There are many things that complex algorithms and artificial intelligence (AI) do very well. It turns out that home flipping is just not one of them. For most families, from Asia to the Americas and Australia, home is probably their biggest lifetime investment. There are a lot of tried-and-tested methods to buy and sell homes. A seller might get his property evaluated, hire an estate agent and list it for sale. Buyers might hire an agent who would show them the sort of homes they are looking for in desirable neighbourhoods. The process can take a few days, weeks or, indeed, even months, as many frustrated sellers and buyers of property will tell you.
Traditional full price realtors in the US can make between 5% and 6% on the sale of every home: split between a listing broker, selling broker, seller’s agent and buyer’s agent. Online real estate firms cut this down to 4% — or even lower, in many cases. There is a lot of money in buying and selling homes because there is a lot of friction in the entire process. If you have ever sold a home and have had dozens of strangers come in and inspect every nook and corner over several weeks only to return for a second or third look at your home, you probably know what a hassle selling through a human property agent can be.
What if a proprietary computer algorithm helped you buy a home at a bargain and then helped you sell it at the highest possible price when you are ready to move on? What if the whole process was completely automated using state-of-the-art semiconductors and great software running on super-fast computers leveraging huge data about local property markets that had been accumulated for decades?
Data is the new oil, and the streets are paved with gold for whoever can monetise it in as many ways as possible. There are already driverless cars and robotaxis on trial in San Francisco and Phoenix suburbs. So, how difficult can it be to make money buying and selling homes using sophisticated algorithms, AI and machine learning?
Online property portals are almost as old as the Internet itself. Initially, they just replaced the classified property sections of newspapers. Over time, they morphed into a platform for real estate agents to find people looking for homes. Online real estate marketplace is where you can find information on just about any home in America, like the last time it was sold, the transaction price, or its estimated current value. Online listings are media-rich with beautiful photos of homes and neighbourhoods, 3D scans of the interiors, as well as water and skyline views for house hunters to browse. Americans call the phenomenon of gawking at online real estate “house porn” that families spend hours on.
At the core of online real estate is a treasure trove of data. It is a capital-light, asset-light business with decent margins. Marketplaces collect data as they scale up and use sophisticated data analytics and algorithms to differentiate from their peers. In 2020, 63% of North American home buyers made at least one offer on a home they saw online and had never stepped into. One reason for the popularity of online real estate is a huge demographic shift. Tech-savvy younger Americans who had deferred buying a home until their late to mid30s are taking the plunge but reluctant to hire traditional brokers. Seattle-based online marketplace Zillow Group has become the icon for house hunters in the US. Earlier this year, it was even featured in a Saturday Night Live comedy skit.
Having collected data for years, online real estate firms such as Redfin Corp and Opendoor Technologies pivoted to make some extra dough on the side by flipping homes. Why not just go in, buy homes by quickly giving sellers a quote, take it off their hands, then turn around, splash on a coat of paint, fix a few things and sell it by re-listing it as a newly refurbished home?
The phenomenon is known as instant buying, or iBuying. For home sellers, iBuying promises a selling experience that is quick, easy and hassle-free. Over the past six years, iBuying firms have used algorithms to make sellers of homes an instant cash offer for their properties in a matter of days. You list your home on an online platform, algorithms discover it, the iBuyer analyses the data you have put up there about your home and you get an offer for your property. It is fast and painless. It may not be as quick as trading stocks commission-free on Robinhood Markets but for something that took a lot of hassle and time, iBuying felt almost as fast.
iBuyers have all the data needed to sell, including what sort of homes people are searching for, which neighbourhoods are suddenly hot, and what homes people are clicking on and liking. They have all the city, regional and national data that goes back years, which even the best property agent in the world with an encyclopaedic memory could not match.
Zillow and their ilk know what sort of photos buyers click on or what families in certain neighbourhoods look for. When you list your property for sale, you probably just post a few random photos of your home and hope the buyer will somehow fall in love with the property. iBuyers, or their algorithms, know exactly what photos to post, what sort of lighting to use when the home for sale is being photographed or what angle the photographer should use when highlighting the kitchen island, living room and front lawn.
Zillow dived into iBuying because co-founder and CEO Rich Barton, a serial entrepreneur and former Microsoft executive, saw competitors making a lot of money flipping homes. If Redfin could do it, Zillow could do it far better. Get great photos, and buyers will take that home off your hands quickly. Barton should know a thing or two about photos. He co-founded travel photography site Trover. Barton had helped found online travel giant Expedia Group within the Seattle-based software powerhouse after convincing Bill Gates that eventually travellers would book plane seats and hotels online.
On Nov 2, Zillow announced that it was shutting down its iBuying operations, having lost US$381 million ($515 million) home flipping in the last quarter and fired a quarter of its 8,000 staffers. “With the price-forecasting volatility we have observed and expect in the future, we have determined that the scale would require too much capital, create too much volatility in our earnings and balance sheet, and ultimately result in a far lower return on equity than we imagined,” Barton was quoted as saying.
Online marketplaces like Zillow have been hoping to become a one-stop shop that real estate agencies could never become, by streamlining and digitalising everything and cutting out the middlemen. They also want to control as much of the process — from listing to the actual sale — taking a cut at each step even as they bring the total cost down as possible. Home flipping through iBuying was an end in itself but a means to an end. Collecting fees from every transaction in the process and providing an array of services that buyers or sellers of home might need was the key to making them the Facebook or Google of real estate.
Mother-of-all home booms
The irony is that one of the largest home flippers in the US was quitting smack in the middle of the mother-of-all booms. Over the past year or so, the country has been experiencing the biggest house price boom in history. US home prices are up 24.8% in 19 months since March 2020. The Case-Shiller National Home Price Index was up nearly 20% in the 12 months to September 2021, the biggest oneyear increase in the country’s history. The National Association of Realtors says home prices recently marked 116 consecutive months of year-on-year price increases. Goldman Sachs expects US home prices to increase another 16% next year.
The best time for flipping homes is when prices are rising, as they have been in the US for the past decade. Covid-19 only helped accelerate the trend as interest rates fell to record lows.
Yet, algorithms failed to work for Zillow. While there was plenty of data that online real estate firms could slice and dice, Zillow just was not getting the competitive advantage it needed to flip the homes it was buying for a quick buck. When humans estimate the price of the home you have put up for sale, they tend to walk around the house because they know what to look for — termites, plumbing, roofing, noise pollution, smelly basements, or stuff that algorithms cannot spot in photos, Matterport 3D scans or text.
Technology aside, Zillow had trouble finding workers who could do basic repair work and deal with soaring raw material costs due to supply chain issues. Contractors were charging an arm and a leg for a paint job or to fix up little things.
Any home flipper will tell you that if you pay more to fix things up, you need to charge a lot more to break even. Even in a hot market with prices up 20%, if you are paying 40% more to fix it up before flipping, it will eat into your margins. Indeed, as Zillow found, in some cases it was difficult to even sell at cost. If you are an institutional home flipper, you just cannot afford to lose US$30,000 on every other home you sell.
Algorithms were also bad prediction machines because they failed to anticipate the recent slowdown in house price increases. Even as the pace of home price appreciation slowed, Zillow kept buying new homes, which left it with huge unsold inventory.
An iBuyer, which anticipates on average a 20% annual increase in prices, can afford to buy a property for US$400,000 and spend US$20,000 to do it up because, eventually, it can sell it for up to US$480,000. If the cost to repair and repaint increases to US$40,000 and the price increase slows to 10% or 8%, the flipper has no margin for error.
Ivy Zelman, CEO of consultancy Zelman & Associates, believes overbuilding in the single-and-multifamily home sector, clouded by dual ownership and institutional capital, could lead to a correction next year as interest rates rise.
No longer the prima donna of online real estate, Zillow now needs to pivot lest it becomes the next Yahoo! or Myspace, online portals that were powerhouses before Google and Facebook became dominant and made them irrelevant. Still, real estate is one big sector that remains ripe for disruption. Clearly, any tech company that takes some friction out of the arduous process of buying and selling homes stands to make a ton of money. The future of home buying is neither just traditional real estate agents nor sophisticated computers and software, but a combination of algorithms and human verification.
Assif Shameen is a technology and business writer based in North America