At a Glance
Max Simkoff, founder and CEO of Doma and former founder of Evolv, discusses in this episode how Doma is reshaping traditional title insurance and real estate transaction processes using machine learning. Max Simkoff argues that the core problem in the traditional title insurance industry is that its workflow has remained virtually unchanged for 150 years—each transaction still relies on manual "small-scale research projects." Doma, by contrast, uses predictive models and computer vision to compress hours of manual work into under three seconds.
Theme 1: Title Insurance — An Industry Unchanged for 150 Years
Max Simkoff points out that the origins of title insurance trace back to a 19th-century U.S. Supreme Court ruling, which established the principle of "caveat emptor" and gave rise to an industry that remains heavily reliant on manual labor to this day.
- Historical Context: In the mid-19th century, the U.S. Supreme Court ruled that sellers were not required to proactively disclose title defects, leaving buyers to conduct their own due diligence. Shortly thereafter, the first title insurance company was founded, operating on the basis of "a group of lawyers reviewing county records to ensure there were no undisclosed claims on the title."
- Current Comparison: Max notes, "Until Doma came along, large traditional title companies were essentially doing these small research projects for every residential transaction — going to the county recorder's office and the assessor's office to confirm there were no outstanding wills or trusts with claims on that property."
- Industry Paradox: Title insurance is also required for refinancing transactions, yet the beneficiary is not the homeowner but the new lender. Max illustrates this absurdity with his own experience — when refinancing the same property with the same bank for a second rate adjustment, he was still charged $1,500 in title insurance premiums. His reaction was: "I'll pay you $50, give you a sworn affidavit under penalty of perjury stating that I haven't incurred any new title claims since the last refinancing — if I'm wrong, just deduct it from my money, don't charge me $1,500."
Theme 2: The Economics of Doma — 80% of Costs Are Labor, Margins Only 10-12%
Max Simkoff breaks down the unit economics of traditional title companies: over 80% of revenue per transaction is consumed by operating costs, the vast majority of which is back-office labor, leaving an average pre-tax profit margin of just 10-12%.
- Revenue structure: The total cost of a home purchase transaction (title insurance + escrow settlement fees) is approximately $2,500-$3,000, priced based on transaction value; a refinancing transaction costs about $1,200 (retail price), while Doma's enterprise clients can get it as low as $700.
- Cost structure: The workload overlap between home purchases and refinancing is roughly 80% (searching records, setting up escrow, scheduling closing dates, signing documents, etc.), but purchase revenue is twice that of refinancing — therefore, purchase transactions generate significant operating leverage.
- Profit margin: Max notes, "Traditional companies spend just over 80% of gross revenue on operating expenses, plus a 3%-5% loss ratio, bringing the combined ratio to 85%-90%, leaving only 10-12 cents pre-tax." The bulk of these operating expenses are back-office labor — checking county records, compiling preliminary title reports, calculating all fees (including loan interest, local settlement taxes, homeowner insurance premiums, etc.), and handling back-and-forth email changes.
- Low customer acquisition cost: Contrary to conventional wisdom, "customer acquisition in this industry is not difficult — most business comes from referrals by lenders and real estate agents, rather than heavy sales and marketing spending."
Theme 3: Predictive Models – From Epidemiology to Computer Vision
Max Simkoff argues that Doma’s core competitive advantage lies in replacing human judgment with data science. Its model inputs are entirely based on property characteristics rather than personal attributes, thereby avoiding discriminatory issues in insurance.
- Prediction Target: When underwriting title insurance, the goal is to predict "the risk of loss from previously undisclosed title claims."
- Data Advantage: Traditional companies conduct exhaustive research on every transaction and record all issues—"They proudly say, 'We did the best research and found all the answers'—and we say, 'Great, we’ll feed those answers into the model.'"
- Examples of Input Features:
- Number of Bedrooms/Bathrooms: In certain areas, properties with 2 bedrooms carry higher risk than those with 1, 3 bedrooms higher than 2, but 4 bedrooms lower than 3—"We don’t care about causality; we only care about correlation. Over 100,000 data points, if this variable has sufficient statistical predictive power, we include it in the model."
- Census Data: The five-year compound annual growth rate (CAGR) of the Housing Price Index (HPI) at the ZIP code level is strongly correlated with the risk of specific categories of title defects. Max emphasizes: "No traditional title examiner, while flipping through a report, would say, 'Let me check the five-year CAGR of the ZIP-code-level HPI'—but we look at this for every transaction."
- Technological Evolution:
- Structured Data: Early on, statistical software like SPSS was used for regression models.
- Unstructured Data (NLP): OCR was applied to extract text from PDF files, followed by natural language models to understand syntax—"Training the model to understand the syntax that title company employees are supposed to understand."
- Computer Vision: The latest application—"Instead of reading text word by word, the model, like the human brain, recognizes the format of a table, the position of shadows, and the signature line at the bottom to identify 'this is a settlement statement,' then extracts fee information based on the geographic location of the text." Result: A task that takes a human 30 minutes is completed by the model in under 3 seconds.
Theme 4: The Entrepreneur’s "Loose Screw" — Refusing to Accept "Impossible"
Max Simkoff describes his core trait as "refusing to accept that anything cannot be achieved within a reasonable timeframe," and believes this is an essential quality for entrepreneurs, though one that needs to be "reined in" by the board.
- Definition of the "loose screw": "I refuse to accept the answer that 'something cannot be achieved in a reasonably short period of time.' Anything."
- The "square peg in a round hole" during the pandemic: On March 13, 2020, Doma had just completed its Series C funding round when the nation entered lockdown, and the real estate market experienced a "cliff-like decline." The company had to decide within 1-2 weeks how to continue investing in R&D without jeopardizing its survival. "We had to lay people off — while also saying, 'We’re going to accelerate spending in another part of the company.' You can’t find out 6-8 weeks later that you made the wrong call and then adjust — that would be a disaster."
- The "successive brick walls" of the startup phase: Doma needed Fannie Mae and Freddie Mac to approve its instant underwriting model, but Fannie Mae’s Selling and Servicing Guide required that "the title insurance company must demonstrate the mortgage is in a first-lien position" and "have a credit rating of at least BBB+." However, rating agencies required "at least two years of operating history" — creating a deadlock. The solution: first convince Fannie Mae that "models can make mistakes, and traditional companies can too, as long as Doma holds more capital," then find a reinsurer to provide an "unlimited balance sheet" as a credit backstop, ultimately securing the rating.
- A double-edged sword: Max admits this trait can push board members "to the point of anger." Lenar’s Executive Chairman, Stuart Miller, once told him: "Max, I’ve enjoyed a lot of success in my career and solved many challenges — but whether it’s a challenge or success, I’ve always insisted on it being 'fun.' What you’re doing today is making it 'not fun' for me right now."
Theme 5: The Next Decade – Deepening the Core Business + Expanding into Two Adjacent Areas
Max Simkoff believes Doma's future strategy is a "dual-track" approach: maintaining a startup's focus on the core business while expanding into two adjacent areas—appraisal and home warranty.
- Core Business: Max still personally reviews the enterprise client pipeline and customer health dashboard on a weekly basis. "I don't look at the top of the list—I look at the bottom. I don't care about clients that are running well; I go straight to 'What is this outlier? Why is this component yellow instead of green?'"
- Adjacent Expansion: Doma has publicly identified two adjacent areas—Appraisal and Home Warranty—"These are problematic, non-digital, friction-filled experiences that sit on either side of the title and closing process. We should own the solution to deliver a truly integrated, horizontal, instant digital homeownership experience."
- Talent Strategy: Max believes the biggest factor for success in 10 years is the team. "If we have the best data science team in the industry, we will build the best data science. We want data scientists to join us and be known for applying their craft in a unique way—if you want to build the most cutting-edge application of computer vision on financial documents, we want you here."
Mentioned Positions
| Position |
Guest Attitude |
Key Data |
| Doma |
Bullish (Founder's Perspective) |
Traditional companies have combined ratios of 85-90%, pre-tax profit margins of 10-12%; Doma's minimum fee for enterprise clients is approximately $700 per transaction; home purchase transaction revenue is $2,500-$3,000, twice that of refinancing, but with 80% overlapping workload |
| Lennar |
Strategic Partner |
Doma acquired its title business in early 2019, bringing in substantial data and cash flow; Lennar Executive Chairman Stuart Miller became a member of Doma's board |
| Chase / PennyMac / Wells Fargo |
Enterprise Clients |
Large centralized lenders that can "switch with one click" to allocate national business volume to Doma |
| Fannie Mae / Freddie Mac |
Regulators/Partners |
Need their approval for Doma's instant underwriting model; Fannie Mae ultimately agreed to accept it under a "controlled pilot" |
| Hippo (Asaf) |
Peer Respect |
Founder Asaf was described by Max as a "highly principled" entrepreneur representative |
| Opendoor (Eric Wu) |
Peer Respect |
Praised for "excellent temperament, remaining calm and kind in the face of major obstacles" |
Judgments Worth Remembering
1. Max Simkoff: The "absurdity" of traditional title insurance lies in the need to repurchase it during refinancing, and the insurance beneficiary is the lender, not the homeowner. Support: A second refinancing of the same property with the same bank still incurs a $1,500 title insurance premium — "I pay you $50, give you a sworn affidavit under penalty of perjury, and if I'm wrong, just deduct the money from me."
2. Max Simkoff: Doma's model inputs are entirely based on property characteristics rather than personal characteristics, avoiding discrimination issues in insurance. Support: Variables such as number of bedrooms and five-year CAGR of zip-code-level HPI — "No traditional title examiner looks at the five-year CAGR of zip-code-level HPI, but we look at it on every single transaction."
3. Max Simkoff: Doma's latest application of computer vision to process financial documents compresses 30 minutes of manual work into under 3 seconds. Support: The model no longer reads text word by word but identifies document types through table formats, shadow positions, and signature lines, much like the human brain, then extracts fee information based on the geographic location of the text.
4. Max Simkoff: Over 80% of traditional title companies' revenue is consumed by operating costs, most of which is back-office labor. Support: Combined ratio of 85–90%, pre-tax profit margin of only 10–12%; purchase transaction revenue is twice that of refinancing, yet 80% of the workload overlaps.
5. Max Simkoff: The entrepreneur's "loose screw" is refusing to accept "impossible," but needs to be "pulled back" by the board. Support: Lenar Executive Chairman Stuart Miller once told him, "You're making me feel unfunny" — Max admits this is where his "genius and dysfunction coexist."
6. Max Simkoff: Doma's "brick wall in series" — needing a credit rating from rating agencies, but the agencies require two years of operating history, while Doma had just been founded. Support: The solution was to first convince Fannie Mae to accept the model, then find a reinsurer to provide an "unlimited balance sheet" as credit backing, and finally obtain the rating.
7. Max Simkoff: The key to success over the next 10 years is talent — "If we have the best data science team in the industry, we will build the best data science." Support: Doma aims to become the top destination for data scientists "known for uniquely applying computer vision to financial documents."
8. Max Simkoff: Hiring is sales — "Time kills all deals." Support: Doma measures the NPS of its hiring process and continuously tracks candidate experience — "If a candidate rejects us because we were a day late in responding, that's our problem."