Roshan SrinivasanDuring the period from 2004 to 2018, property search tools like Zillow and Redfin revolutionized the real estate industry, empowering buyers to find properties on their own. However, there has been a recent slowdown in innovation. AI is now playing a crucial role in property search, making it more personalized and efficient, especially in filtering properties and accessing information.
The Landscape from 2004-2018
2004 was an interesting year for property search tools. Both Zillow and Redfin, two of the most influential players in transforming home buying and selling for end consumers, incorporated in 2004.
Before these platforms launched, the only way in which a prospective buyer could learn about a property was by going through a real estate agent and having them share a MLS (Multiple Listing Services) listing. The National Association of Realtors and local county association of realtors still have a strong control over listings in the United States even today in 2023, but back in 2004, finding a real estate agent was the first step of many to buying a home - not searching for properties yourself like we take for granted today.
Why Property Search was so Powerful
When Zillow and Redfin came out with property search, it was an absolute game changer. Rich Barton, the CEO and co-founder of Zillow had told the Seattle Post-Intelligencer how he originally had gotten the idea for Zillow while still at Microsoft after spinning out his first company of Expedia: "the web [had] been around for now nine years or 10 years, but still I couldn't get pictures of homes and complete listings and prices and addresses…We were trying to answer a simple question. What is that house worth? What should we offer if we wanted to buy it?” (source)
The initial versions of Zillow allowed you to filter for such questions and enabled homebuyers to have more visibility into the ecosystem to make smarter decisions when buying in a new neighborhood.
As property search became more mainstream, these platforms started expanding search capabilities to include more maps-based approaches. In fact, Redfin was a pioneer in map-based search long before Google Maps ever became mainstream. Both Zillow and Redfin were early power users of Mapbox as a software for creating highly custom maps for property search indexing.
More and more data and geospatial solutions also came out which expanded the capabilities of property search for these platforms. WalkScore, which was acquired by Redfin in 2014, is a great example of providing scores for modeling travel distance by walking, biking, driving, and other forms of transit which many search platforms have leveraged in order to provide more context on properties searched. In this particular case, once a user on Redfin or Zillow finds a property they like based on price, number of bedrooms, number of bathrooms, etc., they can view more details about walkability and transit within the Redfin/Zillow report generated for the listing.
Growth of Property Search with Established Brokerages
As Zillow and Redfin began to grow and offer other services such as Zestimates for home value estimation (a highly controversial feature to this day) and agent-matching brokerage solutions respectively, they started to build more end-to-end solutions for homebuyers, agents, and brokers to adopt.
Many more brokerages including Homes.com, Realtor.com (operated under the National Association of Realtors), and even tech-forward agent-brokerages such as Compass, Side, and more have implemented some form of property search for their listings. The idea of property search for listings on the market, rentals, and homes recently sold has been put into practice across the entire industry.
Slowdown in Innovation in the past 5 years
Prior to 2018, most tech-enabled brokerages focused on new core technologies to give agents, brokers, underwriters, home buyers, and home sellers new services whether they be servicing a property transaction, providing a marketplace for agents, or offering other financials tools such as estimation calculators.
But as more and more tools were being built for the top players to stay competitive based on end-user experience, they started losing site of improving core features that could improve ads spend, monetization, and growth since the main focus was on product optimizations. With that, given how interest rates where close to 0%, many brokerages like Zillow, Redfin, and Opendoor started to engage in buying and flipping properties to fund future development. During this period, we also saw more fractional investing platforms like Divvy Homes and Better that were enabling a marketplace for people to buy “stakes” within a property just like buying a single stock - only more illiquid.
As interest rates started hitting 6-7% in 2022-23, many of these brokerages that were growing without assessing risk ended up closing down their buying/flipping businesses and laying off ~15% of their workforce. In the cases of Divvy Homes and Better, they went from Silicon Valley unicorn status with $2B+ valuations to laying off a sizeable percentage of their employees.
The culmination of the slowdown in innovation plus interest rate hikes have forced many brokerages that cut corners for growth during the prior boom from 2018 onwards to reassess how they grow, do business, and continue to innovate.
AI is now a Tipping Point for Change across many industries
As with every industry on the planet, OpenAI’s launch of ChatGPT and subsequent launch of their APIs for GPT4 and GPT3.5 set off an explosion of new startups. The introduction of Llama2 by Meta and other LLMs have created an “arms race” for new AI tooling.
And with the introduction of these models, we will see great improvements to a variety of industries - including real estate. Marc Andreesen describes in his article “AI will Save the World” how AI will help:
- Children with coaching and tutoring at their own speed and development
- Leaders, CEOs, coaches, etc. can make better decisions with intelligence augmentation
- Creatives, artists, and strategists can leverage AI to visualize new scenes and concepts at a greater scale and much faster
Fundamentally, AI is going to help us with personalizing all aspects of our living experiences. That also includes the home purchasing experience!
Applying AI to Real Estate Property Search
Fundamentally, buying a home is a personal choice. Whether a buyer is a first time buyer, looking to buy a vacation home, or maybe invest in a few properties, the criteria they are looking for in a home will all be very different.
Intrinsically, if we just look at the first category of a first time homebuyer, every buyer has a different budget, a different size of home they are looking for, and a different number of bedrooms and bathrooms. Some buyers care about proximity of the house to good school districts for maintaining home value or maybe starting a family. Others care more about being close to the local downtown area to be near bars and restaurants.
For people buying their first home, they may also care if they are buying near a floodplain for damages, or whether they can make a specific addition to a property such as adding a sun room. They may want to know more about how many stories a property is or how often the listing price has changed over the lifetime of the listing.
Each question and layer of research a buyer is interested in today may need to involve a real estate agent, access to an appraiser, lawyer, or the ability to visit the local county office for zoning regulations (depending on the county). However, if we start leveraging LLMs to help with filtering properties, AND we expand the dataset of values possible for properties, we can build a deeper, richer experience to property search to help buyers find the right homes for them.
Leveraging AI for Filtering Properties
In the traditional landscape for property search, a buyer would need to go to a Zillow, Redfin, Homes.com, etc., search for a specific region, enter in the number of beds, baths, and price range they are looking for, and only then will they see a specific list of properties. That said, some properties may be next to a ravine or they might not have a great school district. These are things you only find out after filtering a list of properties.
AI-enhanced filtering allows you to search for properties based on any criteria you want. If you want to search for properties outside a floodplain or properties that have “excellent schools” and “low crime” (both of which Camphor defines relative to the national average), then you can simply ask a question to filter a set of properties based on custom criteria - it’s that simple!!!
Understanding Property Reports using AI
Once you find a property you like, in the status quo, Zillow or Redfin will generate a property report which contains more about the property, transaction history, mortgage values, walk and transit scores, information about school districts, similar properties, and more! The problem is that the user experience is not great! There’s a lot of stats shared with a buyer and they have to personally filter for what they are looking for.
Using AI, property report interfaces can change dramatically. Instead of digging for information, you can just ask a question to receive said info. Often times, platforms like Zillow also leave off information as to not inundate a user, but with AI-powered search, adding this information in is not very challenging - such as being able to search for properties based on a zoning criteria like allowing the creation of a sun room in the backyard!
Applying Semantic Search to Descriptions
The most interesting portion of property reports are the MLS listing descriptions and the ability to look at zoning descriptions as well as other data descriptions for properties. With AI-powered semantic search for filtering and understanding property info, you can now query for “luxury condos with a pool” and other characteristics mentioned in the property listing. Semantic search is something that was never possible before!
How Camphor is Pioneering this Revolution
At Camphor, we’re redefining property search from scratch the same way Zillow and Redfin did in 2004. Our proprietary datalake where we surface zoning, parcel, property financials, weather hazards, schools, and more to stitch together with listings data is the basis for our property search engine. Whether you want to ask a question about “luxury condos in Manhattan under $4M built after 1970” or “3 bed properties in San Mateo, CA outside of a wildfire danger zone”, you can get closer to finding your ideal property on our AI-powered search engine. Check out search.camphor.co for property listings in the Bay Area and New York City today!
Feel free to also follow us on LinkedIn, Twitter, Instagram, and/or TikTok to see more updates. We promise to follow back!