The Ecosystem around Real Estate Transactions
where software and digitization will tend to matter
Real estate makes up about 15-18% of GDP and is by far the largest transaction the average American will make in her lifetime. The existing real estate transactional system is largely an emergent system. If you were to design it from scratch, you probably wouldn’t design the transactional process this way - at least, that’s what every crypto real estate startup and Opendoor-esque market maker will say. But broadly attempts to opt out of the current transactional landscape have been inadequate, inviting myriad new risks and intractable challenges. At the same time, these would-be disruptors have a point: the large digitization of systems and the inventory invites new possibilities.
But at the same time, the relational aspects of the transaction (contra OpenDoor) matter quite a bit. Likewise, it’s not clear what titles on the blockchain gets you. We already have a decentralized storage system for these: local county clerks.
Thus far, technology has mostly succeeded where it emphasizes and allows for the relational aspects of the process to excel.
And short of a market-making utopia or a crypto revolution, it seems that we are still mostly reliant upon the the traditional players whom make up a heavily fragmented, highly regional ecosystem. And so the key question for a wide swath of players in real estate land has simply been: can we create data and workflow pipelines that at the margin make these players more efficient?
These innovations have had large impacts on the transactional experience and enabled most purchasers to start their search on the internet and end their journey with an e-signature.
Zillow and the Digitization of Inventory
Nowadays most house searches begin with Zillow.1 Prior to Zillow, finding a house for sale largely meant driving around looking for signs. Now, most housing searches begin on Zillow’s graph. If you were to make some sort of comparison, you might land on Sears. Sears invented the retail catalogue, Zillow essentially commercialized the same thing for real estate. Plug in your desired zipcode, some general parameters about how many bedrooms you want, and Zillow will give you everything you may need to start your search.
This unlock is taken for granted nowadays, but I think they really captured the first inkling of what the new world would look like in real estate. They are the data layer upon which all other real estate innovations will continue to be predicated. I don’t mean here that Zillow is some sort of primitive layer on top of which things get built, but rather that they’ve had massive impact across the purchasing experience and on the ecosystem. One trite example: a whole photography staging industry has boomed in Zillow’s wake.
As a house search progresses, realtors get involved. Or perhaps more accurately, realtors ensure they get involved. Their acquisition efforts have likewise changed with the new internet paradigm. Social media was of course a large unlock, but perhaps the larger one has been the surface area that an individual realtor can cover. It used to be largely based around your town. Now it can be your entire metroplex, due to how much data is newly available. Larger terrain and the proliferation of data has led to an increased carrying capacity for realtors, nearing 1.6 million in 2022.2
While realtors might be vSaaS users, they are typically not the buyer. Realtors work under a brokerage. And that brokerage is the most important segment for vSaaS in this end of the market.
Brokerages
Brokers tend to be boutique or empires. These tend to attract different types of realtors, but nonetheless, each competes for realtors and have a wide variety of partnerships with the rest of the ecosystem to make the transaction as seamless as possible.
vSaaS here is highly focused on optimizing the back office and ensuring that brokers can work with their realtors effectively. And while you may think most of this functionality can be found in horizontal solutions, Brokermint (and their parent company Boomtown) has found massive success in packaging up all the workflows and presenting it in an interface ready made for brokers to manage their realtors. Within this space, I’m not sure where you could really compete with them. Instead there seems to be some white space to do something Yoni says:3
Keller Williams is a killer business model. There's plenty of space here to build a far better Keller Williams, probably more tailored around a tech-forward realtor while make operations far more efficient. If AI enables the reduction of further overhead associated with brokerage operations, there seems to be plenty of opportunity to pass back some commission to realtors and win a large chunk of the long tail.
Title & Escrow
Alright, you’ve made it to the house purchase and you enter the actual transaction - dealing with a mortgage lender and title & escrow.
Title & Escrow, as the name suggests, ensures funds and title are transferred to the respective parties. These are often highly regional entities due to the the local county filings that must happen.
Qualia is the name to know here, and they’ve been massively successful. As you can imagine, housing funds on behalf of parties during closing opens you up to a lot of accounting risk. Qualia’s software not only handles the accounting but streamlines the entire process around this section of the transaction. They’ve now built or integrated pretty much every aspect of T&E operations - communication, document prep, and connections into underwriting for title insurance.
There’s a whole body of law on why title insurance is deemed necessary by lenders, but the highlights include the Rule Against Perpetuities, handshake deals, liens, and plenty of resulting lawsuits.
Thus far, I haven’t seen any truly impressive innovation in the actual title insurance process. Part of this is due to how concentrated the insurers are (there’s a title insurance version of the Big 4 too) and part of it has to do with how hard it would be to break into the industry itself and win business. There seem to be really interesting things to build, especially given the broader digitization trends around titles. You could probably imagine Qualia deciding to build here - especially as they have established a channel into every important T&E agency.
Mortgage Lending
In the midst of all of this, you get to a bank or lender. Now the lender has probably already given you some sort of pre-qualification, some indicator that you will actually be able to purchase a house in a price range.
But now we are down to brass tacks and also in perhaps one of the true frontiers for real estate transaction tech.
The past decade in real estate transactions has been an attempt by every disruptor to make every other bit of the transaction far more efficient without having to displace the LOS - the loan origination system housed inside a lender and bank.
Blend is probably the best example of this. Blend initially tackled the user interface for the mortgage process; things like applications, rate shopping, and more.4 In many respects, digitizing these aspects was highly critical in order to move deeper into the banks and lenders. After all, if most of the intake information is digitized and not contingent upon re-entering paperwork into the LOS, it stands to reason that the backend of the loan process can be far more digitized too.
None of this would really matter if there wasn’t reason to believe that mortgage lending could be far more automated than it is today. Mortgage lending is fairly rule-based. Does this person’s FICO score, job, and debt-to-income ratio broadly ensure that they can make mortgage payments? Insofar as you have some rule set dictating lending, you can envision a next-generation LOS automating far more of the process.
A couple of things tend to matter for automating lending:
I have to ensure that the data is airtight and collected without a ton of manual operations.
I have to know a fair amount about the asset I am lending funds to purchase. Housing is illiquid, how can I reliably obtain an appraisal?
I have to be able to settle the funds into the proper escrow account and ideally setup servicing operations.
I have to be able to demonstrate that this loan was given in a compliant manner and run reporting on the loans I am giving out.
Suffice to say, the LOS systems don’t exactly support this sort of infrastructure. You need a combination of integrations into products like Blend for information intake, ML and a workflow engine to help ensure you are nearing automation, and a really great risk engine to help ensure that this is all done in a safe manner.
That’s what Vesta is building and it’s perhaps one of the more challenging problems to crack. Displacing a system of record is never easy, but displacing a system of record that trillions of dollars of assets rely upon is a true feat.
If they are successful, you can imagine the customer experience nearing an ecommerce-esque experience and the backend of mortgage lending nearing complete automation.
Throw in an LLM that answers emails and chat popups, and congratulations, you’ve mostly cut out the side of the transaction and relationship that almost nobody wants to spend time on.5
Real estate transactions already involve disguising as much of the behind-the-scenes as possible. Over the next decade, it’s possible to envision software and AI enabling this vision to an even greater extent, cutting the overhead associated with much of the transaction significantly, and… probably encouraging even more realtors to start.
The relational side of the business isn’t going away, the nuances of the real estate transactional chain are here to stay, but instead we probably end up with a bunch of systems talking to each other to complete the transaction once both sides of a house purchase come to terms.
Perhaps in a world without the cultural rite of passage associated with home purchasing, you could envision automating the entire process. But in the meantime, there aren’t any robot realtors to toast your new purchase.
Or Redfin, but let’s focus on Zillow.
Of course, you have a long tail here. But the long tail’s existence has been far more predicated on how easy it is to get your realtor’s license, help your friends with their housing search, and make a 2% fee.
Yes, Blend has been slaughtered in the public markets. I’m a believer of what they are building in the long run. Their real opportunity should lie beyond real estate and across all sorts of bank transformation initiatives. The product is great, the real question (and what has scared every investor) is whether they can survive the mortgage recession and some past financial decisions that weigh heavy upon their future.
Mortgage lenders also don’t want to spend time here. Since, they are bundling these assets into an MBS anyways and passing on servicing obligations down the road, they could care less about the relational aspect for the most part.