’m Jim Freeze, and we’ll see you next time. 


Check out more episodes of The ConversAItion.