Monthly Archives: November 2015

The cost of poor investments

I was recently thinking about the characteristics of strong performing startups and those that perform less well. The goal of the exercise was to identify patterns in the strong and poor performers in order to improve our investment decisions. However, as part of the exercise I discovered that, in addition to the first way of identifying the common traits of successful and less successful startups, there’s a second way to make better investments.

This is by thinking about the cost of making an investment before doing so. The first cost that comes to mind for a poor investment is its financial cost. You lose the dollars that you invest in the company.

While money is certainly an important cost, you can recoup it. Money is made and lost. In fact, you need to take the risk of losing money on certain investments in order to make money on others. Because of this, thinking about the potential loss of money alone when making an investment only goes so far in detracting you from making those investments.

Lost time is a much greater cost than lost money. You can’t make up for the time you lose working on a poor investment. It’s gone forever. And poor investments take up more of an investor’s time than the best performing ones that run smoothly without you.

So if you think about the cost of a poor investment in terms of the time you’ll lose trying to fix things rather than the money you’ll lose, you’re likely to make much better investment decisions.

The underlying assumption here is that you value time more than money. If this isn’t already the case, if you think about it you’ll probably arrive at this conclusion.


Reference checks

I recently led the reference checking process for a C-level hire for one of our startups. In addition to the multiple reference checks which I conducted, two of my colleagues reached out to their network for additional references.

Any individual reference needs to be taken with a grain of salt. The references that a candidate puts forth are likely to be positively biased because the candidate is unlikely to serve up someone who didn’t appreciate their performance as a reference. Similarly, the references of the people who the candidate worked with extensively which aren’t directly put forth by the candidate are often negatively biased. This is because there’s often a reason why the candidate chose to not serve up someone they worked with extensively as a reference.

As a result of these individual biases, the first few reference checks can leave you confused about what a candidate’s strengths and weaknesses really are. There’s more noise than signal after 2 to 3 reference checks.

This is why you need to enter the reference checking process with the clear goal of getting a holistic and accurate picture of the candidate. Getting 2 or 3 references simply because your company’s HR department has this formal requirement isn’t enough. You need to dig deeper. If you do, you soon start to see the signal separating from the noise. You begin to get a comprehensive picture of the candidate with clear strengths and weaknesses.

I can’t prescribe an exact number of reference checks that are necessary to achieve this. You’ll know when it happens. In this specific case, I checked 4 references myself and we checked over 10 references together with my colleagues. And we only stopped when we felt like we had a clear picture of what the candidate would bring to the table, as well as what they lacked. Each candidate has some of both. If you don’t see both sides that means you need more references.

When you have the candidate’s strengths and weaknesses clearly laid out, it’s a lot easier to see whether their strengths warrant overlooking their weaknesses for the specific role you’re looking to fill.

Good founders

Paul Graham who co-founded Y Combinator (YC) together with his then girlfriend, now wife Jessica Livingston recently wrote a post about how Jessica’s involvement was crucial to Y Combinator’s success. I strongly encourage you read the whole piece.

The post carries many important takeaways. One is that although quiet people may not get that much public attention, they may actually be among the greatest contributors to the success of an organization. Another is the importance of women in creating a more considerate and cooperative professional environment.

However, the most important takeaway for me was regarding Jessica’s social radar. Jessica’s ability to identify people with bad character helped YC pick the good people who are most likely to build enduring successful companies. This is a key skill for startup investors. As Paul describes:

“As we later learned, it probably cost us little to reject people whose characters we had doubts about, because how good founders are and how well they do are not orthogonal. If bad founders succeed at all, they tend to sell early. The most successful founders are almost all good.”

Good people don’t succeed more often than bad people in all environments. My belief is that either can succeed in the short run, but that good people win out on a long enough time horizon. If this is true, the fact that startups are long journeys that require years of dedication to build and scale makes them an example of a specific context in which good people have a comparative advantage.

Tough decisions

I recently listened to a New York Times DealBook conference chat with Peter Thiel and Chris Sacca. The full chat is below.

One of Peter’s most successful investments is in Facebook where he was a personal angel investor. The same is true for Chris in Uber where his fund Lowercase Capital was an angel investor.

Both men are very well known for these successful investments. What’s less well known, and which was shared in the chat, is that if Mark Zuckerberg and Travis Kalanick, the founders of Facebook and Uber respectively, had done what Peter and Chris advised them to do in the early stages of their companies, Facebook and Uber wouldn’t be the runaway successes that they are today.

Specifically, Peter advised Mark to accept Yahoo’s $1 billion bid for Facebook back in July 2006. At the time Facebook had $40M in annual revenue. Mark didn’t take Peter’s advice and Facebook is now a nearly $300 billion company. On a side note, one of the reasons why Mark didn’t accept the offer is because, although he would have netted $250 million, he said he wouldn’t know what to do with the money. This is a valuable reminder for many entrepreneurs contemplating early exits.

In the case of Chris, he advised Travis to not take Uber down market where it would compete with taxis on price. Instead, he recommended that it continue to operate with its higher end and more expensive Uber Black service in order to not have to face off with regulators. If Travis had taken Chris’ advice, Uber’s low cost UberX solution which is responsible for the bulk of Uber’s success (in fact, Uber is now trying to further lower prices through UberPool) likely wouldn’t exist.

Both of these examples show that important decisions are very tough to make. With the benefit of hindsight, it’s easy to claim that Facebook clearly shouldn’t have sold out early, and that Uber clearly needed to offer a lower priced transportation alternative to Uber Black. However, in the moment, the decision is never easy.

As these examples show, when it comes to tough decisions, even very intelligent people with aligned incentives can disagree. Sometimes founders are right, and sometimes investors are right. However, it’s ultimately the founder’s responsibility to make the tough decision. As in these cases, hopefully they’ll be right.


The Wall Street Journal recently had a great piece on Sequoia Capital’s Scouts program. Basically, it’s a program whereby Sequoia gives its scouts around $100K a year to invest in promising startups. The scouts consist of entrepreneurs that Sequoia has backed in the past, academics, former Sequoia partners, and anyone else that Sequoia believes may have the ability to identify and access great startups.

Most of the gains from successful investments are shared by the scout making the investment and Sequoia’s LP’s. Sequoia’s GP’s and other scouts only get a small fraction. The fact that the returns from the investment go to the scout rather than Sequoia’s GP’s ensures that the scout has sufficient financial motivation to perform well. The scout essentially gets free money from Sequoia with no downside in the case of a poor investment, but significant upside in the case of a good investment. In this way, the scout effectively becomes the GP of an angel fund where Sequoia is the main LP.

There are no formal strings attached to Sequoia’s capital. Since the investment is made by the scout’s investment vehicle, the startup is free to take capital from whichever source it chooses in the future. However, Sequoia hopes that the scout will repay Sequoia’s goodwill by giving Sequoia priority access to the company’s future larger rounds. Companies like Stripe and Thumbtack, which Sequoia invested in after its scouts Sam Altman and Jason Calacanis referred the companies to them, suggest that the program is working. Companies like Uber, Optimizely, and Zenefits also received original investments through the scout program but Sequoia didn’t participate in their follow-on rounds.

The economics of the scout program also make a lot of sense. The article shares the names of 105 scouts. Some likely no longer serve as scouts, and there are likely others beyond the names on the list. However, assuming that there are ~100 scouts and that each is given $100K a year to invest, that’s a total of $10M per year. In exchange, Sequoia gets the opportunity to find out about great performing startups before other investors, as well as a scout who can vouch for the entrepreneur to take Sequoia’s money over that of other investors. Let’s try to place a value on this.

Sequoia led Stripe’s $18M Series A and Thumbtack’s $12.5M Series B rounds respectively. I don’t know how much Sequoia invested and at what terms, but assuming that each round was for 20% of the company and that Sequoia invested $10M in each round, that’s 11% and 16% of the companies. Stripe and Thumbtack are said to be valued at around $5B and $1.3B right now. Assuming 20% further dilution in each round, Sequoia’s 11% original stake in Stripe is now 4.5% after 4 rounds. Its 16% original stake in Thumbtack is now 8% after 3 rounds. That’s a total paper value of 4.5%*$5B + 8%*$1.3B = $225M + $104M = $329M. These calculations ignore the returns from the follow-on rounds in which Sequoia likely participated to at least retain its pro-rata.

Let’s be very conservative and say that these are the only two return-producing companies that Sequoia invested in as a result of its Scout program over the last 5 years. The actual number is likely much larger. The total cost of the Scout program over 5 years is $50M, and the total investment amount we assumed is an additional $20M. So for a total investment of $70M, Sequoia is sitting on a paper value of $329M. That’s a minimum 4.7X return. This calculation also ignores the paper value of the investments made by the scouts which, although they only produce a small return for Sequoia’s GP’s, are likely shared with its LP’s on the same terms as the core fund.

The scout program makes both logical and economic sense. In startup ecosystems where capital is plenty, having a strong and wide network that gives you priority access to startups is an important competitive advantage.

A guide to marketplaces

The two largest internet company exits to date in Turkey, Yemeksepeti and Gittigidiyor, were both marketplaces. Although marketplaces are more difficult to scale than e-commerce companies since they require up-front investments to simultaneously acquire and activate both sellers and buyers, this initial barrier makes them more defensible at scale.

Our strong belief in marketplaces has led us to invest in multiple marketplace companies in different verticals. These include Bitaksi, Dugun, Buldumbuldum, Modacruz, Hemenkiralik, and Kapgel.

In addition to operating in different verticals, each company’s business differs along dimensions like its order frequency, average order value, take rate, supplier fragmentation, commoditization of the underlying inventory, ability to intercept the payment flow, and many others. VersionOne, a VC fund with a strong emphasis on marketplace investments, has a great primer on different types of marketplace companies and their underlying characteristics here.

High output management

I first read the book High Output Management by former Intel CEO Andy Grove 2 years ago. Since then, I have regularly revisited certain chapters whenever I faced a problem that was addressed in the book.

It is the most practical book that I have read on management. Andy has an amazing ability to solve management problems which appear complex at the surface by identifying the system dynamics which govern these problems, and thinking about each component of the system in isolation and how they interact with one another. He uses this approach to offer practical recommendations in areas like how a manager can increase the output of their organization, and the related question of how a manager should allocate their time across different activities.

Ben Horowitz recently wrote the foreword to the new edition of High Output Management. If you like the examples in the foreword, you’ll love the book.

Distribution vs innovation

I recently read an excellent post by Alex Rampell, a General Partner at Andreessen Horowitz, which highlights the importance of distribution. It’s called Distribution vs. Innovation and its core premise is that “The battle between every startup and incumbent comes down to whether the startup gets distribution before the incumbent gets innovation.”

Sometimes, innovation and distribution happen hand in hand because a product doesn’t need to rely on established distribution channels for its growth. The innovation inherent in the product causes it to spread through a combination of on-platform referrals and off-platform word of mouth. Facebook is a great example of this.

At other times, a specific innovation needs established distribution channels to reach users. In this case, taking the innovation directly to those that control the distribution channels is unlikely to produce a positive outcome for the startup. This is because the distribution channel is harder to build than the innovation. TiVo’s digital video recording innovation which relies on cable companies with access to content for its distribution, and TrialPay’s offers platform which relies on payment processors with access to merchants for its distribution, are great examples of this which Alex highlights in his article. Another example is online education marketplaces partnering with reputable universities with access to certificate or degree-producing educational content in order to achieve their distribution goals.

In those cases where a startup needs established distribution channels to spread its innovation, a startup can capture more value if it can establish its own distribution network before innovating. As Alex highlights, Stripe building a payment processor with merchant relationships is a great example of this. Stripe can now build and distribute its own innovations over its own payment processing network.

It’s arguably more challenging for a startup to build another cable company (although Chamath of Social Capital is trying to build a carrier) or another reputable university than for it to build another payment processor. So building your own distribution network isn’t a strategy that works in all contexts. But if you want to capture more of the value produced by your innovation, the more control you have over your distribution channels the better.

Google Cardboard

I recently received my Google Cardboard virtual reality kit in the mail. I purchased it for $30 at Knox Labs.

Google Cardboard is basically a piece of cardboard that transforms into a set of basic virtual reality goggles when you place your smartphone inside. Once in the virtual reality experience, you move left and right by tilting the goggles and there’s a single button which serves as a click function.

Due to its simplicity, I imagine that Cardboard offers a much more basic virtual reality experience than more advanced goggles like the Samsung Gear VR and Oculus (scheduled for release before the end of 2015 and in the first quarter of 2016 respectively). I have yet to try out the more advanced systems. However, even the Cardboard’s basic experience is enough for a user to grasp the full potential of virtual reality.

My first virtual reality experiences came from browsing through the Cardboard app‘s pre-installed use cases. These include experiences of visiting different cities and walking through different museums. Although not the same as being there in person, these experiences were much better than simply browsing images of the locations on the web.

I also tried a car racing game called Infiniti Driver’s Seat VR. Although it was a short demo experience, it was a lot more fun than the popular Need for Speed series that I used to play on my PC as a kid.

If this is what virtual reality looks like with Cardboard, I can’t wait to experience it with more advanced systems.

Exit interviews with voluntary leavers

I recently had an exit interview with a departing executive from one of our startups. If an executive is fired from their role, such interviews may not be that useful because the employee may have a lot of anger as a result of their involuntary departure. This may bias their communications about what’s working and what isn’t working at the company to the point that these communications aren’t useful.

However, when an employee leaves voluntarily, they don’t carry this anger. As a result, taking the time to understand the reasons for their departure can serve as a valuable source of information for the company to identify what it’s doing well and areas for improvement. An employee may feel uncomfortable sharing this information while they’re still working at the company because they still have skin in the game. However, after they’ve left, there are no potential repercussions from sharing. This makes them much more likely to openly do so.

This was certainly the case during our talk. I learned a lot of valuable information about the inner workings of the company. Although these views represent the perspective of a single person, the lack of bias in their communication relative to the views of existing employees and involuntary leavers makes them very valuable. I strongly recommend that company executives and investors perform exit interviews with voluntary leavers, and debate their findings to improve the performance of their business.