What this covers

The conversation is between host Patrick and Pat Dorsey, who runs a long-horizon investment firm for endowment and foundation clients. It begins with how such a firm is built. Dorsey declines to manufacture products, citing Morningstar data on the gap between asset-weighted and time-weighted returns. He argues for designing around a ten-year horizon, treating performance as an output of people and process, and weighting analysts' voices equally so jargon doesn't win debates. He also describes his early mistake of imitating other investors, and the circle of competence discipline that replaced it. Risk, for him, means permanent capital impairment rather than a bad month.

The middle of the discussion is about method. Dorsey treats a DCF as a blunt instrument useful for value driver analysis, not for its point estimate. He lays out a 15% IRR hurdle, a higher hurdle for volatile currencies, and the case that growing businesses should be seen as collections of projects. He also argues that operating leverage is where the linear projection fallacy leaves mispricing.

The business examples run across sectors. Aircraft parts broke the firm's thesis when a parting-out market emerged, and Boeing and Airbus show that share is not the same as advantage. Uber's scandal-driven share loss illustrates fungibility, while OpenTable's decline shows that a network effect needs continued investment. Other threads include Workiva's CAC spike, subscription transitions at Adobe and Autodesk, and attention aggregation by Netflix and Facebook. The claim that human capital is undervalued because R&D is expensed is among the more contested.

Sharpest takeaway

Dorsey argues that durable competitive advantage and qualitative, human-capital-driven insight are where markets misprice businesses, and that valuation should be dynamic and matched to the business type rather than mechanically sorted into growth or value.

  • Growing businesses should be valued as dynamic collections of projects, not with a 60-cent-dollar margin of safety suited to static asset businesses
  • Markets price clean, linear historical data efficiently but misprice non-linear items like operating leverage and the hidden value of human capital
  • Competitive advantage must be actively maintained; network effects, brands, and switching costs decay without investment

The argument · threads11 threads · 66 claims
0.85

Valuation should rest on value drivers and scenarios, with discount rates and optionality treated as judgment.

9 pointscentrality 5/5
  • DCFs are blunt tools where one change can swing the output dramatically, so the firm places very little confidence in its point estimate. The DCF's real value is framing the value drivers, identifying what moves the needle, and running scenario analysis.

    “So we actually place very little confidence in our point estimate. What it helps us do, however, is frame what are the value drivers here?”

  • The growth-versus-value classification is not useful to the firm because it does not focus on classic low P/E, low P/B value stocks. But the framework is valuable for investors generally because a business must be analyzed in the context of its economics: a DCF on a no-terminal-value cigar butt is asinine, and dismissing Amazon for a high P/E while it reinvests every dollar at high returns is equally insane.

    “trying to say that Amazon is worthless because the P/E is high, when they're reinvesting every cent of cashflow at a high return on capital against a massive market opportunity, that's equally insane”

  • The firm uses a full three-statement DCF to unpack a business's cash economics and an IRR framework, plus a market-based approach that applies a reasonable multiple to three- and five-year-out free cash flow and EBIT forecasts, with the two methods expected to triangulate. It uses a minimum 15% IRR hurdle on EBIT free cash flow.

    “we use the DCF, full three statement model, as well as an IRR framework. And those two do two very different things.”

  • A firm that does not buy 60-cent dollars should view a business as a dynamic collection of projects. The 60-cent-dollar margin of safety fits static enterprises with hard assets like real estate or oil and gas, but a growing business producing cash through successive projects needs a dynamic valuation view.

    “if you view a business as a dynamic collection of projects, which is what any business is, you cannot look at it as a 60 cent dollar”

  • Valuation should be the last step because no valuation output is worth anything unless its inputs are sound. Analysts should spend most of their time debating margins, growth, working capital efficiency, share-gain pace, and operating leverage, since these are what they can have confidence in, rather than the discount rate or final number.

    “evaluation is always the very last thing we do, because no valuation output is worth squat, unless the inputs are worthwhile”

  • Facebook's DCF fair value excludes any value for WhatsApp, which is not yet monetized. Dorsey treats it as a judgment call: whether it is worth zero or something above zero, and the likelihood that value materializes within the firm's lifetime determines whether to keep holding Facebook.

    “that estimate doesn't incorporate any value for WhatsApp. And then we have to make a qualitative judgment on is WhatsApp board zero or something above zero”

  • Dollar General's value is essentially fixed to its store-growth plan, so trading at fair value tomorrow is simply the outcome. Facebook's value depends on unpriced optionality such as WhatsApp, which makes its valuation more judgment-dependent.

    “Dollar General is not going to suddenly open six times more stores than we suddenly expected”

  • No one, including the speaker, knows the current cost of equity, so discount rate choices such as 8 versus 10 percent are judgment calls rather than precise inputs.

    “Should the discount rate be eight or 10? I mean, neither you nor I, nor the good Lord himself, knows what the current cost of equity is.”

  • Bill Miller's maxim that all the data is in the past but all the value is in the future is the origin of Dorsey's thinking that quantitative information is priced efficiently while value depends on qualitative future insight.

    “All the data's in the past, but all the value is in the future.”

0.71

Switching costs and customer lock-in, built through scale and human capital, create durable compounding moats.

9 pointscentrality 4/5
  • Unlike 20 years ago when TV viewing was zero-sum between networks, mobile devices let people use phones while watching TV, creating extra ad inventory and effectively new hours of attention. Platforms such as Google, Facebook, and Netflix then aggregate that attention, unlike physical distribution businesses that aggregate demand.

    “people have a phone in front of them while they're watching TV, that has in essence created more ad time”

  • Sticky vertical software such as Bloomberg and Advent can exercise pricing power and take an unfair share of economics because users have no real substitute. This creates a pricing umbrella that invites competitors, but raising prices modestly by 2 to 4% a year while improving the product keeps customers from leaving.

    “you can create a pricing umbrella that creates a profit pool for your competitor to come in and try and take it”

  • A business with 6% of a very fragmented market can gain scale by acquiring competitors and consolidating, while the next largest firm has only 2% and is far behind. That gap makes the consolidation flywheel accelerate.

    “that flywheel will only accelerate”

  • Scale lets distributors pool variants: infrequently purchased parts turn more often at a larger distributor, so it carries them more efficiently than a smaller rival. Netflix similarly can afford heavy content spending because it aggregates a large audience to monetize it.

    “The bigger you get, the more frequently your stock of infrequently used widgets will turn”

  • Through primary research with landscapers, the firm tested the intuition that consolidated purchasing saves time that landscapers value. Landscapers said they would pay about 3% more for the same brick if they did not have to drive around, supporting SiteOne's value proposition.

    “Would you pay 3% higher for that brick if you didn't have to drive around? Oh, sure. I would”

  • Competitive advantage is created by human capital through brand construction, network effects, and product designs that are hard to switch away from, not by building factories. Human capital is only valuable if it is monetized; Twitter and Snap have lots of talent and loved products but have struggled to monetize.

    “competitive advantage is created via human capital, is created via the construction and maintenance of a brand”

  • Market share is often conflated with competitive advantage, but it is not the same thing. Boeing and Airbus have high share in a duopoly with high barriers to entry, yet returns on capital are only okay because functionally equivalent planes let airlines switch freely; economics depend on whether a business can differentiate and consolidate a fragmented market.

    “market share is often conflated with competitive advantage, which it's not”

  • The key threat to Facebook's network effect would be declining user engagement, along with spammy ad load, failure to innovate tools like Facebook Live, slowing advertiser tools, or capital misallocation, all of which would undermine monetization.

    “If user engagement went down, I mean, engagement is the cynic one-on of a network effect business, right?”

  • Competitive advantage is the output of human capital creating brands, network effects, and switching costs, so the investor evaluates the product's difficulty to leave rather than headcount or number of engineers.

    “competitive advantage is created via human capital, is created via the construction and maintenance of a brand.”

0.65

Long-term process and talent beat market timing, so the firm should act as a craftsman.

7 pointscentrality 4/5
  • A new asset management firm should first decide who it wants to be in 10 years and design for durability, because its endowment and foundation clients are perpetual entities that require a perpetual timescale. Many founders focus on next year's career instead.

    “know who you want to be when you grow up. Think about what should this firm look like in 10 years”

  • Aircraft parts companies the firm owned showed no linear trend in parts sales despite rising passenger miles. The cause was a parting-out market for old planes, fueled by cheap capital and a wave of relatively new retired aircraft from Boeing and Airbus ramp-ups, which let airlines buy used parts far below new OEM prices and broke the firm's thesis.

    “if you're an airline and you can buy it for X from a dead plane and you have to buy that same part brand new for 10X from the manufacturer, I think that manufacturer sales are going to be weak”

  • Investment performance is an output of people and process, and if either is wrong a firm has lost before starting. Good firms iterate on process, reflect on what fails, and never say we have always done it this way; they also keep voices equal, because analysts who know more jargon can weaponize information and win debates without being right.

    “the performance is an output. Performance is an output of people in process”

  • If the best and brightest people no longer prefer to live in the U.S., the country loses its scarce human capital, which is why Dorsey is concerned about the current immigration debate. He estimates that for decades five or six of ten top people would have picked the U.S.

    “if that goes down, then we at the margin loops because that's the scarce resources, is human capital”

  • A large firm such as Franklin Templeton must act as a manufacturer of products, while a smaller asset manager is more like a craftsman whose core job is curating human capital. Each requires different skills and structures, and neither is inherently good or bad.

    “if you're Franklin Templeton, you have to be a manufacturer, but if you're a smaller asset manager, you're craftsman or craftswoman”

  • Creating investment products in response to investor demand usually means launching at the exact right time for the manager and selling to the exact wrong people, because investors make poor timing decisions as shown by the gap between asset-weighted and time-weighted Morningstar returns. The firm therefore has no ambition to manufacture products.

    “if you're creating a product in response to investor demand, you're probably creating it at the exact right time and selling it to the exact wrong people”

  • Morningstar data on asset-weighted versus time-weighted returns show that investors tend to make poor decisions about when to invest, buying after strong performance and leaving before recoveries.

    “if you look at the Morningstar data on asset weighted versus time-weighted returns, people tend to make pretty bad decisions about when to invest and when not to invest.”

0.47

Concentrated portfolios accept binary outcomes and define risk as permanent capital impairment.

6 pointscentrality 3/5
  • Investing in equities in volatile currencies such as the Brazilian real or South African rand requires a significantly higher hurdle rate than U.S. equities because inflation differentials mean those currencies depreciate against the dollar over two-to-five-year periods. Because clients are dollar-based, the firm must be paid for this currency risk.

    “Our hurdle rate for investing in a Brazilian or South African equity is significantly higher than for a U.S. equity because over time the real and the rand will depreciate”

  • A concentrated portfolio of binary biotech bets has all-or-nothing outcomes, so the manager must understand the science well enough that winners offset losers. Concentration magnifies the consequences of each binary outcome.

    “a concentrated portfolio of biotech bets, where it's like a completely binary upside and downside”

  • Dorsey founded the firm to run an unconstrained, concentrated portfolio because the Morningstar style box is not useful and concentration lets him deploy more human capital per idea. Large endowments and foundations suit this model because they allocate across five to fifteen managers and their interests align.

    “once concentration, because you're going to be one of five, 10, 15 managers, then the interests align”

  • Risk depends on the client: for funds of funds evaluated monthly, risk is a bad month and volatility matters; for Dorsey's firm with long-horizon endowment clients and a five-plus-year underwriting horizon, risk is permanent capital impairment.

    “for us risk is permanent capital impairment”

  • Founding his own asset management firm was Dorsey's biggest leap of faith because he had never invested his own money and publishing research differs from managing money. Friends estimated his odds of success at roughly 20%, and he wishes he had started a decade earlier.

    “Your odds of success are maybe 20%, but that's better than 2%”

  • A small position for a concentrated portfolio like Dorsey's is about five or six percent, which means each holding must be understood deeply because the firm cannot diversify away unknown local risks.

    “a small position for us is five or 6%.”

0.44

Mispricing hides in non-linear operating leverage and qualitative insight, not linear financial projections.

5 pointscentrality 3/5
  • Quantitative information is fairly well priced because databases and computing have made clean data widely available, so markets arbitrage it efficiently. Qualitative insight is more often mispriced, and that is the space where the firm seeks an edge.

    “as databases have become more robust over time... that information should be more efficiently priced. And then that's what we find too.”

  • Clean historical financials that can be linearly projected rarely reveal mispricing because most people also predict linearly. Mispricing shows up in non-linear dynamics such as operating leverage in high-fixed-cost businesses, as with newspapers in the late 1990s that looked great until they did not, and with Facebook and Chegg.

    “we rarely find that if we look at historical financials and they look really clean and neat, and you can linearly project the passion of the future, linearly the equity is not that mispriced, because most people predict linearly”

  • Facebook's R&D as a share of revenue jumped because a large part of the WhatsApp purchase price was allocated to R&D, making linear analysis suggest no operating leverage. Because content creation has near-zero marginal cost, the 30% margin printed in 2015 was not representative, and meaningful operating leverage could be forecast.

    “a lot of the purchase price for WhatsApp had been allocated to R&D. And so R&D spend as a percentage of revenue really jumped up”

  • Chegg's historical financials look awful because they reflect the asset-intensive textbook rental business the company sold. The performance of the asset-light, high-incremental-margin Chegg study business is buried in the segment results.

    “the historical data looks awful and it's because they just sold the business”

  • Human capital is now the scarcest resource, but its output is poorly reflected on financial statements because R&D is expensed rather than capitalized. A rail line or steel mill appears on the balance sheet, while the value of Google's engineers does not.

    “the output of human capital is poorly reflected on financial statements”

0.41

Facebook's moat endures through disciplined spending, user data, and non-fungible social connection.

4 pointscentrality 3/5
  • Facebook projected about $14 billion of CapEx next year, double the prior year, and expense growth of 45 to 60%, without disclosing what the spending is for. Dorsey's firm views this as possibly political cover after congressional scrutiny and would consider itself wrong if expenses ramp ahead of revenue and margins deleverage.

    “Like they say they're going to spend $14 billion next year in CapEx, which is double this last year's CapEx.”

  • Pat Dorsey's firm surveyed digital ad conference attendees on Facebook's return on ad spend versus Google and targeting improvement; about 95% of responses were positive, with one representative quote calling Facebook the platform God would have invented. Facebook's voluntarily disclosed user data gives advertisers measurement that Google's inferred intuition does not.

    “it's rare that you do primary research and the work comes back 95% positive, and that's what we got”

  • Facebook management has shown willingness to accept short-run harm for long-run benefit, such as reducing ad load, even though most of its growth came from raising ad prices. Dorsey views this discipline as a sign of durability, while spammy experiences or stalled advertiser tools would be warning signs.

    “management being willing to take actions that might be harmful in the short run, but are beneficial in the long run, such as moderating ad load”

  • When products are fungible, a brand scandal can cause sudden share loss because customers can switch with no cost: Uber lost 15 to 20 points of U.S. share and Lyft ran a campaign on it. Consumers do not care about the brand of a credit card or car ride, whereas Facebook is less fungible because no analogous platform exists for connecting with people.

    “this will formerly unassailable business suddenly loses 15, 20 points of share in the U.S. because the founder does something dumb and because the products are fungible”

0.38

Competitive advantages like network effects must be continually fed to persist.

3 pointscentrality 3/5
  • A network effect does not persist on its own; it must be cared for and fed through continued investment. OpenTable's share declined because its owner under-invested in tools, while competitors such as Reserve innovated and OpenTable extracted too many rents from users.

    “a network effect has to be cared and fed. Cared for and fed. They don't just exist and perpetuate without anything.”

  • Investors should ask when the knowledge they are generating will pay off, and begin building knowledge now about regions and industries that will matter in five years. Waiting until later means the research will take years to complete when the opportunity arrives.

    “think about when will this knowledge that I'm generating payoff”

  • Workiva's defensibility comes from replicating Excel in the cloud for many simultaneous users with linked data across the enterprise. External reporting teams are small, so a rip-and-replace is feasible, but internal enterprise reporting has a much larger addressable market and deeper switching costs because data links and users multiply.

    “where things get interesting for this business and where the TAM gets much larger is internal reporting”

0.37

Growth requires reinvesting in customer acquisition rather than harvesting the existing client base.

7 pointscentrality 2/5
  • A very high LTV-to-CAC ratio, around five, signals that a company is milking its existing client base rather than acquiring new clients, so it should spend more on acquisition. Diligent Board Books exemplified this: it was proud of being profitable yet was not growing and should have reinvested in its salesforce.

    “there's an upper ceiling at which point that sends a signal that the company is more interested in milking the current client base than growing clients”

  • Workiva has about 96% client retention and 106% revenue retention because it keeps upselling clients. It replaced markup-PDF workflows with Donnelley, growing from zero to about 50% share in roughly six years and reaching about 80% of Fortune 500 external-reporting teams.

    “they have about 96% client retention, 106 revenue retention... they've got about 80% share of the fortune 500 right now”

  • Businesses should focus on the customer who pays the bills rather than on Wall Street or bankers, as Facebook, Google, and Amazon did for a long time. Doing what is right for the customer tends to work out in the long run regardless of what bankers say.

    “take care of your customer and ignore Wall Street”

  • The key SaaS metric is the ratio of customer lifetime value to customer acquisition cost. A delta of three or four times means the company should reinvest every cent of its cash flow rather than generate free cash flow.

    “if you're getting three or 4X delta between those two, you should be reinvesting every cent that you have”

  • Workiva's customer acquisition cost spiked about a year and a half to two years before the interview because it pivoted from external reporting to the Sarbanes-Oxley market, where no product exists and buyers do not budget for a replacement for an internal process. Returning to enterprise internal reporting and reorganizing sales geographically brought CAC back down.

    “There is no SOX product. There is no product for SOX reporting. It's a whole bunch of kluged together, internal processes.”

  • A live quiz app with a large engaged audience could monetize by letting users watch an ad to get a second question or another chance at a lottery-style prize, exploiting the dopamine rush of trying again. Dorsey expects many users would take this option.

    “you only get a chance to play the lottery if you watch this ad”

  • Subscription SaaS models pay the company over time rather than through a large upfront license, increasing the customer lifetime value by 1.5x to 2x but delaying cash. The market often punishes this transition, as with Adobe and Autodesk, though those drawdowns can become opportunities if the product remains sticky.

    “Companies that go from licenses subscription often see a 1.5 to 2X increase in the lifetime value of the customer, but you don't get all the cake today.”

0.30

Investors should stay within their circle of competence, especially for foreign and local-consumer businesses.

5 pointscentrality 2/5
  • After a week in India, Dorsey concluded the firm will probably never invest in businesses tied to local consumer tastes, which it cannot evaluate from afar. Indian exporters that compete globally with global value chains are within its circle of understanding and warrant periodic on-the-ground research.

    “we're probably never going to invest in businesses that are tied to consumer taste in India”

  • Japan's corporate governance and capital allocation are improving, with many global businesses. However, Japanese companies generally prefer in-person visits and rarely take calls, so investors must fly over to make the initial visit that opens future conversations.

    “Japanese companies prefer visits and especially to break the ice.”

  • Dorsey's early mistake was trying to be other investors he was close to, buying things they would have bought. The fix is to define what you are good at, stay within your circle of competence, and suppress fear of missing out on ideas you lack competence to assess.

    “I tried to be other people. I was too influenced by other investors I was close to.”

  • Dorsey admits he did not own Amazon because he did not appreciate AWS's power until recently. AWS may be worth a very significant fraction of Amazon's share price, and Amazon checks every box of the firm's criteria, though he has no good thoughtful reason for not owning it.

    “I didn't appreciate the power of AWS until recently. That's the real reason.”

  • Beyond measurable currency risk, the larger hidden risk for a foreign investment is what the investor does not know because they do not live there. A business tied to local consumer behavior can change without the investor noticing, which is a tall order for a concentrated portfolio.

    “What's not quantifiable, but it's just as much of a risk is what you don't know, because you don't live there.”

0.27

Chegg's rebuilt model of paid curated answers and organic search drives growth and value.

4 pointscentrality 2/5
  • Unlike Facebook, which monetizes user content without paying to create it, Chegg must pay substantial licensing fees to publishers such as Pearson and McGraw-Hill. This gives publishers some leverage, making Chegg a less attractive zero-marginal-cost business than Facebook, though the publishers' own struggles kept renewal costs flat.

    “Now Chegg has to pay money, good money, big money for those licenses to get that content”

  • Students will pay about $14.95 a month for Chegg because the value of a higher grade or passing a required course outweighs the cost, and because they trust a verified answer used by peers more than crowdsourced answers on Reddit. The cost-benefit calculation favors paid, curated content over free alternatives.

    “the marginal benefit of paying 14.95 a month for Chegg and knowing it's the right answer because all your friends have used it”

  • Chegg began as a textbook rental pioneer that Amazon and Barnes & Noble crushed; after the venture-backed founders were ousted, the new CEO recognized that the brand was the main remaining asset. The business was then rebuilt around a digital library of step-by-step answers to end-of-chapter questions, licensed for 27,000 ISBNs and indexed on Google.

    “they have exclusive licenses for 27,000 ISBNs and answered every single question and indexed it on Google”

  • Because most college students copy and paste questions into Google, and Chegg ranks as the first organic result, its user base grew 2.5x in three years while marketing costs stayed flat. Organic search distribution substitutes for paid acquisition.

    “the modal college student today copies and pastes... Chegg comes up as the first organic result, which is how their user-base has gone up 2.5X in three years with marketing costs being the same”

0.24

Understanding businesses requires customer immersion and open debate over winning arguments.

3 pointscentrality 2/5
  • Ray Dalio's Principles reinforced Dorsey's view that a team with diverse opinions should seek the right answer rather than win arguments; the goal is truth, not assigning blame for who was wrong. Dorsey found the book's stance on discussion powerful despite the firms being very different.

    “A discussion is about seeking the truth.”

  • The growth-versus-value classification is not useful to the firm because it does not focus on classic low P/E, low P/B value companies, though it can be useful for investors who need to analyze each business within its own context.

    “trying to say that Amazon is worthless because the P/E is high, when they're reinvesting every cent of cashflow at a high return on capital against a massive market opportunity, that's equally insane”

  • You cannot understand a business without sitting in the customer's shoes, and the best way to do that is to visit industry conferences and talk to participants, as the firm did with an 800-student survey for Chegg and landscaper interviews for SiteOne.

    “You can't understand the business unless you sit in the customer's shoes, and the best way to sit in their shoes is go talk to them.”