Category: Startups

When David Ulevitch tweeted a screenshot of this post by Lead Edge Capital, I had a laugh. It’s quite true that companies report the best metrics they can conjure.

When taken from the perspective of an investor in public companies, this commentary is understandable. We all remember constructs like WeWork’s Community Adjusted EBITDA.

From the perspective of an early-stage entrepreneur, however, there are many unknowns. Good cash profits may be an end-state for the business, but other leading indicators must be used to understand and communicate the reality of the business and its future.

Lead Edge describes metrics well, though a business goes beyond its metrics. Imagine a list of layers from 1 (the most lagging indicator, cash profit) to n (the most leading indicator of all, the vision and initiative of the founder). It may look roughly like:

1. Cash profit
2. Revenue
3. Rate of revenue change over time
4. Product development that will secure future revenue
5. Velocity of product development that will lead to greater capability
6. Quality of team that will execute that product velocity
7. Ability of team to hire great teammates
… and so on …
n. Vision and initiative of the founder

It’s crucial to remember that this is an information problem. Just because an iconic company like Amazon once had no profit (and at the very beginning, no revenue or product!) does not mean that it won’t amount to something great in the end.

For an outsider looking in, the more layers deep they can accurately observe, the more accurate their predictions about future outcomes will become. In theory, with a proper understanding of each layer 1→n, investors, hires, and founders alike could see the future, if not for unexpected external events such as changes to a market, technological changes, or economic factors.

A fun exercise may be assigning a “lag time” to each layer. For a generic startup, perhaps cash profit is a decade lagging from action taken today. It takes a while for team quality to become product quality, to become sales quality, and so on. Perhaps revenue is a year or two lagging from product development.

The investment process of one firm that focuses on hypothetical layers 1-5 would have a strictly longer “lag time” having their assessments (in)validated than a firm that figures out how to properly see and analyze layers 6-7 as well, as long as both firms are on par in terms of decision-making capability.

Many of the entrepreneurs I work with are well aware of this reality. The best investors will offer the most attractive terms when they can look towards the nth layer and see the core of the organism that is a business for what it really is. Likewise, the best hires will want to join something special as early as possible if they can pull back the curtain and see what’s about to happen. Deeper understanding of each layer leads to efficient markets, and helps everybody while hurting nobody.

In the long run, entrepreneurs benefit from the robust nature of “nth layer conviction.” If an investor backs a startup because of its growth metrics, the investor may lose conviction if circumstances change. But if the conviction comes from the nth layer, the founders themselves, support networks will tend to have more enduring support through the inevitable bumps in the road.

The responsibility to strive towards “nth layer thought” is shouldered by everyone. Entrepreneurs should take great pain to communicate the entire stack to those who can hear and synthesize the message.

Given a stack that’s well-communicated by entrepreneurs, investors need to act on information that’s more qualitative than quantitative in nature. This comes in a number of forms, for example:

• Aligning decision-making process — ultimately one investor on a team is likely to have 10x the visibility into the deeper layers of a company. Giving that one investor leeway to act on that information is critical, since the qualitative data is harder to communicate than the surface level quantitative metrics. For example: Benchmark famously has the partnership vote on deals, but the lead partner still has the authority to invest even if the rest of the partnership does not reach a majority Yes vote. The partners trust each other to know the truth and act on it when there’s an intangible spark that’s hard for newly-introduced colleagues to understand.
• Spending time with teams more than data rooms — there is certainly room for innovation in the “relationship business.” The venture world tends to rely on rotating guests of dinner guests to mix and match potential future founders. But observing founders beyond formal events and pitch meetings is an art that has yet to be mastered.
• Building deep context long before funding events — at Contrary, we take great pain to build deep personal relationships with founders years before they start companies. We’ve even made large growth investments in late-stage unicorns having known the founder since they were an undergraduate student. Building the formal systems to engage with founders, executive teams, and early hires personally requires a degree of preparation and longevity to be successful, but when it works, the information advantage is profound.

Luckily there is strong financial incentive for an investor to gain an edge. We will never go back to the era where truly exceptional companies would receive a single term-sheet in the early days. The venture ecosystem has significantly professionalized (and capitalized) itself over the past decade. Modern investors need to either 1) generate superior relationships and power to win deals that are known to be high-quality by others, or 2) generate the information advantage to build higher conviction earlier. Failing to do one of these will lose deals, and lose money.

What better way to win than by being a Nth Layer Investor rather than a 1st Layer Investor?

How should we think about competitive moats for companies like Spotify that aren’t quite marketplaces yet still have data and multiple parties involved? Is defensibility (or lack thereof) inherent to the market structure, or can you create stickiness out of thin air? Is there another category of defensive strategy yet to be fully explored?

For Spotify, the traditional network effects model doesn’t quite work as it would for Facebook or Twitch. Sure, you have consumers on one end that incentivize labels to add more content to the platform. You can also add original content, or build things like playlists that are mildly high-friction to transfer.

But labels can raise prices until they capture most of the value. Competing platforms can offer the same content at a different price. And switching-cost fundamentally doesn’t get stronger with scale. In fact, product friction often incurs a tradeoff between usability and stickiness.

Spotify ultimately leans heavily on product design, branding, loyalty, and other forms of “weak moats”.

But I think there’s a whole set of stickiness strategies that companies big and small can exploit to raise switching costs and slowly eat more and more of the user stack.

Example 1: “blank slating”

We’ll stay with music streaming products for now.

The data you generate when playing music can only be so useful: Spotify can personalize playlists to fit your tastes, but Apple Music can easily do so too, especially if you import your playlists. Both companies have large data sets to make recommendation engines good. As much as you think you’re a musical snowflake, most people’s tastes aren’t actually that unique.

But what Apple Music will never have is the list of songs you have saved but have not listened to in a while. So Spotify can always hand you a radio station full of “favorites you haven’t listened to since last year” that perpetually surfaces novel yet nostalgic content.

This relies on data that cannot carry over between competing products. It’s a tree rooted within the bounds of the walled garden. If you switch services, that entire set of data insights becomes a blank slate.

Example 2: “stored data”

Literally every talking point I’ve heard about Superhuman relates to 1) speed, 2) keyboard shortcuts, or 3) the “luxury software” effect.

You may assume that I’m another VC Superhuman evangelist, but I’m surprisingly lukewarm user. Many apps like Spark have plenty of keyboard shortcuts and add-on features. I find myself spamming the ESC key more than I should, and I really wish I could have all accounts in one view like I can with other apps.

But this post isn’t about the pros and cons of different email clients. It’s about untangling which features actually contribute to a business model that’s unbeatable in the long-run.

The most salient elements of Superhuman’s moat are the ease of use, favorable aesthetic/UI, counterparty profile panel, and handy features like read receipts, reminders, and send later.

Unfortunately speed, aesthetic, and simple features are not very defensible. Spark is already ahead on half these dimensions, plus they have team functionality.

What is defensible, though, is the long list of emails you’d like to be reminded about and the emails you currently have queued to send at a future date. Followups and executing planned outbound are both very important.

“Stored data” must be painstakingly carried over to another service. Or slowly transitioned over time. In either case, the user just doesn’t want to make a switch.

If I were on the product team at Superhuman, I’d be doing everything I could to make Reminders and Send Later first-class citizens. Both in the typical email use-case, but also in the “Note-to-self On Steroids” use case.

The beauty of this stickiness-driver is that it’s pulled out of thin air. There’s no trade-off to be made.

(At this point you might be thinking “wait, what about the exclusivity? The invite system! The elusive prestige of a private product!” Sure. The status/signaling element is valuable right now. Just as it was for Facebook when FB only existed at Ivy League schools. But that sort of moat is more of a growth-hack than a form of long-term defensibility. Nobody thinks FB is elite or exclusive anymore. It’s the opposite.)

Lessons learned

I’ve had a tough time coming up with original examples above. At the concept’s core: there are sometimes small niches where you can increase product stickiness without making a friction/usability tradeoff. That’s what contributes to defensibility where there is otherwise not much.

Facebook might inherently be a network-effects driven social product. Spotify and Superhuman have to work a little bit for it. But it can be built!

Whether it’s the “blank slating” approach, the “stored data” approach, or something else entirely, your job as a founder or PM is to find subtle ways to create product-driven stickiness out of thin air.

Every once in a while, Silicon Valley is ridiculed for reinventing the wheel. You’d think there are only so many wheels left to reinvent, but apparently we’re still in the first-bite phase of software eating the world.

There are plenty of examples. Dropbox famously reinvented FTP servers. Lyft reinvented public transportation. Kindle reinvented libraries. Soylent reinvented, well… Soylent. People-free this time.

Why do pessimists think this way? Of course there’s general ignorance. But the pattern of criticism seems too specific to be random. It’s also lazy to assume that Valley outsiders “just don’t get it.”

So what subtleties are lie behind the curtain? There are a handful of relevant things I can imagine:

• Aversion to changing norms
• Dispersion of existing power hierarchies
• Setting precedent, e.g. about public vs private services
• Cost of switching platforms/infrastructure

As I’ll mention later, each of these can be reverse-engineered to find ways to solve problems and capture value. But first, let’s walk through some examples of these ulterior motives playing out.

Lambda School is the most recent and striking example of such criticisms. Check out this hit piece published by none other than The Guardian.

It’s so interesting because there’s some grain of truth to each gripe, even if it’s not coherently argued by the journalists covering it.

On changing norms: LS is guilty of the same crime as the Thiel Fellowship. Normalizing alternative career paths and dropping out is scary to a lot of people. Although it can be a positive signal in the Valley, it’s rightfully a strong anti-signal elsewhere.

On dispersion of power hierarchies: not only are universities and modern hiring pipelines vulnerable to LS-style disruption, literally everyone with a college degree stands to lose out too because such credentials are largely (but not entirely) zero sum. Read Bryan Caplan’s excellent book The Case Against Education for more on this. Even if its title is gratuitously provocative.

On setting precedent: LS, much like Airbnb and Lyft, is a lightning rod for statist vs individualist conflict. It’s the charter schools debate playing out at the college level. You’d be reasonable to assume that LS-style thinking will seep into many parts of society, and that could be very good or very bad depending on your viewpoint. For one of those groups, it’s best to explain this away as “nothing new here — ISAs already exists in state-run form in Germany.”

On switching cost: If you’re enrolled in college or sourcing talent based on university talent or credentials, your life simply gets harder. Either you switch to LS which incurs sunk-cost and hassle, or you don’t and you might be on a suboptimal path.

My point is that both operators and investors should think more deeply about the “reinventing the wheel” trope. It’s a hint about the structure of underlying incentives.

I don’t think any of this is intrinsically good or bad. I do however think it’s perfectly natural.

There are a couple different lessons you could draw from this post. On the surface level, visceral gut-reactions are probably more rational than you’d otherwise think. That’s important to consider when building and communicating a product.

Taking it one step further, you can follow these patterns to find wheels actually worth reinventing (meaningful startup ideas). This framework can help you filter through such ideas. Otherwise you’d just waste your time tearing down Chesterton’s fence everywhere you look.

Addendum: The Pessimists Archive is an excellent podcast that dovetails well with these themes.

This is a followup post. See Part 1: the Product and Market!

“Founders matter most” is one of the most prolific philosophies in early stage VC. Most investors agree that the best companies are started by incredible people with ideas that aren’t so exciting at first glance.

Airbnb is one of the canonical examples here. Notice that most rejections cited the product or market. It’s possible that Team was the issue and some investors lied to not burn bridges, but I don’t think that’s likely given PG’s recommendations and the team’s storytelling ability.

So how should we think about early-stage teams when the product or market haven’t proved themselves yet? Here are a few things I look for. You can use this as a scorecard or checklist:

Deliberate learning. How does a founder take feedback? Are they coachable? Will they adjust to feedback in real-time? Best of all, do they aggressively seek out feedback when none is offered?

Recruiting ability. There are many ways that you can build a “recruiting unique value prop.” People like Saku are smart enough that I’d work for them to access novel, interesting ideas at the office. Other founders (think Steve Jobs) sell a magnetic vision. Some have a track record or network that makes success seem certain. Many strategies work. It’s a question of whether or not the founder is exceptional enough at whatever strategy fits them and the business.

Urgency/GSD/raw energy. Smart people can fall into the trap of being too intellectual rather than action-driven. Making the right decisions for the right reasons is obviously important, but at the end of the day, founders need to push hard and go fast for years.

Being very blunt or forward. It’s always a “no” when you don’t ask. Blunt and forward personalities tend to get straight to the core issue, put more shots on goal, and find common ground more quickly in my experience.

Making use of tools. Does the founder use software or people to be relentlessly resourceful? Have they hacked anything to their advantage? Do they use a relationship manager to be amazing at followups? Do they send over a link to mystartup.com/deck instead of a 40 character long Google Drive link? Have they automated energy-draining activities by scraping websites for sales leads? Reinventing the wheel indicates many many bad things like inflexibility and lack of prioritization. Founders should focus on their comparative advantage.

Being good (but not necessarily great) at everything. Hiring for “spikiness” is common advice. I think looking for what’s exceptional about someone and evaluating them on that dimension generally works well assuming they meet the minimum hustle/culture bar. But founders quickly become “editors” more so than “writers” as their companies scale. A strong intuition across all fields helps identify talent and align the company’s various specialists. This is partially why having a complementary cofounder is so important. (To be clear: founders should still be spiky, but those spikes should ideally be additive rather than trade-offs with other traits.)

Founder-market fit. People aren’t blank slates. What experience and personality traits are best for this business and team? Marketplaces may need hustlers, for example. Does the founder understand this and consider it in team-building plans? One big challenge is determining whether this company is the right fit or whether the founder is just a hammer that sees a nail. (E.g. I know the music space really well and I like math so I’m gonna start a music analytics company. Is this good, or is it just the random-chance combo of the things they happen to know best?

Are the cofounders ready to marry each other? This was alluded to above, but team dynamics matter a lot. Is there a complementarity? What happens if they get a divorce? A surprising number of companies have done very well despite cofounder relationships ending.

Do I want to partner with this person for 10 years? Investors (hopefully) spend a lot of time working with founders as a sounding board, advice-giver, and connector. If you’d have a hard time working with a founder for whatever reason, it’s probably not worth making the commitment.

What did I forget to mention? What metrics or litmus tests do you use to think about founding teams? Say hi! Email/Twitter on homepage.

Update: I made a video explaining these trends in more detail!

From The Palace to League of Legends, social spaces on the internet are defined in part by the technology supporting the medium. There are a number of trends that will reshape online social spaces in the next several years:

• Everyone now has phones in their pockets at all times
• Voice and facial recognition is becoming more mainstream
• AR and 5G will enable more content rich apps
• Norms around IRL interaction are changing (e.g. Pokemon Go. More on that here.)
• Better batteries and graphic chips make mobile streaming and gaming more reasonable
• Infrastructure and dev tools for mobile apps, streaming, and games are getting much better at all levels of the stack

We’re seeing early hints of these coming together with Fortnite, for example.

We can take a look at Fortnite’s predecessors to better understand how things have changed. Call of Duty, Halo, Battlefield, etc. had very similar features. 5-10 yrs ago, friendships were forged remotely over private lobbies, zombies, boosting, team-based S&D, deathmatch, co-op story mode, etc. You’d go home from school alone (or with one friend for split-screen) and rendezvous with everyone else online. Fortnite’s ‘hangout spot’ effect is really nothing new. But it is special, in my view, for three reasons:

1) Seamless cross-platform functionality. I don’t know how Fortnite was built, specifically. But it’s amazingly scalable. The ability to play on your computer, Xbox, or iPad removes location/situation as a barrier. Plus you don’t have to deal with the entry costs of a full console. This means there’s no bifurcation between Xbox Live and PSN, or more generally, mobile and console. In the past, you couldn’t play with friends unless they were on the same ecosystem — try convincing mom to buy you a whole new console. Fortnite overcomes this large barrier by working well everywhere by default.

2) Relatively simple mechanics. First Person Shooters are usually super complex. The cartoon-like mechanics lower the computational load (good for cross-platform engineering!) and soften the learning curve. Reducing hardware constraints and making the theme more mass-appealing grows the addressable market. Plus a larger segment of the market can actually compete from day one. This bumps retention and virality.

3) Better customization with skins, emotes, etc. The OG FPS games had basic customizations like 4-letter “Clan Tags” that let you identify with a group or “Clan” of players. There was limited ability to make your gun a different color or your character look different. This was a brilliant yet nascent way to encourage and support “hangout spot” socialization. Fortnite takes this to another level with the numerous skins, weapons, and emotes you can earn or buy. You’re not just using the gun or clan-tag your friends have as you would in the OG FPS games. In Fortnite, you’re expressing yourself with an individual aesthetic that you design. That’s a powerful way to build engagement.

I think Fortnite is a great example of distilling the best of past titles into one simple, focused game. The monetization and design is eerily reminiscent of pure-mobile games too. Adding an in-game currency (V-Bucks) that distances USD from in-game goods is a classic mobile revenue strategy. In the past, console and mobile were pretty distinct ecosystems with hugely different economics and distribution systems. It looks like the two will begin to merge. I expect Fortnite to be one of the first, but not the last!

But Fortnite is just one example. There are a lot of levers to pull when it comes to online spaces: accessibility, identity, ease-of-use, social norms, the possibilities of the medium, the focus created by a central activity/goal, and the things you can signal can all influence the way that users interact online.

I expect the traditional and mobile distribution and monetization models to continue merging, and new or underused (e.g. Houseparty or group FaceTime) social spaces to continue picking up users. That’s partly why I’m so bullish on tools like Discord — some media don’t support “social space functionality” but are still great to build community around. Discord adds that social space layer.

I wonder whether we’ll converge into one unified, app-enabled social space a la the Facebook/Oculus/Snow Crash vision. The trends listed above makes me think that we’ll tend towards a variety of smaller single-purpose spaces with richer content. This is supported by the current social media landscape. Each big social space — Reddit, Twitter, Snapchat, Facebook, iMessage, FaceTime, Instagram — owns a unique “job to be done.” Current fragmentation is a feature, not a bug. More context on that concept here. We haven’t converged into the single unified platform that some thought we would 20+ years ago, and I don’t see anything that will change that going forward.

Given that I spend most of my time nowadays working with first-time founders and students interested in venture, I figured I’d write a post covering my common advice on getting really good. If anything, this serves as an exploration of trade-offs I’ve dealt with at one point or another.

Although I learned the most exciting and nuanced tips and tricks from talking to great people, there’s still a ton that you can do on your own. I’m constantly amazed at how young some top startup nerds are. There’s so much content out there that you can get learn surprisingly fast if you separate the signal from the noise. Apologies in advance if this turns into a sales pitch on joining a startup and focusing on getting really good at one thing.

Principles

As is true in everything, there are many different ways to be really good. In general, I think we bias towards covering our weaknesses and being well-rounded. Especially if you have a strong deliberate practice/growth mindset. But if you want to be really good, it often helps to be the absolute best at one or two things, and merely acceptable at the rest. You could spend most of your time building a great network. Or becoming a top-tier salesperson. Or learning the detailed history of Silicon Valley and the empirical results of what works when building startups. It doesn’t really matter what you choose. But it’s incredibly hard to do everything when so young.

One useful strategy here is to prioritize the skills with a strong snowball or halo effect. If you get very knowledgeable about one specific vertical, you can be known as “the X guy/girl” for your X of choice. Then you have an excuse to talk to people and build cool things. Or if you focus on building a network of smart, like-minded people, it’ll get much easier as time goes on. Having an already strong network makes future networking easier and frees up time for other pursuits.

I think there’s not much choice to be had in your comparative advantage. You should do what you do enjoy the most. I can’t stand “networking” so I focus on what I do best: reading, learning, and expressing interesting viewpoints to friends. Building a reputation around knowing a little about everything has made the other relevant startup/venture skills easier to get.

Read the right materials

I’m a big fan of Semil Shah’s post about context as a tool for having good ideas and operating effectively. Most knowledge is subtle and not specifically written down.

The best way to get context is to follow current and historical thought. I’d read through all the classics: Paul Graham, YC, a16z, First Search, Founders at Work, Hacker News, tech Twitter, AVC, Above the Crowd, Chris Dixon, Brad Feld, Elad Gil, 20 Minute VC, and many other that I’m forgetting.

The raw volume of freely available content is humbling. I used to be bummed that I got rejected from MIT, but there’s a silver lining that I’m extremely grateful for: I have a huge amount of time to focus on stuff beyond coursework. Of course CS@Illinois is no walk in the park but I have a lot more flexibility to sit down and consume whatever info I want. Being careful about 1) minimizing “gen ed” time sinks that could be replaced with short books and 2) cutting out extracurriculars I’m not in love with has made a huge difference in productivity and happiness.

With a small number of brillant exceptions (Jason Lemkin on Quora, Vitalik Buterin on Medium, to name a couple), everything on Hackernoon, Quora, CNET, Business Insider, etc. isn’t worth the time. I find that Techcrunch, WSJ, Recode, etc. are fine for keeping track of big picture stuff, but you don’t learn much from them. Newsletters like LAUNCH ticker, Term Sheet, and StrictlyVC are better.

I’m a staunch supporter of learning things completely outside of your field. Not just “tech, but in a different vertical.” Really out there. Think political science, biology, military history, ethics, law, and comedy. More on the importance of this here.

Starting something

This is definitely the best way to learn every skill you need in entrepreneurship and venture. I’ve always known that I have more of an investor personality, but literally all of my mentors told me not to go straight into venture. I was nearly stubborn enough to ignore them but luckily I found the ideal startup internship and jumped on that opportunity.

Let me say: I’m amazed at how much I learned in just three months. It really is the fastest way to learn. I’d try to go as late-stage as possible as long as you can do a little bit of everything. The startup I was at had 7 people. I’m very fortunate to have done everyone’s job at some point.

To take a step beyond my own personal experience, there are a few specific characteristics to unpack from the generic term “startup.”

The speed of execution helps keep things challenging and interesting. There’s more room to experiment, the feedback loops are tighter, and you’re forced to overcome perfectionism and ship quickly.

The extra responsibility of a startup role adds a very qualitative sense of purpose or necessity. At a big tech company, you don’t need to worry about most of the little details, but those tiny parts of the business add up quickly. The utter lack of a safety net or acceptable “not my problem” scenario is a forcing function. You’ll have to think hard about asking the right questions and prioritizing.

Finally, there’s an enormous value to working on small teams that don’t build narratives. Outside of a company, all you know are the stories. It really is mind-blowing how revisionist or over-simplified most info is. Part of this is to build prestige among outsiders. It sounds better if the challenges you overcame were scary and hairy. It’s also because some things like interpersonal conflict and doubt of the company mission can’t be shared externally. That’d violate norms or expose confidential info. By being on the inside of the machine, you know all the juicy details of what’s happening. That was much more valuable than I had anticipated.

If you want to start something of your own, working at a startup is a no-brainer. Hop between startups each summer (and each semester part-time!) If you want to go into venture, I do think the empathy and tactics you learn are worth the time doing something other than your eventual goal. The street-cred doesn’t hurt either.

Investing

One of my major learnings over the past couple years at Contrary has been the value of getting as much real practice as possible. If you’re asking yourself “is this founder in the top 10%, 1%, or 0.01% in terms of clarity of vision?” you need to have enough data points to draw a normal distribution in your head and decide how much of an outlier the founder is.

This also applies to metrics. Is a 5% week over week growth rate good or bad? Is there a big difference in benchmarks between verticals? What CAC is reasonable for this sort of product? The answers to these questions depend on experience and are hard to observe from the outside of the decision-making room.

Talking to a very large number of founders also helps you understand the edge-cases that pattern matching and percentile-ranking don’t cover. What if a founder is a management consultant who taught themselves to code, and is building a social app which has historically been a questionable category? How do you begin to reason from first principles if you don’t have much info to go off of? You develop a valuable intuition for different situations by talking to as many people as possible.

Another thing to optimize for is the “legitness” of the investing you’re doing. Other founders/investors have suggested building a Fantasy Portfolio where you choose startups you know about and pretend that you invested. This is fairly popular in the public markets. It works better there because you have access to much of the same information as other investors, and you always have the option of investing in public markets. A Fantasy Portfolio helps prove that you can find companies, but convincing them to take your money is a different ballgame.

Through a handful clubs or nationwide student organizations, you can give out grants or invest small sums of ~20k. While these are a good way to build a network and meet founders, you can’t learn as much on the decision-making or value-add side. Working on a returns-driven team with larger check sizes (ideally over 6 figures) forces rigor. Just as exams force you to study and thoroughly learn the material in a way that lectures and homework can’t.

Of course I’m biased, but I think Contrary is the best way to do that while still in school. It’s much better than working remotely for a venture fund doing diligence work, in my opinion. Joining a VC in the summer would be best, but I don’t think it outweighs the opportunity cost of joining a startup when school’s out.

I was recently given a few interesting questions on big-picture tech and investing principles. Here are some thoughts:

On big tech co defensibility

I think many investors over-weight for the theoretical aspects of a moat and under-weight practical ones. To be fair, good investors use mental-models to understand and generalize concepts. But it’s very difficult to discount an idea that is conceptually elegant though flawed in detail.

Network effects are the top-of-mind example here. In my view, Facebook — the poster-child of network effects — isn’t nearly as defensible as other tech giants that have ecosystem effects (Google search is much better partly because of data harvested from peripheral businesses like Android and Gmail) or brand equity (Coca Cola is otherwise indistinguishable from Pepsi).

A trite, tweet-sized rebuttal to FB’s moat would sound something like “how did MySpace’s network effects turn out? Or AOL’s, or Yahoo’s?” While NFX are great when you have them, they disappear just as fast once the flywheel starts spinning in the other direction.

I can imagine FB’s core social utility being split off into a number of separate products. We saw this with Instagram before FB acquired them. We saw Snap fill the private/ephemeral need, and not even FB’s massive network effect could save Poke from massive failure.

I also think FB is acutely aware of this: they’re investing more in video streaming to lock content-creators to the platform. They’re also drawing inspiration from WeChat with Messenger by tying in tools to make it a platform rather than a social network feature.

Mark Zuckerberg is one of the last people I’d want to bet against. But I can say the same for Bezos, Page, Brin, etc.

On the most important characteristics of founding teams:

Of course the ability to attract talented hires and being hard-working are useful traits. Solving a burning problem in a large and growing market is great too. Just as being tall is useful to a pro basketball player. I view these more as prerequisites, not forward-looking predictors of success.

Many investors and operators alike seem to favor category creators — companies that build something totally new. Zero to One, as Thiel would say.

But historically, many of the best companies were not the first to the punch. Going by market cap: Apple did not build the first computer or smartphone, Amazon was not the first to sell things online, Microsoft wasn’t the first to make an OS, Google was just a clone of Alta Vista, and Facebook was like MySpace or Friendster.

Even smaller companies like Flexport or Quora can be glibly described as “freight forwarding, but with software” or “Yahoo answers, but with good content.”

To be clear: this does not (and should not!) diminish the achievements or impact of any tech companies. I’m actually humbled by how transformative incremental changes can be. I see this as a testament to the power of pure focus and insightful iteration on the learnings of others.

There are probably other characteristics higher on the list of greatness principal-components. But this is one of the most underrated and misunderstood in my opinion.

On evaluating teams:

First, I look for something wrong with the team. Not to find a reason to pass, rather to make sure that they’re not “too perfect” or cookie-cutter. At risk of sounding like an armchair philosopher, founders with a vision for something truly innovative will look a little strange or contrarian. Founders simply following trends generally opt for more traditional or “prestigious” career paths before starting a company. As a quick aside, I think this is what tech people get at when they use the “don’t hire people who went to business school” mantra as a shorthand for describing outsider-ist tech culture.

Second, I try to understand how the team will continue to attract stellar people both inside and outside of their current areas of expertise. There are a lot of potential levers to pull here. Perhaps everyone is super cool or brilliant (or both!) and other great people will want to just be around the team. Or maybe the founders are respected engineers who only have street cred with other engineers, for example. Then how will they recruit and retain salespeople?

Third, I want the team to be thoughtful about the cultural diversity vs tribalism paradox. Especially in the early stages, teams need to be able to move quickly, be agile, and have laser-focus on the vision. To get through the early-stage grind where you spend all your time with the same people, it helps to be “like-minded.” As you grow, however, that can incubate serious problems (Uber) and you’ll have a harder time cultivating different viewpoints and skill-sets. Ideally founders have thought through this and will be able to handle the team culture through different stages of the company lifecycle. You could even generalize this to being deliberate about scale and knowing how to learn about management.

There just isn’t enough enough time in the day to take advantage of every opportunity. Below I’ve listed some ideas that I wish I could start in another life. I’d invest in all of them if the right team is committed. Let me know what you think! I could dig into these ideas for hours.

Online mega church

Senior citizens are often lonely, immobile, and religious. They’re also very loyal customers. You can read more of my learnings about seniors here.

Think of how successful some televangelists have been. And they just sell videos. Now that seniors are quickly adopting mobile devices, you can layer on Netflix-like content selection, social features to build community (and network effects!), on-demand streaming, interactive components, in-app donations, etc. Just as middle schoolers can practically live their social lives in Fortnite teams or CoD lobbies, on online religious gathering place could be the killer app for seniors.

My thinking on this space has many layers I haven’t mentioned here, and I am extremely interested in advising or investing in companies in this space. Reach out if this is relevant to you — seriously. I have a couple ideas on distribution as well…

A usable prediction market

I don’t understand why there aren’t more prediction markets (primer on the concept here). The prediction markets that do exist — the financial markets — are incredibly impactful and employ millions of people. Interesting PM use cases include prediction-based corporate governance structures, better predictions of real-world events, and more accurate securitization of private risk.

I was excited by the launch of Augur, but nobody really uses it. I haven’t worked out why Augur — or a centralized service for that matter — hasn’t caught on. Sports betting is a technically a form of prediction markets and that’s a huge industry. I see no reason people wouldn’t enjoy on similar real-life events (will Kanye run for president and get more than 500k votes?) I think that a prediction market with a Robinhood-esque ease-of-use or social features could get big. (Perhaps this is something that Robinhood should just build on their own!)

This could take the form of a consumer-focused product, or a toolkit for companies to integrate prediction markets into their own products. (Beware, this is a slippery slope leading to Futarchy. Kidding.)

Better tools for remote teams

Zoom, Slack, and G-Suite have made managing remote teams much easier over the past several years. I can’t imagine leading Contrary without such convenient ways to work and communicate. But there are still unsolved problems. Documenting everything for your team through a wiki or messenger is slow and painful enough to not get done most of the time. And you can’t look over at someone’s desk to see if they have a minute to chat. We need better information collection (some companies have attempted auto-transcription of meetings) and less ambiguous/distracting ways to get someone’s attention remotely.

The holy grail of remote tech is feeling like you’re with someone IRL. No matter how much time I spend with Contrary’s venture partners over calls or text, the friendships don’t feel quite right until we meet in person. When seeing someone in-person I’ll feel like we’ve been friends for a while. But in-person interaction is that final stepping stone. This seems to be an artifact of how we’re built as humans. Getting around that somehow would be huge since team culture is so important.

Tools or a platform for job offer negotiation

Most college kids don’t negotiate their job offers. I think that’s a colossal mistake because 1) wages tend to be sticky so a marginal difference in starting salary will add up over a career, and 2) an extra few thousand dollars saved now will compound into 5x that by the time the students’ future kids are growing up. Every little bit of cash matters early on in life.

I think there are a couple causes of this problem. College students may not know that they can negotiate offers, or they may be too timid or anxious about risking their offer. Even though in reality, nothing ever happens to the offer and employers won’t think less of you as long as you’re not overtly greedy. Or maybe students are just burned out after 4 years of hoop jumping and want to put the job search behind them.

I don’t know what the best model would be. Career coaching is available to most students, but nobody uses it and it’s got a mediocre reputation. The school cares about getting you in a job, not optimizing those little comp details (similar to the real estate agent incentive problem). There are a couple product angles you could take. First, you could offer an end-to-end job hunting service that closely coaches you on what to say to recruiters (don’t be the first to name a salary number! Blame a family member for needing time to think and compare!). You’d charge a percentage of bonuses you can negotiate to de-risk from the user’s perspective. Second, you could create a library of example emails, Glassdoor-like negotiation benchmarks (some companies are more flexible than others), and a community for navigating the post-offer process which is crowded out by resources for the job-seeking process. Paid access to such a library would be clearly ROI-positive from a user perspective in theory, but I bet it’d be hard to market in practice.

The long term vision would be to slowly eat more and more of the recruiting “stack” and eventually match users with companies. You already know how Indeed, WayUp, etc. work. I think there’s an opportunity to build lasting value-add by being a uniquely trusted brand that is on the job-seeker’s side from the start (helping with negotiations), and by having a better understanding of why employees do or do not accept an offer since you’re involved in that process.

High quality food vending machines

Consumers want 1) convenient food, 2) better quality than fast food, and 3) more unique or “local” eating experiences. The rise of fast-casual, food trucks, and single-purpose micro-shops supports these trends. But there are a couple problems you run into in the food industry that make restaurants bad fits for venture. Margins are low, the businesses are not scalable, and the moats are dependent on brand and food. I want to take the food truck craze one step further with automated vending machines that cook food within them. You’d start with simple (but restaurant-quality) foods like dumplings. The machine would have an internal steamer to cook the dumplings on the spot. To maintain freshness, you’d have a full-time employee prep ingredients and deliver them to all of the vending locations. You could expand into pizza, eggs + hashbrowns, shawarma, etc.

The key thing here is keeping quality up — most vending machine food is bad. But if customers would walk up to a machine on their morning commute and get fast-casual quality meals, you may even be able to charge a premium. Of course there would still be a cost challenge: the machines would be expensive, and you still need to hire someone to do constant maintenance and restocking. That’s mitigated by the lack of labor cost on a per-machine basis. This would be really hard to grow, but I think the market is so big that the slog would be worth it.

A robust way to signal conformity and conscientious  

I’m a fan of Bryan Caplan’s The Case Against Education. It argues a viewpoint that may seem obviously wrong or even upsetting, but is very difficult to actually find steel-man refutations of. To oversimplify one component of the book’s thesis, education signals three things: intelligence, conformity, and conscientiousness. Of course we have the SAT or IQ tests to show off intelligence. But you can’t show a potential employer that you’re conformist enough to be an obedient employee and conscientiousness enough to get the job done.

That’s a big part of why we spend tens or hundreds of thousands of dollars and 4 prime years on education. If there were a stronger, more honest signal of conformity/conscientiousness, then you could potentially skip the brute force signalling that makes up much (most?) of modern education.

(If you’re thinking “but what about everything you learn?” then I agree that’s also part of school, but go read the book. There are some incredible empirical results that may sway your viewpoint.)

I’m purposely leaving this proposal vague. There are a lot of different angles you could try. Perhaps a company should build a stronger reputation system that tracks detailed feedback across employers (hiring managers rarely call references beyond your most recent job). Or you could build a more comprehensive marshmallow test for adults. Or something totally different that I’m not alluding to here.

A socially acceptable, well-branded nootropic

I’m worried that nootropic vitamins and supplements will go the way of Google Glass, Segway, and Soylent: useful products, but something you’re a little too embarrassed to use. Cutting edge research-based companies tend to have very utilitarian marketing strategies (WTF is a ketone ester and why do I need it?) and the nootropic/supplement/IMF/biohacking space isn’t really “cool” or consumer-ready yet. I assumed that the company formerly known as Nootrobox, now HVMN, would be the ones to make it happen. Maybe timing isn’t right, or their execution was off.

Consumers seem to gladly regulate themselves using drugs (coffee, alcohol, and nicotine, specifically) so I don’t think there’s anything fundamental preventing nootropics from happening. Vitamins are common too. I can’t explain why drugs and vitamins haven’t been combined into a mainstream nootropic. It has to be the marketing. (Semi-related: Kin is a startup I’m excited about).

A dating app for hyper-targeted user-defined subcommunities

There are two competing trends I’ve observed in the dating space: 1) an absurd number of niche dating services have small communities, and 2) it’s difficult to filter for what you actually care about on the popular platforms like Tinder or OkCupid. Match Group, the owner of Tinder, Match.com, and OkCupid, has tried to cover the major user segments at the brand level. Is there a way to cover more user segments and give users more choice at the app level?

Nearly all new dating startups are cookie-cutter and terrible. I’m interested in a platform that solves the filtering problem by offering a marketplace of user-generated subcommunities, a la Reddit. There would be an “everyone pool” or master-list of profiles to sift through, paired with the option to “join a circle” such as tall people or neoliberals who enjoy the theatre. I’m skeptical of top-down data science that matches people together based on survey questions — I think letting users generate their own taxonomy could lead to interesting results and potentially form a lasting differentiation. The default “all” list helps users join the platform, and the unique subcommunities lock users in.

What is it that makes software startups so able to disrupt incumbents? Why does competition seemingly matter less in software compared to other industries?

Software companies are very maneuverable. Especially new entrants. There are relatively small CapEx requirements to maintain existing products so pivoting and expanding doesn’t require as much of a operational/financial tradeoff as compared to non-tech. You can keep your previous product running for next to nothing. Plus you can instantly push new changes to existing users.

Software can instantly scale to fill any niche or unlocked growth opportunity. The marginal cost of servicing a new user is basically zero, and it’s instant. Larger companies are held back by old processes, bureaucracy, hyper-focus on a larger revenue stream, etc.

Network effects (which many competing tech companies rely on) can be unraveled. I’ve always had the hardest time arguing this point. Friends have pushed back against me, saying that the whole point of a network effect is that it can’t come undone easily. I wasn’t quite articulate enough to counter that argument, but Marc Andreessen says it well in Elad Gil’s book:

Marc: I think network effects are great, but in a sense they’re a little overrated. The problem with network effects is they unwind just as fast. And so they’re great while they last, but when they reverse, they reverse viciously. Go ask the MySpace guys how their network effect is going. Network effects can create a very strong position, for obvious reasons. But in another sense, it’s a very weak position to be in. Because if it cracks, you just unravel. I always worry when a company thinks the answer is just network effects. How durable are they? To your point on data network effects, I would just say that we don’t see it very often. We see a lot of claims, and very little evidence. The reality is, there’s a lot of data in the world, and a lot of ways to get data. We have not seen very many data moats that actually make sense, even in science.

You can G2M through many channels. Technology companies often go to market through several different hyper-focused channels. All you need to do is find one LTV/CAC positive channel and pour money into it. Non-software companies are often stuck having to do generic lifestyle marketing and branding which doesn’t easily attribute or confirm ROI.

Software is less tied to external forces like regulation, macroeconomic trends, and cultural norms. I’d argue that the whole point of some software companies is to get around or hedge against the effects of these things. Software firms compete more so on product and distribution.

Software is feature-driven. Coke and Pepsi taste the same in blind trials — their slow-changing brands and distribution drive sales. If a software company builds a better feature set, they can win customers on the basis of their product or service’s functionality .

Data is fungible. Users can switch between WordPress and Medium, for example, at will and with no cost. This is a part of what drives software companies to be platforms: if your value is the aggregation and network effect rather than the info itself, users have a cost to switching (fewer eyeballs)

Great software companies tend to be category creators. I disagree with the Thiel view that startups must go zero-to-one or have a grand secret — most amazing businesses can be reframed in incremental terms, like Netflix is just TV but online, or Facebook is MySpace but for your college friends. Most software companies simply don’t compete much on price or quality. To be fair, this point is a bit of a truism: categories tend to be defined by the most prolific companies, not vice-versa.

All I’ve done here is enumerate the some factors in play. The challenging part is figuring out how these different forces work together, in which situations they appear, and what properties of the market emerge from them.

Of course these points are simplified. Many enterprise software companies, for example, have strong lock-in effects on data. It’s very difficult to export your task management or CRM tooling into another system. Perhaps digging more into these exceptions would make for a good followup post.

As I’m sadly nearing the end of my time at Umbrella and looking forward to what’s next, I wanted to write down some of the interesting things I’ve learned.

I originally joined with one personal mission: add as much info as possible to my understanding of what I want to do with my life. The role was for a wear-every-hat gap-filler. And I really liked the team — I could see myself fitting in well with the culture. To be honest, I wasn’t in love with the problem space at the time, but it was new and mildly intriguing.

As the days went on, however, I ended up becoming rather interested in senior living. Growing old creates problems in basically every part of your life, and most modern tech-enabled companies aren’t designed to work with seniors. I’m increasingly convinced that there are a truly gigantic number of opportunities and interesting problems out there.

Starting with some things about seniors in general:

The financial calculus is harder. Many seniors live on fixed income and have a lot of uncertainty surrounding how long they’ll live and whether or not they’ll have unexpected medical costs. This makes seniors reluctant to spend on anything that’s not necessary.

Word of mouth carries lots of weight. Seniors are very deliberate in evaluating their experience as customers. It can take weeks for seniors to get comfortable enough with a product to recommend it to friends — and they often have a smaller set of people they talk to. These dampen viral coefficients for a nascent single-player company like ours. It’s unknown how strong word of mouth will be for a more advanced senior-focused tech product. On the flip side, careful recommendations often have a higher impact and make friends likely to convert.

Trust and security are huge concerns. A disturbing number of Umbrella’s members have been previously scammed by unethical contractors or even had their bank accounts emptied by fraudsters. Some sales leads aren’t comfortable giving us a credit card number or basic personal info, and for good reason given the level of abuse targeting seniors. Figuring this out across marketing, sales, and product has been an interesting challenge.

Serious attention to detail. There were multiple occasions where I’d talk to a potential customer and say something like “we charge $29/month,” then after talking for another few minutes, I’d reiterate that “Umbrella only costs $30 bucks a month.” Then I’d inevitably be interrupted: “wait, I thought you said $29, not $30?” Inconsistent information over the phone and sloppy landing pages are a big turn-off.

It’s harder to consistently communicate. While most of us are available through text, email, or phone calls, seniors often only use one or two of the three. It can be tough to design a system that works when any given user might require that you use a specific channel.

You can’t take all tech product concepts for granted. Ideas like recurring subscriptions (what if I don’t use it this month?) and on-demand work (so who are these people just showing up?) are alien. Seniors usually don’t keep up with tech news that might familiarize them with companies like Uber or Netflix, so you can’t even rely on common context to build a brand image and value proposition.

Bad design is really really bad. You can’t hide much behavior (linking an address to a map application, for example.) Hamburger menus, swiping, and press-and-hold are out of the question. Animations are probably too confusing as well. Other complex products may be able to get away with those design elements, but not Umbrella. Building a crystal-clear interface that is acceptable to everyone on the seniority-spectrum is hard.

You earn a lot of goodwill, at least as far as tech companies go. Serving seniors is a strong mission that unifies potential partners, customers, and teammates. While still one hell of a challenge, BD/sales/recruiting gets 3% easier. And that can make a huge difference.

Your customers love you more than you thought was possible. One of the best parts of working at Umbrella has been the immediate and significant impact we’ve had on our members. Every week we’d get a call or note saying “I don’t know what I’d do without you,” or “Umbrella is a godsend.” It’s a lot easier to push through an intractable bug in your code when the users are so genuinely grateful for what you’re doing.

You work with lot of fascinating characters. Seniors don’t seem to care what anyone thinks of them. Between the hilarious age-gap moments, borderline-crazy 500+ word emails, and heartbreaking reminders of our own mortality, seniors are by far the most human group of customers I’ve ever served.

Thoughts on the space overall

There are a couple strong tailwinds that make me very bullish on senior-focused startups.

First, seniors are becoming more and more digital. While plenty of our members don’t have email addresses or cell phones, that’s changing — fast. You can easily Google around for trends and stats here. By the time a startup begins to scale up several years from its starting point, this will be much less of an issue. So now is a probably a great time to get rolling.

Second, unique senior-specific UX and sales challenges create a surprisingly solid moat. Think about everything I listed above — I suspect that X but for seniors will be a viable startup idea generator. We already see early hints of this from GoGoGrandparent, the Uber for seniors that went through YC. (They’re double dipping on the idea generators: Uber for X and X for seniors!)

I’m convinced there will be several $1bn+ companies out there focused on senior verticals. One of the running jokes at the office was that an online megachurch would be an insanely profitable business. While that’s not a business we would ever want to build, just remember how successful televangelists are. Add on the scale of the internet, modern phone/tablet distribution, and social-network-like community features and you’ve got something huge.

AARP is investing $40mm in senior-focused companies. I’ve also heard more non-specific but senior-related chatter recently from mainstream VCs. I’m looking forward to seeing how the space develops! Let me know if anything catches your eye.

Shoutout to Sam, Lindsay, Emma, Erin, Samra, Caroline, Megan, Manuela, and David for time that went by way too fast! I’ll miss you all ❤