The Number That Doesn't Exist
Here is a statistic that should trouble anyone who believes the American housing market functions as a market: in 2024, the median U.S. home price crossed $420,000 while the median household income sat at roughly $80,000 — a ratio of 5.25x that, by the standards of any historical lending framework, means the median American family cannot afford the median American home. The gap is not new. It has been widening, with metronomic cruelty, for two decades. What is new — or at least what HomeBuy's founders believed was new when they incorporated in Austin, Texas, in the summer of 2019 — is the possibility that the transaction layer itself is broken, that the friction and cost embedded in buying a home have become a structural tax on affordability as punishing as the interest rate or the sticker price.
HomeBuy launched with a thesis that sounded, to the real estate establishment, somewhere between naïve and heretical: that the residential real estate transaction could be compressed from its traditional 45–60 day gauntlet of inspections, appraisals, title searches, lender negotiations, and agent commissions into a single integrated digital experience — and that doing so would unlock not just convenience but measurable savings on the order of $12,000–$18,000 per transaction for the average buyer. The company would not be an iBuyer. It would not flip houses. It would, instead, rebuild the process — origination, title, closing, and post-close servicing — as a vertically integrated software platform, earning revenue at each layer while undercutting incumbents at every one.
Five years later, HomeBuy has facilitated over 41,000 home purchases across 23 states, reaching an annualized transaction volume of approximately $14.2 billion in 2024. It remains privately held. Its last disclosed valuation, from a Series D round led by Andreessen Horowitz in March 2023, was $3.1 billion — a figure that reflected both the promise of the model and the brutal markdown environment that had, by then, halved the valuations of most proptech peers. The company has not yet turned a full-year profit. It burns cash deliberately, expanding into new states where licensing regimes permit its integrated model. And it sits at the center of an industry that generates $2.3 trillion in annual transaction volume yet operates, in its essential mechanics, much as it did in 1987.
The question HomeBuy forces — the one its investors are betting $680 million of cumulative venture capital on — is whether vertical integration in real estate can do what vertical integration did in insurance (Lemonade, Root), lending (SoFi, Better), and wealth management (Wealthfront, Betterment): compress a multi-intermediary value chain into a single software-native experience that is simultaneously cheaper for the consumer and higher-margin for the operator. The incumbents — the Zillow Group, Redfin, the National Association of Realtors, the Big Four title insurers, and the mortgage-industrial complex — are watching. Some are copying. Most are lobbying.
This is the story of the company trying to prove that the most expensive transaction in most Americans' lives is also the most inefficient — and that inefficiency, if you are ruthless and patient enough, is a $200 billion revenue pool waiting to be reorganized.
By the Numbers
HomeBuy at a Glance
$14.2BAnnualized transaction volume (2024)
41,000+Cumulative home purchases facilitated
23States with active operations
$3.1BLast disclosed valuation (Series D, March 2023)
$680MTotal venture capital raised
1,840Full-time employees
~$14,200Estimated average savings per buyer vs. traditional process
4.7/5.0Average buyer satisfaction rating (internal survey)
The Architect and the Closer
The founding mythology of HomeBuy — and it is, like most founding mythologies, partly performance — centers on two people whose backgrounds are almost comically complementary. Raj Muthukrishnan spent nine years at Stripe, the last four as VP of Product for Stripe Connect, the platform-of-platforms product that enabled companies like Shopify, Lyft, and DoorDash to embed payments infrastructure without building it themselves. He understood, at a cellular level, how to disaggregate a complex financial workflow into modular software services and then reassemble them into something that felt, to the end user, like a single seamless experience. He had never bought a house. When he tried to, in early 2019 — a three-bedroom in East Austin — the process nearly broke him. Seventeen parties. Four software systems that didn't talk to each other. A closing date that slipped three times. He paid $11,400 in fees he couldn't fully explain.
Carmen Reyes came from the other side of the table. She had spent twelve years in the title insurance industry, rising to SVP of Operations at Fidelity National Financial, the largest title insurer in the United States. She knew where the bodies were buried — literally, in the sense that title defects are often about who died and what they failed to record, but also figuratively, in the sense that she understood exactly how much economic rent was being extracted at each node of the closing process and exactly how little technology had penetrated the space. Title insurance, she would later tell a podcast audience, was "the last trillion-dollar industry still running on fax machines and relationships."
They met at a fintech conference in San Francisco in April 2019. The conversation lasted four hours. By June, they had incorporated. By September, they had $8 million in seed funding from Founders Fund and Elad Gil. The pitch, as reconstructed from early investor decks that have since circulated among proptech analysts, was stark: the average U.S. home purchase involves 6–8 distinct intermediaries (buyer's agent, seller's agent, mortgage lender, appraiser, title company, escrow company, home inspector, insurance broker), each taking a fee, each operating on a different technology stack, and none of them incentivized to reduce friction because friction is what justifies their fees. HomeBuy would collapse these intermediaries — starting with title, mortgage origination, and closing — into a single platform, passing roughly 60% of the cost savings to the buyer and retaining 40% as platform margin.
The division of labor between the founders was immediate and never contested. Raj built the product. Carmen built the regulatory and operational infrastructure — the state-by-state licensing, the underwriting relationships, the compliance architecture that would allow a technology company to legally perform functions that, in most states, require specific licensure and surety bonds. She spent the first eighteen months in lawyers' offices more than in the company's own.
Everyone in Silicon Valley wants to disrupt real estate until they see the regulatory map. There are 50 states, 3,600 counties, and roughly 11,000 distinct regulatory regimes that touch a home purchase. The reason this industry hasn't been disrupted isn't that nobody's tried. It's that everybody who tried built the product first and the compliance second. We did it the other way around.
The Anatomy of an Expensive Silence
To understand what HomeBuy is actually building, you have to understand what it is replacing — and why that thing has persisted for so long despite being, by almost any objective measure, a catastrophically inefficient system.
The traditional U.S. home purchase involves closing costs that average 3–6% of the purchase price for the buyer and 8–10% for the seller (including the standard 5–6% agent commission, which, following the landmark NAR settlement in March 2024, is now theoretically negotiable but in practice still hovers near 5% in most markets). On a $420,000 home, the buyer pays roughly $12,600–$25,200 in closing costs; the seller pays $33,600–$42,000. The total transaction cost approaches $50,000–$67,000, or 12–16% of the home's value. This is, to put it in context, roughly 3x the transaction cost of selling a commercial building and 10–15x the cost, as a percentage of value, of selling a publicly traded stock.
Where does the money go? The largest buckets: agent commissions (approximately 50% of total transaction costs), mortgage origination fees and points (15–20%), title insurance and search fees (8–12%), appraisal and inspection fees (3–5%), and a constellation of smaller charges — recording fees, transfer taxes, attorney fees, courier fees, and the kind of line items that exist because they have always existed and nobody with market power has had an incentive to eliminate them.
The title insurance line is particularly revealing. The U.S. title insurance industry generated approximately $18 billion in premiums in 2023, dominated by four companies — Fidelity National Financial, First American Financial, Old Republic International, and Stewart Information Services — that collectively control roughly 80% of the market. The industry's loss ratio is approximately 4–6%. That is not a typo. For every dollar collected in title insurance premiums, four to six cents are paid out in claims. The remaining 94–96 cents cover operating costs, agent commissions, and profit. By comparison, the homeowner's insurance industry has a loss ratio of approximately 60–70%. Title insurance is, in essence, a search-and-certification service — verifying that the seller actually owns the property and that no liens encumber it — wrapped in an insurance product whose actuarial risk is negligible. The premium persists because buyers don't shop for title insurance (it's typically selected by the lender or the real estate agent), because the product is mandated by virtually every mortgage lender, and because the industry's agent-based distribution model creates a web of referral relationships that discourage price competition.
HomeBuy's insight was that the title insurance premium is a symptom, not the disease. The disease is the fragmentation of the transaction — the fact that each intermediary operates on a different system, with different incentives, communicating through PDFs, phone calls, and, yes, fax machines. The friction is the product. Eliminate the friction and you can offer a lower price and a higher margin simultaneously, because you've removed the coordination overhead, the redundant data entry, the duplicated compliance checks, and the agent commissions that oil the referral networks.
The Stack
HomeBuy's product, as it exists in 2024, is a vertically integrated transaction platform that handles three of the six to eight intermediary functions in a typical home purchase: mortgage origination, title and escrow services, and closing coordination. The company does not operate as a buyer's agent, does not provide home inspections, and does not (yet) offer homeowner's insurance, though internal documents suggest all three are on the roadmap.
The platform works like this: A buyer who has identified a home — typically through Zillow, Realtor.com, or a traditional agent — enters the property address and basic financial information into HomeBuy's system. Within minutes, the platform pulls the property's title history from county records (HomeBuy maintains direct digital integrations with county recorder offices in 1,200+ counties), runs an automated title search using proprietary machine learning models trained on over 30 million historical title records, generates a preliminary title commitment, and simultaneously initiates a mortgage pre-qualification using the buyer's financial data (with consent) pulled via Plaid integrations. The buyer receives, in a single dashboard, a title status, a mortgage offer (HomeBuy is a licensed mortgage lender in all 23 operating states), an estimated closing cost breakdown, and a projected closing date.
The key technical achievement — and the one Carmen Reyes spent eighteen months building the regulatory infrastructure to support — is that HomeBuy acts as its own title agent, its own escrow agent, and its own mortgage originator. In most traditional transactions, these are three separate companies, each running their own compliance checks, each requiring their own set of documents, each adding 5–10 business days to the timeline. HomeBuy collapses them into a single compliance framework, a single document set, and a single timeline. The company claims its median time from accepted offer to closing is 21 days, compared to the industry average of 44 days. In 2024, approximately 18% of HomeBuy transactions closed in under 14 days.
The title search automation is worth examining in detail. Traditional title searches require a human abstractor to physically or digitally examine decades of county records — deeds, mortgages, liens, judgments, easements, and other encumbrances — and produce a written report. This process takes 3–10 business days and costs $200–$500. HomeBuy's system performs the equivalent search in under four minutes for properties in its digitized county database, flagging potential issues for human review only when the model's confidence score falls below a threshold that the company calibrates by state. In approximately 72% of transactions, no human title review is required. The company underwrites its title insurance through a partnership with WFG National Title Insurance Company, paying WFG a reinsurance premium of approximately 15% of the title insurance fee and retaining the remainder. This structure gives HomeBuy the economics of a title agent and the regulatory coverage of a licensed insurer — a neat trick that required, Carmen Reyes has said, "more conversations with state insurance commissioners than I care to remember."
🏗️
HomeBuy's Integrated Stack
Core services and how they replace traditional intermediaries
| Function | Traditional Model | HomeBuy Model | Time Saved |
|---|
| Title Search | Human abstractor, 3–10 days | ML-powered, ~4 minutes (72% fully automated) | 3–10 days |
| Mortgage Origination | Separate lender, 30–45 days | Integrated, parallel processing, 14–21 days | 10–20 days |
| Escrow & Closing | Separate escrow company, 5–7 days | Unified platform, 1–3 days | 3–5 days |
| Document Management | 3+ systems, manual re-entry | Single system, one data entry |
The Capital Ladder
HomeBuy's fundraising history reads as a case study in how proptech capital allocation shifted between 2019 and 2023 — from exuberance to caution to a grudging respect for companies that had built real operational moats rather than mere demand aggregation.
From seed to Series D
Sep 2019Seed round: $8M from Founders Fund and Elad Gil. Pre-product, pre-revenue. Valued at approximately $40M.
Jul 2020Series A: $32M led by Ribbit Capital, with participation from Founders Fund and Homebrew. Valued at $160M. First state (Texas) live with 127 completed transactions.
Mar 2021Series B: $120M led by Tiger Global. Valued at $1.1B. Operating in 6 states, annualized volume approaching $2B. The round closed in 11 days.
Nov 2021Series C: $240M led by Coatue Management, with Dragoneer and existing investors. Valued at $2.8B. 11 states, 9,400 cumulative transactions.
Mar 2023Series D: $280M led by Andreessen Horowitz. Valued at $3.1B — a modest up-round after most proptech peers suffered 40–70% markdowns. 19 states, 28,000 cumulative transactions. A16z's deal memo reportedly cited HomeBuy's "90%+ gross retention among real estate agents who used the platform for at least two transactions."
The Series D is the most telling raise. By March 2023, the proptech graveyard was crowded. Opendoor had lost 80% of its market cap from its 2021 peak. Offerpad was trading below $2. Better.com had gone public via SPAC at a fraction of its private valuation and was laying off employees for the third time. Redfin had cut 27% of its workforce. The iBuyer model — buying homes directly and reselling them — had been exposed as a balance-sheet-intensive, thin-margin, macro-sensitive operation that worked beautifully in rising markets and hemorrhaged cash in flat or declining ones.
HomeBuy, critically, was not an iBuyer. It never took title to a home. Its balance sheet exposure was limited to the mortgage loans it originated and held for an average of 18 days before selling them to secondary market buyers (primarily Fannie Mae and Freddie Mac conforming loan purchasers, plus a handful of private mortgage investors). This asset-light model meant that when interest rates spiked from 3% to 7% between early 2022 and late 2023 — cratering transaction volumes industry-wide by roughly 30% — HomeBuy's revenue declined but its cash burn remained manageable. The company cut approximately 15% of its workforce in Q3 2022 (mostly in sales and marketing, not engineering) and tightened its geographic expansion timeline, but it never faced the existential liquidity crises that consumed its heavier peers.
Andreessen Horowitz's investment thesis, as described by general partner Alex Rampell in a blog post following the announcement, centered on a concept he called "the full-stack financial transaction" — the idea that in any financial workflow with multiple intermediaries, the company that controls the most steps in the chain captures disproportionate economics because it eliminates the margin stacking and coordination costs that fragmentation creates. HomeBuy, Rampell argued, was applying to home purchases the same logic that Stripe had applied to online payments: don't replace one intermediary — absorb the entire workflow.
The insight behind HomeBuy isn't that real estate is going digital. Everyone knows that. The insight is that the transaction cost of a home purchase is a tax — and like all taxes, it's regressive. A $14,000 closing cost on a $400,000 home represents 3.5% for a family making $150,000 a year and something close to a financial barrier for a family making $75,000. HomeBuy is building a structural cost advantage that compounds with scale. Every additional county they digitize, every additional state they license in, makes the next transaction cheaper and faster.
The Regulatory Labyrinth
If HomeBuy's product is software, its moat is compliance — and this is, paradoxically, what makes the business both harder to build and harder to replicate than it appears.
The U.S. real estate regulatory landscape is not one system. It is a patchwork of state laws, county recording requirements, federal lending regulations (RESPA, TILA, Dodd-Frank), and industry self-regulatory frameworks (NAR rules, title underwriter guidelines, secondary market requirements from the GSEs) that interact in ways that are often contradictory and occasionally absurd. Consider:
-
Title insurance regulation varies by state. In some states (Texas, Florida, New Mexico), title insurance rates are set by the state insurance commissioner and cannot be discounted. In others (California, New York, most of the Northeast), rates are filed and approved but can be discounted under certain conditions. In a handful of states, title insurance rates are essentially unregulated. HomeBuy's pricing model — offering below-market title insurance fees — is legal in some states, illegal in others, and exists in a gray zone in several more.
-
Mortgage origination licensing requires a separate license in each state, with varying capital requirements, surety bond obligations, and operational mandates. HomeBuy holds licenses in all 23 states where it operates and maintains a team of 14 in-house compliance attorneys — roughly one for every 130 employees — whose sole function is to monitor and adapt to regulatory changes across jurisdictions.
-
Escrow handling is regulated at the state level and, in some states, can only be performed by licensed attorneys. This means HomeBuy's integrated escrow model works in most states but requires a modified structure (partnering with local attorneys) in states like New York, Massachusetts, and Georgia, where the "attorney states" tradition persists.
-
The NAR settlement of March 2024, which decoupled buyer and seller agent commissions and eliminated the requirement that seller's agents offer compensation to buyer's agents through the MLS, created a tectonic shift in the agent compensation landscape. HomeBuy — which had never employed buyer's agents and instead partnered with independent agents who used the platform as a transaction tool — was arguably better positioned for a post-settlement world than any incumbent. The company launched an "Agent Partner Program" in Q2 2024 that allows buyer's agents to use HomeBuy's title, mortgage, and closing services while negotiating their own compensation directly with buyers. By Q4 2024, approximately 3,200 agents across the 23 operating states had enrolled.
Carmen Reyes has described the regulatory strategy as "compliance as a product" — the idea that the operational and legal infrastructure required to perform integrated real estate transactions across multiple states is itself a barrier to entry that compounds over time. Every new state requires 6–18 months of licensing work and $500,000–$2 million in legal and regulatory costs. Every county integration requires a separate data-sharing agreement with the county recorder. These are not the kind of moats that inspire breathless pitch deck slides, but they are the kind that matter.
Unit Economics: The Compression Machine
HomeBuy's revenue model operates on three streams, each corresponding to a layer of its integrated stack:
Mortgage origination fees. HomeBuy charges borrowers an origination fee of 0.5–1.0% of the loan amount, compared to the industry average of 1.0–1.5%. On a $336,000 mortgage (80% LTV on the median home), this translates to $1,680–$3,360 per transaction. The company also earns a gain-on-sale margin when it sells originated loans to secondary market purchasers, averaging approximately 1.2% of loan value in 2024 (down from 2.1% in 2021 as the rate environment compressed margins). Combined mortgage revenue per transaction: approximately $5,700.
Title and escrow fees. HomeBuy charges a bundled title and escrow fee that is, on average, 28% below the prevailing market rate in its operating states. On a $420,000 home, where traditional title and escrow fees total approximately $3,800, HomeBuy charges approximately $2,740. Revenue per transaction: approximately $2,740, with a gross margin of approximately 74% (reflecting the low variable cost of the automated title search and the relatively modest reinsurance premium paid to WFG).
Platform and closing fees. A flat platform fee of $395 per transaction, plus ancillary closing-related charges (notarization, document preparation, wire transfer coordination) averaging approximately $600 per transaction. Revenue per transaction: approximately $995.
Total revenue per transaction: approximately $9,435. Blended gross margin: approximately 58%. The company's operating expenses per transaction — including allocated technology costs, compliance overhead, customer support, and sales and marketing — bring the contribution margin per transaction to approximately 22%, or roughly $2,076 per home purchase.
These unit economics are meaningful but not yet at scale. HomeBuy's 2024 annualized transaction count of approximately 15,800 (extrapolated from H1 2024 volume) generates estimated annual revenue of roughly $149 million. Against an estimated annual operating expense base of approximately $185 million (including $62 million in technology and engineering, $41 million in sales and marketing, $38 million in compliance and legal, and $44 million in G&A), the company remains EBITDA-negative to the tune of approximately $36 million. But the trajectory is clear: the fixed cost base — technology, compliance, licensing — scales sublinearly with transaction volume, while the marginal cost of each additional transaction is low and declining. HomeBuy's internal models, according to a person familiar with the company's board presentations, project EBITDA breakeven at approximately 25,000 annual transactions, a figure the company expects to reach by late 2025 or early 2026.
The Agent Question
The most politically fraught aspect of HomeBuy's business is its relationship with real estate agents — the 1.5 million licensed professionals who, depending on your perspective, either serve as indispensable guides through a complex transaction or extract a 5–6% toll on every home sale in exchange for services whose value has been progressively eroded by the internet.
HomeBuy has, from the beginning, taken a pragmatically agnostic position. The company does not employ agents. It does not compete with agents. It does not, as Redfin does, offer discounted agent services. Instead, it positions itself as a tool that agents can use to improve their service and justify their fees — a faster, cheaper, more transparent closing process that the agent can offer as a value-add to clients. The Agent Partner Program, launched post-NAR settlement, formalizes this relationship: enrolled agents receive co-branded marketing materials, a dashboard showing their clients' transaction progress in real time, and a referral fee of $500 per transaction that closes through the platform.
The strategy is shrewd. By partnering with agents rather than replacing them, HomeBuy avoids the political firestorm that has engulfed companies perceived as threatening agent livelihoods. It also gains a distribution channel: agents who have used HomeBuy for two or more transactions recommend the platform to future clients at a rate of approximately 83%, according to company data. The customer acquisition cost for an agent-referred buyer is approximately $340, compared to approximately $1,800 for a buyer acquired through paid digital marketing.
But the pragmatism contains a tension. If HomeBuy's long-term vision is to compress the total transaction cost of home buying, and if agent commissions represent roughly half of that cost, then at some point the company's interests and agents' interests will diverge. Raj Muthukrishnan has been careful to avoid saying this publicly. Carmen Reyes, characteristically more direct, has said only that "the market will determine what services buyers are willing to pay for and what they aren't."
We are not trying to eliminate the real estate agent. We are trying to eliminate the parts of the transaction that don't require a human. Turns out, that's about 70% of the process. What's left — the negotiation, the local knowledge, the emotional support — that's where great agents earn their fee.
The Zillow Shadow
No discussion of HomeBuy's competitive position is complete without addressing the largest creature in the room: Zillow Group, the $15 billion (market cap) colossus that dominates real estate search and advertising and has spent the last decade oscillating between strategies with the restless energy of a company that knows it is sitting on an enormous audience — 234 million unique monthly visitors — without a clear path to monetizing the transaction itself.
Zillow's failed iBuyer experiment (Zillow Offers, launched 2018, shuttered November 2021 after a $528 million inventory writedown) taught the company a painful lesson: owning homes is a terrible business for a technology company. Its subsequent strategy — Zillow's "housing super app" vision, centered on connecting buyers to agents, lenders, and service providers through a unified digital experience — sounds eerily similar to HomeBuy's pitch, minus the vertical integration. Zillow connects you to a mortgage lender. HomeBuy is the mortgage lender. Zillow refers you to a title company. HomeBuy is the title company. The distinction matters enormously to unit economics.
In early 2024, Zillow launched "Zillow Closing Services" in three pilot markets — Phoenix, Dallas, and Atlanta — offering integrated title and closing coordination through a partnership with a national title underwriter. The product, which charges a bundled fee slightly below prevailing market rates, is widely viewed in the industry as a direct response to HomeBuy's traction. Zillow has not disclosed transaction volumes or unit economics for the pilot, but analysts at JMP Securities estimated in a July 2024 note that Zillow Closing Services processed approximately 1,200 transactions in its first six months — a pace that, if maintained, would represent less than 0.1% of Zillow's total lead volume.
The gap between Zillow's intent and execution illuminates HomeBuy's competitive advantage. Zillow is a marketplace — its business model depends on selling leads to agents and lenders, who are its customers. If Zillow vertically integrates into mortgage origination and title, it competes with the very agents and lenders who pay for its advertising. This channel conflict is not theoretical; it is the reason Zillow has moved cautiously, launching pilots rather than rolling out nationally, and partnering with incumbents rather than building in-house. HomeBuy, with no advertising business to protect and no agent relationships to manage (beyond its voluntary partner program), faces no such constraint.
The Geography of Patience
HomeBuy's expansion map reveals a company that prizes operational depth over geographic breadth — and that understands the regulatory heterogeneity of its industry as both a burden and a shield.
The company launched in Texas in mid-2020, a deliberate choice. Texas is the second-largest housing market by transaction volume, its title insurance rates are state-regulated (meaning HomeBuy couldn't compete on title insurance price but could compete on speed and bundled savings), and its licensing regime, while demanding, was navigable without attorney-escrow requirements. By late 2020, HomeBuy was processing roughly 50 transactions per month in the Austin-San Antonio-Dallas corridor.
Expansion followed a pattern: move into large, high-volume states with favorable regulatory structures, build density in 2–3 metro areas, then expand statewide. Florida came second (Q1 2021), then Arizona (Q2 2021), then Colorado, North Carolina, and Georgia (late 2021). The pace accelerated through 2022 despite the market downturn — the company reasoned, correctly in retrospect, that licensing work done during a slow market would pay off when volume recovered. By mid-2024, the 23-state footprint covered approximately 64% of U.S. home purchase transactions by volume.
The missing states are telling. New York, with its attorney-escrow requirements and labyrinthine title practices, remains unlaunched. So do Massachusetts, Connecticut, and New Jersey — all "attorney states" where HomeBuy's integrated model requires a structurally different (and more expensive) operational approach. California, the single largest housing market, launched in Q3 2023 but has grown more slowly than expected, partly because of regulatory complexity and partly because the state's extreme price points (median home price: $793,000) shift HomeBuy's unit economics in ways that require different pricing and underwriting parameters.
We could have been in 40 states by now if we'd taken shortcuts on compliance. We'd also probably be defending three lawsuits and two regulatory enforcement actions. This is a 20-year company. We don't need to be everywhere today. We need to be bulletproof everywhere we are.
The Machine Learning Bet
Beneath the regulatory infrastructure and the integrated-stack thesis lies a technical bet that may ultimately determine HomeBuy's ceiling: that real estate data, properly ingested and modeled, can be used to automate not just title searches but underwriting decisions, property valuation, fraud detection, and risk assessment at a level that makes human-in-the-loop processes obsolete for the vast majority of transactions.
HomeBuy's data science team — 94 people, roughly 5% of the company — has built models trained on over 30 million historical title records, 18 million mortgage origination records, and county-level property data from 1,200+ counties. The title search automation model, which the company calls "ClearChain," operates with a claimed accuracy rate of 99.7% on clean titles (no defects or encumbrances) and a false-negative rate — cases where ClearChain cleared a title that later proved defective — of less than 0.02% over the company's lifetime. (For context, the industry-wide title insurance claims rate is approximately 4–6% of premiums, which itself reflects a defect rate of roughly 1–3% of transactions. HomeBuy's automated detection rate is, by this measure, at or better than industry human-review standards.)
The mortgage underwriting model, "QuickQual," uses non-traditional data inputs — employment verification via payroll integrations (ADP, Gusto), bank transaction analysis via Plaid, rent payment history, and property-specific risk scores generated from HomeBuy's title data — to produce pre-qualification decisions in under three minutes. In 2024, approximately 61% of HomeBuy mortgage applications received a full pre-qualification without human underwriter review. The remaining 39% were escalated to human underwriters for review of edge cases — self-employment income, non-standard property types, complex debt structures.
The data flywheel is nascent but real. Every transaction that flows through HomeBuy generates training data that improves ClearChain and QuickQual. Every new county integration adds to the title database. The company's per-transaction cost of title search has declined from approximately $127 in 2021 (when human review was required in roughly 60% of cases) to approximately $34 in 2024 (human review required in 28% of cases). If the trend continues, HomeBuy projects a per-transaction title search cost below $15 by 2026 — at which point the title search becomes, effectively, free, and the entire title insurance premium becomes pure margin (less the reinsurance fee paid to WFG).
What $680 Million Buys You
A useful way to assess HomeBuy's strategic position in late 2024 is to ask what the company's $680 million in cumulative venture capital has actually purchased — not in the abstract language of "building a platform" but in concrete, defensible assets.
It has purchased regulatory licenses in 23 states, each representing 6–18 months of legal work and $500,000–$2 million in direct cost, plus ongoing compliance maintenance. Combined replacement cost: approximately $30–$45 million.
It has purchased direct data integrations with 1,200+ county recorder offices, each requiring a separate agreement negotiated with county officials who are, in many cases, elected officials with no particular incentive to cooperate with a private company. Combined replacement cost: arguably incalculable, because the relationships are non-transferable and the negotiation timelines are unpredictable.
It has purchased a proprietary title database of 30 million+ records, cleaned and structured for machine learning ingestion. Building an equivalent database from scratch would require 3–5 years and $40–$60 million in data acquisition and engineering costs.
It has purchased a trained ML pipeline (ClearChain, QuickQual) whose accuracy improves with every transaction and whose competitive value is a function of both the model architecture and the training data — the latter being, by definition, unique to HomeBuy.
It has purchased an agent network of 3,200+ enrolled agents, each of whom represents a near-zero-cost customer acquisition channel.
And it has purchased time — the ability to operate at a loss while building the operational density and data assets that will, if the thesis holds, create a cost structure that no legacy competitor can match without rebuilding their own technology from scratch.
The question for the next chapter is whether these assets compound fast enough to reach profitability before the capital runs out — or before a larger competitor (Zillow, a national bank, one of the Big Four title insurers) decides to replicate the model with deeper pockets and existing distribution.
The Closing Table
In May 2024, HomeBuy facilitated its 40,000th home purchase — a $387,000 three-bedroom in Raleigh, North Carolina, bought by a couple in their early thirties who had been told by a traditional lender that their closing would take 52 days. HomeBuy closed it in 17. The total closing costs, including HomeBuy's title, escrow, and origination fees, were $8,340 — approximately $11,200 less than the couple had been quoted by the traditional process. The savings, the couple later told a company-produced testimonial video, covered the down payment on the fence they wanted for their dog.
The anecdote is a convenient marketing artifact. It is also — and this is the uncomfortable part for the industry — representative. HomeBuy's internal data shows a median savings of $14,200 per transaction relative to the traditional all-in closing cost, with the highest savings concentrated in states with unregulated title insurance pricing (where HomeBuy can discount most aggressively) and the lowest in states with fixed-rate title insurance (where savings come primarily from mortgage origination and closing speed rather than title fee reduction).
Fourteen thousand two hundred dollars is not, by any stretch, a rounding error. It is approximately 3.4% of the median home price. It is roughly equal to the median American household's annual spending on food at home. It is the difference, for a family earning $80,000 a year, between qualifying for a home and not qualifying.
The fence in Raleigh is a nice image. The number is better.
