Tag - tech

Anatomy of a Managed Marketplace
In Pursuit of Non-Linear Growth
Debunking The Skill Gap Myth in Daily Fantasy
Multiple Compression In The “Winter” & Why It Matters
The Importance of Being Dumb

Anatomy of a Managed Marketplace

The following article was originally published on Techcrunch on May 25, 2017.

Managed marketplaces (also known as end-to-end or full-stack marketplaces) have been one of the hottest categories of venture investment over the past several years. Recent examples of high-flying managed marketplaces include The RealReal, Opendoor, Beepi, Luxe and thredUP, which have collectively raised nearly a billion dollars. They garner a lot of press because the consumer experiences are often radically different than what’s previously been available in the market.

But there is confusion over what a true “managed” marketplace is. It’s fairly easy to spot a true managed marketplace if you know what you’re looking for. Managed marketplaces typically adhere to the following characteristics:

  • A value-added intermediary (the “management” or “service”) that provides a superior experience versus more traditional peer-to-peer marketplaces, brick and mortar or even a legacy service provider.
  • An introduction of additional risk into the business model; examples might include pre-purchasing and holding inventory or via investing in services related to the buyer/seller that are an incremental, variable cost before any profits have been realized (money goes out before it comes in).
  • A take-rate (gross margin) that is a significant premium versus other buy/sell options in the market in order to offset the premium service level or risk transfer that has occurred.

It’s important to note that many of today’s ubiquitous marketplace companies such as Uber, Airbnb, Grubhub and others are “lightly” managed, by which I mean they invest resources in quality assurance, background checks and verifying reviews. But these services are typically a de minimis expense on the company’s overall operating cost structure — often even considered as part of the customer (or merchant) acquisition cost — and therefore do not classify as a fully managed service.

For Airbnb, these “light” costs might include the costs related to verifying a user’s home address, or the customer service costs of resolving disputes. For Grubhub, the light management might include the costs related to updating menus, but they are not fully managed in that they are not taking ownership of the food or food prep themselves (although Grubhub has begun rolling out delivery, which would qualify as a managed service). This infographic shows the primary marketplace categorizations:

In order to build a successful, sustainable managed marketplace, the take-rate margins must be high enough to support that value-added intermediary and the subsequent amount of services and risk the marketplace is providing. What makes these marketplaces so powerful is that they can drop a comparable amount of contribution margin to the bottom line while investing the higher take-rate revenues into customer experience, reducing friction and product improvements.

Additionally, if, over time, these marketplaces can develop technology that significantly reduces or eliminates the costs of providing these value-added managed services, they can continue to justify higher take-rates and build high-margin businesses that are worth a premium to their traditional service provider comparables or peer-to-peer businesses.

As a primer, here’s a quick chart of take-rates in the re-commerce industry amongst both managed marketplaces and traditional marketplaces. Can you guess which ones are actively managed?

Value innovation and risk

Now that we’ve identified that managed marketplaces are effectively business model innovations, it’s useful to go through a few on a case by case basis to identify each of these innovations and be able to properly identify managed marketplaces in the future.

Opendoor is a managed marketplace in the real estate industry that is an on-demand tool for selling your home. They utilize numerous data sources to offer real-time bids on a home, typically without ever stepping foot in it. Basically: Click a mouse, sell a house. They charge the typical 6 percent brokerage commission plus a risk-adjusted service fee (about 2-3 percent extra, on average, up to 6 percent).

Value-Add Innovation: A consumer no longer has to wait to sell their home. They don’t even so much as have to engage a real estate brokerage. Opendoor reduces the friction of selling a house from possibly months (and multiple showings) down to minutes. The company will also perform all maintenance/changes demanded by a licensed inspector.

Risk Innovation: Unlike brokerages such as RE/MAX or Century 21, which take zero capital risk on a transaction but collect 3 percent from each of the buy/sell sides of a transaction, Opendoor is buying inventory and holding it on their books. The effect of this is that they can offer an extraordinarily differentiated experience to their home sellers, who traditionally rely on peer-to-peer markets (typical MLS listings, with a realtor advising).

Take-Rate: To justify this level of risk (holding inventory) and service (managing maintenance), Opendoor charges a take-rate on average 50 percent higher than in a traditional real estate transaction. While a 50 percent premium may seem marginal given the delta in other categories, the large transaction sizes in real estate mean that the take-rate premium on a $500,000 house is $15,000 of incremental gross margin. That is a significant amount of money to manage maintenance and some risk.

Case in point is the below estimation of Opendoor’s revenue and cost structure on an average $220,000 home (their sweet spot) with a 9 percent brokerage fee, a 50 percent premium to market rates:

Source: Inside Opendoor: What 2 Years of Transactions Tell Us

For Opendoor, when all is said and done, their average net profit, $8,320, or 3.8 percent of the home’s original selling price, is still greater than the 3 percent an agent at a traditional brokerage earns. And, they are able to provide a substantially differentiated experience. It’s a powerful model.

TheRealReal is a managed marketplace in the luxury consignment space focused on clothing, jewelry, handbags, even art. The experience differs from eBay, for example, in that sellers need to provide zero effort other than sending their goods to a TRR warehouse (no photography, no descriptions, no customer interaction) and buyers take comfort in TheRealReal’s quality and authentication services, which they fully guarantee.

Value-Add Innovation: Rather than having to post online listings and photographs of items, pay for a third-party authentication or even deal with shipping, TheRealReal simply collects an item from a consignor and sends them a check once it sells. For sellers, it’s a true “set it and forget it” experience and is multiples more convenient than dealing with online auctions (or even price comparing between local thrift shops).

Risk Innovation: In order to provide a frictionless experience for the seller and an aesthetic, trusted experience for the buyer, TheRealReal is forced to frontload all those costs into their own overhead. They expense per-item charges for photography, copy writing and logistics before an item sells. If the item fails to sell, TheRealReal is forced to eat those overhead costs. Therefore, if they inaccurately forecast demand for certain items, they could end up burning more money than they’re able to recoup on sales.

Take-Rate: In order to justify its cost structure, TheRealReal (and other comparable marketplaces) command take-rates of 30 percent, effectively triple what non-managed, peer-to-peer marketplaces charge as a commission to sellers.

Luxe is a managed marketplace for drivers that reduces all friction associated with parking: finding a lot, searching for a spot, returning to the lot, paying the cashier and waiting to exit. Operating as an effective always-on, mobile valet service, drivers are met at their destination by a Luxe agent who takes the keys and parks a driver’s car. Upon leaving, the driver requests their car in-app and are met by a Luxe agent who delivers their car at their exit point.

Value-Add Innovation: Luxe fundamentally reimagines parking by providing any driver with an on-demand valet who will meet them across a large radius of major cities. Unlike traditional parking or even mobile parking marketplaces such as SpotHero* (which require a driver to park their own cars), Luxe reimagines driving to be destination-focused: drive to your ultimate end-point, not a parking lot. In theory, its product could save drivers time and enable them to avoid inclement weather.

Risk Innovation: In order to provide uninterrupted, on-demand service, Luxe is forced to employ numerous valets across each geography in which it operates. Irrespective of what these valets are actually paid, it is a considerable human capital cost that Luxe is forced to bear ahead of any realized demand. This is in contradistinction to a sharing economy marketplace model such as Airbnb or Uber who are not burdened with human capital costs, but rather pay transactional commissions on any given home-owner or driver.

Take-Rate: In order to counter-act the considerable human capital expense of staffing valets across a city, Luxe should be forced to charge a material premium compared to average hourly parking rates in a particular city. It is therefore quite surprising that they advertise an average of $5/hour for their service, especially when the average hourly rate in NYC, for instance, is $11-15/hour. Several months ago, reports from San Francisco suggested that prior rates of $5/hour have now increased to $15/hour or a $45 daily maximum and, as of this month, they have suspended the valet service. Given the considerable variance in parking costs by neighborhood, it is hard to assess their exact take-rate premium, but I’d estimate it would have had to have been about 200 percent of traditional parking take-rates to be profitable.

Each of these companies is built on the vision that technology will ultimately be able to deliver increased automation and better margins. For example, that one day TheRealReal’s item authentication will be entirely algorithmic or that Luxe will be able to predict the real-time flow of drivers, thereby reducing its human capital costs. Because these representative companies are all still relatively young startups, those tech-driven narratives have mostly only begun to play out.

Beepi versus Carmax

Beepi, a managed marketplace for used cars, recently closed its doors after burning through nearly $300 million in the span of two years. Unlike eBay Motors, which is a peer-to-peer experience, offering no concierge services (although it does offer some self-service options such as free Carfax reports), Beepi was a full-service platform promising rigorous inspections on cars, a 10-day no-questions return policy for a purchased car and, for sellers, a promise that if one’s car didn’t sell in 30 days, Beepi would buy it outright for the appraised value.

The seller fees for this risk-free service? Approximately 9 percent with Beepi… versus a $125 fee for eBay motors, or about 1.25 percent on a $10,000 car, nearly an 800 percent differential.

So with a premium 9 percent take-rate, how did Beepi fail?

The best insight into their failure may come from a similar model with considerable success. It turns out that the nation’s largest retailer of used cars is also arguably one of the most recognized managed marketplaces in the world: Carmax. Give Carmax 30 minutes to inspect your car and they will buy it, even if you’re not purchasing one of theirs, with a “no-haggle,” take it or leave it offer. Thirty minutes is pretty efficient, pretty darn close to on demand.

The extraordinary thing about Carmax is that the Company’s gross profit per used car sold basically doesn’t change even when the year’s average price per car sold moves up or down by 5 percent in any given year. Which means that Carmax is actually less focused on their take-rate per car, but instead focused on their profit per car; their commission is a function of the profit they expect to earn.

Rationally, this makes sense as well — consumers value convenience but have a cognitive dollar limit they are willing to trade for that convenience. By inverting their take-rate to be a function of their profit expectations, Carmax is able to offer more for higher-end cars where a 10 percent difference between the car’s BlueBook value and Carmax’s offer would likely be too extreme for a customer to accept. For high-priced assets, a flat tax is fundamentally unlikely to work.

Used cars are curious assets in that they depreciate so quickly that even 60 days can have a demonstrable impact on value. This is where Carmax excels. In their most recent annual reporting, Carmax notes having sold more than 600,000 cars in the prior year with (then) present inventories at about 55,000. That’s a retail turn of about 11, meaning that a car moves off Carmax’s lot every 35 days or so, allowing them to more accurately price cars andmake higher offers, being less exposed to the less predictable volatility of depreciation.

In a post-mortem on Beepi, Carlypso founder Chris Coleman suggests that in addition to the noted reasons (depreciation effects, and cognitive pricing differential), the approach was inherently flawed from a customer acquisition perspective. Specifically, that while there are customers looking to simply sell their car for cash, most car owners are looking to trade-in a car, because they still need a car and there are tax benefits to doing so; a platform that has to pay marketing costs for both the buyer and seller in all transactions is at a significant disadvantage.

At the end of the day, a managed marketplace model for used cars does work. Carmax is only one of thousands of proof points: tens of thousands of dealerships across the country hold inventory, inspect cars and reap a profit. Beepi’s failings appear to be the result of poor execution, mispricings and maybe even some bad luck around a financing that fell through.

Automation and Shutterstock

Because managed marketplaces involve a heavy “service” component to improve the overall experience, one of the expectations of the sector is that as artificial intelligence and automation continue to evolve, the human capital cost of providing the service will decrease if software can assume more of those responsibilities.

But an area of struggle with managed marketplaces is that very few digital managed marketplaces are actually public companies, reducing the visibility into their overall economics and processes — and making it hard to test the assumption that service costs should come down over time. Luckily, there’s at least one: Shutterstock (NYSE: SSTK), a marketplace for photographers to sell their images, bills itself as a “trusted, actively managed marketplace,” in that “each image is individually examined by [their] team of trained reviewers.”

On the spectrum of managed marketplaces, Shutterstock is undoubtedly on the lighter end — with the financial risk from its active management being only the human capital cost of its QA reviewers. Nevertheless, it would seem like this hypothesis is an ideal one to test on a company such as Shutterstock, which isn’t dependent on an unproven technology such as self-driving cars to reduce their cost of providing a service, but could presumably leverage proven, inexpensive image recognition technologies to do much of the quality assurance, copyright detection and tagging that the human reviewers do.

Yet, that doesn’t appear to be borne out by Shutterstock’s financials. To test the automation hypothesis, I decided to look at the company’s revenue versus the cost to generate that revenue. Shutterstock defines their costs of revenue as “royalties paid to contributors, credit card processing fees, content review costs, customer service expenses, infrastructure and hosting costs…and associated employee compensation.” I would assume that credit card processing fees as a percentage of revenue are relatively flat (if not slightly reduced year over year) and that cloud-hosting fees also scale mostly proportionately to demand (revenue). I’ve defined “True COGS” below as the aforementioned expenses to providing their service, minus the contributor royalties:

Surprisingly, rather than decreasing over time, these True COG costs appear to be increasing. Meaning that the same picture that used to take 10.5 percent of revenues to process now costs nearly 15 percent.

There are a number of possible explanations here. It’s certainly possible that these increased costs are because the company is investing heavily into automation, the effects of which simply haven’t been borne out yet while they streamline their QA process. It’s also possible that the number of photos the company maintains makes it more expensive to process each incremental picture — for any variety of reasons.

The learning from this Shutterstock case study, a company which is now 14 years old, is that it’s improper to simply assume that the substantial service-related costs that managed marketplaces incur in their early stages will decrease with “scale,” either through execution or software automation. As with any company, there is always room for improvement, but the above analysis of Shutterstock would imply that it’s nowhere near as easy as flipping a switch.


Managed marketplaces are a quintessential venture investment, allowing entrepreneurs to recast consumer experiences while leveraging venture capital subsidies to hold much of the risk inherent in these managed models.

From a unit economic perspective, the potential automation of much of the service labor that goes into these platforms could be significant. Investors and operators need to remain sensitive that it is ultimately the technology, not heavy services, that will long-term cultivate highly desired business models and margins. But, that future automation could also lower the barriers and defensibility of these companies, allowing peer-to-peer players to launch a comparable offering with similar software.

In my mind, the sustaining managed marketplaces will not only re-imagine the experience they’re approaching, but be focused from the outset on building a data moat around their product, thereby ensuring that they remain the platform of choice, even if software innovation begins to level the overall playing field.

Special thanks to Josh Breinlinger and Rebecca Kaden for their feedback on this article.

*Chicago Ventures is an investor in SpotHero.

In Pursuit of Non-Linear Growth

It’s hard to imagine much can be added to the discussion around marketplace theory and metrics. And yet there’s a concept – an obvious one – that rarely seems to be referenced – and is one of the main drivers of our investment thesis at Chicago Ventures: searching for non-linear growth.

It’s a characteristic that exists at Kapow, Shiftgig, Spothero and Curbside in our portfolio, amongst others and is extraordinarily powerful in early stage marketplace growth.

The Core Theory:

From what we’ve observed, non-linear growth is typically spurred by relationships on one side of the marketplace that allow it to launch and scale seamlessly across multiple new markets with compounding efficiency. In our experience (the companies listed above) that works best in geo-scaling marketplaces – because of the successive new market launches – where instead of building a new market from scratch in each new geography, one half of the market comes already solved at launch.

So for example:

  • As Kapow launches its real-time corporate events booking marketplaces in new cities, its demand side is pre-stimulated. How so? Because it works with large enterprises such as Google, HP, Deloitte, etc who boast dozens of regional offices, it can launch new cities featuring those clients with minimal sales effort.
  • As Shiftgig rolls out its on-demand temp labor booking marketplace from city to city, its demand side is also pre-stimulated by large national clients, for example a national big box retailer or hospitality provider that has temp needs in dozens of cities across the US.
  • Spothero is able to turn on new cities effectively overnight because of its enterprise partnerships with the largest parking operators in the country. While new entrants might have to put boots on the street in each new city to try and sign independent lot operators, Spothero can aggregate a large supply of inventory from its existing partners in the top cities in the country.

The benefit to this model is that it allows each new market to launch with incremental traction above each prior market launch. For example, here’s a breakout of one of our marketplace investments and each market growth (numbers have been changed slightly to protect confidentiality):


This graph reflects months from launch to hitting an arbitrary revenue milestone, represented by the bold line. In this case, it took market 1 over 11 months and took market 4 only 2 months.

Let’s be clear: every marketplace wants to show a similar graph (reduced time to X) whenever they’re launching new markets – whether that be new geographies or new verticals, whatever. And many are able to show that, even starting from a base of zero because their playbooks have become well refined and each incremental launch has a faster growth rate than the prior.

But by betting on non-linear growers, we feel that companies can reduce the risk incumbent in new market launches because they start from a non-zero base.

By aggregating these markets together, you yield a growth curve that begins to look extremely compelling:


And each successive market launch will start at a compoundingly higher base – enabling the growth curve to bend closer and closer to a 90-degree angle.

When you understand this, it may not surprise you that the majority of our marketplace investments have some b2b component – simply because few customers can scale purchasing or selling across multiple markets (whether that is geographies or verticals). Even consumer facing companies such as Spothero and Curbside, whose end customer is a mass-consumer, have strong b2b relationships on the supply side, with lot operators and big box retailers, respectively.

Umm….so what about Uber?

Yeah – it’s a good point. Airbnb too. The two greatest digital marketplaces of the past decade are both built on seemingly linear growth (in that there was no supplier or buyer that could drive outsized liquidity in new city launches).

When I considered this issue for some time, I came to the following two observations:

  1. Both of these platforms benefitted from unprecedented levels of virality and PR. And they deserve all the credit in the world for that. The consequence of that virality is, for example, that in Uber’s case, they were able to methodically selected new market launches by analyzing latent demand – the # of times the app had been opened in geographies where they weren’t operational. Understood this way, they actually weren’t starting from a base of zero at all because there was latent, pre-stimulated demand in each new geo (whose customers simply needed an e-mail or notification to alert them of the launch).
  2. Both of these platforms enabled micro-entrepreneurship in that an enterprising supplier could purchase cars for others to drive or property managers could list multiple properties (or purchase others specifically for Airbnb) so that the supply side was not growing on a 1+1 incremental basis.

With these understandings, Uber/Airbnb actually are non-linear growers, and their greatness is that they achieved that from a harder to scale base.

But no matter how you look at it, it’s hard to build a liquid market. At Chicago Ventures, we feel there’s a benefit – or even a greater margin of error, so to speak – by focusing on companies that can pre-solve for one half of the market. And that said, I’d give anything to be an investor in Uber 🙂

Debunking The Skill Gap Myth in Daily Fantasy

Daily Fantasy Sports are under a full on assault: allegations of “insider trading,” a hyper aggressive (if not overwhelming/annoying) advertising strategy, and multi-directional claims that DFS can’t be beat because it involves too much skill.

There are lots of legitimate concerns which I will address in future posts over the coming weeks. But the most damning – that DFS suffers from “too much skill” – is inherently flawed.

Here’s why:

Too Much Skill

A McKinsey study from September is making headlines after it noted that 1.3% of DFS players – the sharks – account for 91% of the winnings.


One of my favorite daily reads, Ben Thompson of the newsletter Stratechery, used that data to note the following:

In fact, as the McKinsey article concludes, the fact that Daily Fantasy does require skill is one of the biggest threats to DraftKings and FanDuel: the flipside of a few players earning most of the money is that there must be a counterweight — some number of big fish willing to lose and lose substantially.

Here’s the thing: while these numbers make for a great click-bait headline, in actuality it’s a non-issue. Why? Because betting markets fundamentally demand winners and losers.

One could make the claim that in a fairer world, the winners would win less and the losers would lose less. But let’s unpack that: for starters, the headlines are deeply skewing the numbers. While it’s true that the “Big fish” are losing 44x more than the “Minnows,” they are also playing stakes 66x higher. Keeping in mind that Fanduel spreads daily games with $1,035 buyins, the idea that affluent bettors regularly playing $300-$1k buyin lose $1,000 over the course of “half an MLB season” is not that surprising. If anything, based on my experience, it appears light.

Let’s also not forget that for the “minnows” who lose a full 50% of the money they deposit – $25 of $49 deposited – that wagering on DFS is entertainment. It’s a little known fact that traffic to online poker sites during the late 2000s was inversely correlated to new airings of American Idol: when the show came on across time zones, traffic would accordingly drop in those geographies. Low stakes betting is entertainment. And given the low buyin amounts (down to $1) I’d be shocked if, on an hourly basis, it were more economical for these players to go to the movies than lose at DFS.

I had a former member of the Full Tilt Poker Board of Directors in my office a couple of weeks ago. I asked him to verify the often-cited statistic that approximately 95% of online poker players were losers. He confirmed this. Going back to those McKinsey results…a full 15% of DFS players appear to winning – currently 3x the rate of online poker players.

But let’s take a step back and all agree on something else: DFS is in its absolute infancy. The level of scrutiny being placed on DFS, especially given its youth, is unparalleled by any other betting market (skill or pure gambling) in the history of the world. That means that a lot of the less salacious factual realities – though entirely normal in the adolescent development of any market – are being distorted as structural flaws. When in reality they are fundamental pre-requisites to building a healthy ecosystem.

As always, I draw a lot of inspiration from my years in the poker industry. And here’s what I observed: when markets are imbalanced, with strong financial incentives on either side, unless those markets demand some genetic prerequisite to entry, the financial opportunity will move those markets closer to equilibrium.

And here’s what I mean – in plan English. As an example, there are millions of brilliant, hungry, deeply incentivized college student with lots of disposable time and some disposable income. Many will experiment with DFS. Most will lose; some, however, will recognize that only 1.3% of players are using advanced statistical modeling and will hunker down for weeks building their own models until they start winning. The effect of this is that win-rates for the top 1.3% of players will begin to decrease. And instead of 1.3% of players winning 91% of all the payouts, the market will move towards 5-10% (my guess) winning 91% of the payouts.

The Historical Ecosystem Cycle:

In a recent Bloomberg Masters in Business podcast featuring Nate Silver, the most frequently used word when responding to questioning was: “historically.” History is paramount. And in the case of DFS, we have precedent of ecosystem evolution by looking at the history of financial markets, poker, and gaming. As noted my analysis is primarily influenced by the evolution of online poker.


Two major trends will begin to develop: (a) A formalized, professional ecosystem of third party DFS training apps will begin to emerge, (b) The sites themselves will invest in educating players and normalizing skill levels. Case in point is that we’ve already begun to see this with Fanduel’s acquisition of Numberfire.

Here’s a breakdown of what really happens in these stages –

Stage One – Early Adopters:

  • DFS is currently holding somewhere between Stage One and Stage Two.
  • Early adopters – those who understand the game theory of lineup assembly and experimented early have a significant skill edge in DFS. The winners have benefitted greatly from the influx of players to the ecosystem.
  • An increased focus on DFS – between its media coverage, heavy advertising, and mainstream profiles of large winners (especially if they can build a spectator/televised component) – will cultivate an aspirational user base and begin to inspire players of all skill levels to improve their lineup performance.

Stage Two – Third Party Apps:

  • Consumer grade player analytical software and lineup generators will begin to make their way to market (I would know I’ve been pitched on several).
  • Subscription training sites, founded by pro DFS players will begin to emerge offering unique insights into lineup formation (I would know I helped to build Cardrunners – a similar business in the poker world). RotoAcademy is one example but there will be more.
  • Peer to peer coaching marketplaces connecting DFS pros directly with recreational players willing to invest in their lineups will emerge.

Stage Three – Platform Sponsored Education:

  • As platform growth slows because market expansion slows, the site operators will re-prioritize on maximizing customer revenues rather than raw customer acquisition.
  • The platform operators, cognizant of the increased revenues/rake as average skill level improves will launch their own efforts to draw the ecosystem closer to golden mean (where everyone breaks even long-term).
  • The operators will either: (a) Launch their own, recreational player focused, free to learn, educational platforms just like Full Tilt Poker Academy, (b) Will subsidize learning on third party applications such as Truly Free Poker Training and/or (c) Purchase late night programming slots on mainstream television channels such as Fox Sports to air informercial-like educational programming.
  • The effect of all this sponsored education is that the baseline performance of even low volume recreational players increases.

Stage Four – Platform Defection:

  • With the skill gap closing across the board two important shifts begin to take place:
  • Lesser winning players, who had previously fed off utter amateurs, no longer find it time-effective or profitable to drive volume on the platform. They are economically incentivized elsewhere and will leave the platform.
  • The winningest players on the platform who have built bankrolls of winnings in the millions of dollars will no longer be able to drive large ROIs on the platform because the stakes are not high enough. They will divert their attention off platform, either to professional sportsbooks or private markets that open.
  • These two defections serve to actually increase the skill gap on the platform as the winningest players and breakeven or lightly winning players leave the platform. In an ideal world where the market continues to grow and customer acquisition remains stable, this shift actually creates an opening for the next generation of players to move from “minnow” to “shark,” replicating the cycle of the ecosystem.

The Structural Issues:

There are legitimate structural claims against DFS, most importantly that novice players receive no protection from high stakes sharks in that multi-entry allows for lineups to be replicated without friction across buyins.

That said, this blog assumes (and I am taking for granted) that those structural issues (which are real) can be solved, either by limiting the number of entries per day (as Fanduel has done) or by creating a more dynamic pricing system or by eliminating identical lineups. Or, in reality, by some solution I haven’t even considered.

I don’t know how they’ll be solved, but I believe they can be. The purpose of this blog is to disprove the detractors who believe DFS’s skill gap is permanent. It is not. It will evolve.

Multiple Compression In The “Winter” & Why It Matters

To say that today was a rollercoaster would be an understatement. While I recognize I’m a relative novice to the financial markets, watching fast selloffs and panic selling is always mesmerizing – no matter how many times one has seen it.

I don’t know how Bill Gurley always seems to have impeccable timing, but last Thursday night (before Friday’s market pullback which started this whole mess) he ripped off an 8 piece tweetstorm where he warned that recent corrections in the public markets could have an adverse impact on the tech startup world. Re/code chronicled those thoughts in a piece entitled Winter is Coming, You Ready? Within the tweetstorm, Gurley made the following two points:

At the close of today’s markets, I decided to do a quick analysis to see how true that is. Especially with a few of the stocks I know very well: marketplaces such as Etsy, Grubhub, Shutterstock & Homeaway. As some of you may recall, I penned Unpacking Etsy’s S1 on TechCrunch a few days ahead of its IPO back in April. Here’s how those stocks stacked up back then on a GMV, Revenue & EBITDA basis –

Recall that on the day of its IPO, $ETSY actually spiked to an Enterprise Value of over $3 billion dollars so imagine an EV/GMV of close to 2x and an EV/Revenue of nearly 15x.

I re-ran the numbers at the close of today’s market and updated the company financials to reflect their trailing twelve months (2nd half of 2014 & 1st half of 2015) as opposed to just 2014. Here’s what they look like now:

As is evident, Gurley is 100% correct. Its not these companies aren’t growing (see below) its that their multiples have been cut by 50-65%. Only Homeaway seems to have avoided the carnage, most likely because it is the eldest, most profitable, with the strongest balance sheet. At the same time, to those of us in the tech world, it is the stock most likely to be entirely disrupted – by Airbnb – but no matter. The market is looking for stability, not growth. You can see for yourself: the market is applying nearly identical multiples to both Etsy & Homeaway – two companies at very different stages of growth and maturity.

Here’s why this matters: Most VCs (including Chicago Ventures) have typical rules of thumb we like to apply to companies when considering an investment that have (at least roughly) some correlation to public market comparable. For example for marketplace businesses, we’ve historically looked for a valuation to be anywhere from 1.5-3x annualized GMV. For an e-commerce business it might be similar. Its dependent on growth rates, and we expect to pay higher than public market multiples because our startups are growing 2-10x year/year, not 60%. But as public market multiples come in, we’re forced to reevaluate. For instance, if we’re considering a Shutterstock clone (bad example, but you get the point) and we’re forced to say – “Well, it really only has $800M market potential, not $2.5 billion potential” – then it drastically affects entry price.

How long current multiples will last is anyone’s guess – or whether they’ll have a snap correction. But one thing seems clear to me right now: my generation of younger VCs who were trained to assess top line GMV growth first will soon be forced to master other, more fundamental aspects of a business’ economics.

For those keeping score at home, here’s my assessment of those 4 public co’s trailing twelve month financials. I moved quickly to get this out the door before the market close & there could well be errors:


The Importance of Being Dumb

Stupid is as stupid does.

–       Tom Hanks, “Forest Gump,” 1994

Nearly every VC firm in today’s market is focused on showcasing their brilliance: white-papers, 50 slide powerpoint decks, marketing departments and prescient theses that can be referenced for credibility. In fact, when I’ve met with elder VCs to seek mentorship, I’ve often been asked some variant of “what makes you a good picker of companies?” The implication is that I need to prove I’m smarter than the herd.

I’m now convinced that’s wrong. Great investors don’t need to be smarter: they need to be dumber.

To be more nuanced, one needs to be entirely comfortable in looking dumb – in feeling comfortable being wrong. Great firms (venture firms, startups, big cos, etc etc) will be cultivated when team members are comfortable being vulnerable enough to say dumb things, make dumb decisions, yet find collaborative support even when they are wrong.

At first glance, this approach might appear to be simply a question of a firm’s risk tolerance (should we make this investment?!? It seems too dumb to work!). But I think it runs far deeper – coming face to face with cognitive dissonance itself. Several months back, Roy Bahat, founder of Bloomberg Beta (a seed fund I respect immensely) noted to me that their investment process was rooted in a culture that focused on mitigating the cognitive bias of fearing looking dumb:

“A lot of this has to do with the psychology of the team. Peoples’ need to “feel smart” and, more importantly, avoiding looking dumb… that drives a lot. In our structure, it’s designed for me to be dumb. As long as *someone* around the table is smart enough on each deal to say yes, we win.”

Early in my tenure as a professional poker player, I realized that I was far more adventurous, creative, and risk-taking while playing online poker than I was while playing in casinos (my winnings online ultimately exceeded those in a casino by a factor of probably 1,000x). Given that my opponents in a casino were on average actually worse than my online competition, and I was focused on only a single table in a casino as opposed to 4-12 simultaneous games online, I was miffed. I identified two causes of this incongruity:

  1. I would keep a very large bankroll in my online poker accounts and this large number gave me the confidence to make mistakes knowing that I had ample reserves. In a casino, I rarely kept more than a couple of buyins in my pocket at any time and often felt cash constrained.
  2. Online,  I was largely anonymized. I could experiment with moves, run crazy bluffs, and trust my instincts to make daring call-downs knowing that I would never be blushing out of embarrassment, that no one would ever “ask to see my cards,” or berate me for my decisions.

The truth is that 99% of the time my play in both worlds looked identical. But the outlying 1% made all the difference – the confidence to tap inner creatively, reimagine a situation, and make a decision entirely antithetical to the standard move in a certain context. The greatest players I ever encountered: Tom Dwan, Jason Strasser, Vanessa Selbst, Phil Ivey, just to name a few, became great because in those moments where 99.9% of players would follow a formulaic, auto-pilot approach of the “optimal play,” they would catch themselves and ask: how would my opponent react if I did something entirely novel and unexpected? What would happen if I changed the assumptions underlying what is “optimal?”

They were willing and confident to look dumb because they instinctively knew you could never be great by being slightly smarter or slightly more optimal than the field. Being slightly smarter will undoubtedly win money – but it won’t yield greatness. To be fundamentally great you need to be fundamentally different. Simply put: you need to look dumb.

The underlying question is how one gets there.

Greg McKeown writing in what is probably my favorite book of the past several years, Essentialism, stresses the importance of playfulness to drive creativity:

Our modern school system, born in the Industrial Revolution, has removed the leisure – and much of the pleasure – out of learning. Sir Ken Robinson, who has made the study of creativity in schools his life’s work, has observed that instead of fueling creativity through play, schools can actually kill it: “We have sold ourselves into a fast model of education, and it’s impoverishing our spirit and our energies as much as fast food is depleting our physical bodies… Imagination is the source of every form of human achievement. And it’s the one thing I believe we are systematically jeopardizing in the way we educate our children and ourselves.” The idea that play is trivial stays with us as we reach adulthood and only becomes more ingrained as we enter the workplace.

When we play, we are engaged in the purest expression of our humanity, the truest expression of our individuality…Play expands our minds in ways that allow us to explore: to germinate new ideas or see old ideas in a new light. It makes us more inquisitive, more attuned to novelty, more engaged…Play broadens the range of options available to us. It helps us see possibilities we wouldn’t otherwise have seen and make connections we would otherwise not have made. It opens our minds and broadens our perspective. It helps us challenge old assumptions and makes us more receptive to untested ideas. It gives us permission to expand our own stream of consciousness and come up with new stories. Or as Albert Einstein once said: “When I examine myself and my methods of thought, I come to the conclusion that the gift of fantasy has meant more to me than my talent for absorbing positive knowledge.”

When more traditional thinkers look at startup culture – stocked pantries, ping pong tables, arcade games, frequent parties, kegs in office, open space for chatting – and express confusion, or even disdain, they are missing its root function: to cultivate fun; to cultivate play. There many ways to keep employees happy – health benefits, lenient vacation policies, etc. And those are also undoubtedly important. But they don’t promote playfulness within an office setting. Playfulness will drive creativity – and will drive extraordinary thinking.

It took me years to realize it but what made my friends so darn good at poker is that they were always having so much fun when they played. Looking back, I now realize that I played my best poker when I stopped trying to be technically perfect – and instead focused on the pleasure of interpreting puzzles and analyzing challenging situations.

When a group of people are engaged in creative thinking, they’re all, intrinsically, exposing themselves – their ideas, their thoughts, their essence. And the vast majority of ideas expressed in a creative setting are rejected – frankly, because they’re dumb. The catch is that the winning ideas may well actually be smart (although they’re often controversial) – yet they can only emerge in an environment where team members are comfortable being vulnerable, being dumb.

The firms and companies that can cultivate this culture have a shot at greatness. All others are playing for second place.

Copyright © 2014. Created by Meks. Powered by WordPress.