The Science of Re-Upping

baseball, follow on

Soooooooo… (I know, what a great word to start a blogpost) I started this essay, with some familiarity on one subject. Little did I know I was going to learn about an entirely different industry, and be endlessly fascinated about that.

The analogy that kicked off this essay is that re-upping on a portfolio company is very much like re-signing a current player on a sports team. That was it. Simple as it was supposed to sound. The goal of any analogy was to frame a new or nuanced concept, in this case, the science of re-upping, under an umbrella of knowledge we were already familiar with.

But, I soon learned of the complexity behind re-upping players’ contracts, as one might assume. And while I will claim no authority over the knowledge and calculations that go into contracts in the sports arena, I want to thank Brian Anderson and everyone else who’s got more miles on their odometer in the world of professional sports for lending me their brains. Thank you!

As well as Arkady Kulik, Dave McClure, and all the LPs and GPs for their patience and willingness to go through all the revisions of this blogpost!

While this was a team effort here, many of this blogpost’s contributors chose to stay off the record.


The year was 1997.

Nomar Garciaparra was an instantaneous star, after batting an amazing .306/.342/.534. For the uninitiated, those are phenomenal stats. On top of batting 30 home runs and 11 triples – the latter of which was a cut above the rest of the league, it won him Rookie of the Year. And those numbers only trended upwards in the years after, especially in 1999 and 2000. Garciaparra became the hope for so many fans to end the curse of the Bambino – a curse that started when the Red Sox traded the legendary Babe Ruth to the Yankees in 1918.

Then 2001 hit. A wrist injury. An injured Achilles tendon. And the fact he needed to miss “significant time” earned him a prime spot to be traded. Garciaparra was still a phenomenal hitter when he was on, but there was one other variable that led to the Garciaparra trade. To Theo Epstein, above all else, that was his “fatal flaw.”

Someone that endlessly draws my fascination is Theo Epstein. Someone that comes from the world of baseball. A sport that venture draws a lot of inspiration, at least in analogy, like one of my fav sayings, Venture is one of the only types of investments where it’s not about the batting average but about the magnitude of the home runs you hit.

If you don’t follow baseball, Theo Epstein is the youngest general manager in the history of major league baseball at 26. But better known for ending the Curse of the Bambino, an 86-year curse that led the Red Sox down a championship drought that started when the Red Sox traded Babe Ruth to the Yankees. Theo as soon as he became general manager traded Nomar Garciaparra, a 5-time All-star shortstop, to the Cubs, and won key contracts with both third baseman Bill Mueller and pitcher Curt Schilling. All key decisions that led the Red Sox to eventually win the World Series 3 years later.

And when Theo left the Red Sox to join the Chicago Cubs, he also ended another curse – The Curse of the Billy Goat, ending with Theo leading them to a win in the 2016 World Series. You see, in baseball, they measure everything. From fly ball rates to hits per nine innings to pitches per plate appearance. Literally everything on the field.

But what made Theo different was that he looked at things off the field. It’s why he chose to bet on younger players than rely on the current all-stars. It’s why he measures how a teammate can help a team win in the dugout. And, it’s why he traded Nomar, a 5-time All Star, as soon as he joined, because Nomar’s “fatal flaw” was despite his prowess, held deep resentment to his own team, the Sox, when they tried to trade him just the year prior for Alex Rodriguez but failed to.

So, when Danny Meyer, best known for his success with Shake Shack, asked Theo what Danny called a “stupid question”, after the Cubs lost to the Dodgers in the playoffs, and right after Houston was hit by a massive hurricane, “Theo, who are you rooting for? The Dodgers so you can say you lost to the winning team, or Houston (Astros), because you want something good to happen to a city that was recently ravaged by a hurricane.”

Theo said, “Neither. But I’m rooting for the Dodgers because if they win, they’ll do whatever every championship team does and not work on the things they need to work on during the off season. And the good news is that we have to play them 8 times in the next season.”

You see, everyone in VC largely has access to the same data. The same Pitchbook and Crunchbase stat sheet. The same cap table. And the same financials. But as Howard Marks once said in response how you gain a knowledge advantage:

“You have to either:

  1. Somehow do a better job of massaging the current data, which is challenging; or you have to
  2. Be better at making qualitative judgments; or you have to
  3. Be better at figuring out what the future holds.”

For the purpose of this blogpost, we’re going to focus on the first one of the three.

To begin, we have to first define a term that’ll be booking its frequent flier miles for the rest of this piece – expected value.

Some defined it as the expectation of future worth. Others, a prediction of future utility. Investopedia defines it as the long-term average value of a variable. Merriam-Webster has the most rudimentary definition:

The sum of the values of a random variable with each value multiplied by its probability of occurrence

On the other hand, venture is an industry where the beta is arguably one of the highest. The risk associated with outperformance is massive as well. And the greatest returns, in following the power law, are unpredictable.

We’re often blessed with hindsight bias, but every early-stage investor in foresight struggles with predicting outlier performance. Any investor that says otherwise is either deluding you or themselves or both. At the same time, that’s what makes modeling exercises so difficult in venture, unlike our friends in hedge funds and private equity. Even the best severely underestimate the outcomes of their best performers. For instance, Bessemer thought the best possible outcome for Shopify was $400M with only a 3% chance of occurring.

Similarly, who would have thought that jumping in a stranger’s car or home, or live streaming gameplay would become as big as they are today. As Strauss Zelnick recently said, “The biggest hits are by their nature, unexpected, which means you can’t organize around them with AI.” Take the word AI out, and the sentence is equally as profound replaced with the word “model.” And it is equally echoed by others. Chris Paik at Pace has made it his mission to “invest in companies that can’t be described in a single sentence.”

But I digress.

Value itself is a huge topic – a juggernaut of a topic – and I, in no illusion, find myself explaining it in a short blogpost, but that of which I plan to spend the next couple of months, if not years, digging deeper into, including a couple more blogposts that are in the blast furnace right now. But for the purpose of this one, I’ll triangulate on one subset of it – future value as a function of probability and market benchmarks.

In other words, doubling down. Or re-upping.

For the world of startups, the best way to explain that is through a formula:

E(v) = (probability of outcome) X (outcome)

E(v) = (graduation rate) X (valuation step up from last round) X (dilution)

For the sake of this blogpost and model, let’s call E(v), appreciation value. So, let’s break down each of the variables.

What percent of your companies graduate to the next round? I shared general benchmarks in this blogpost, but the truth is it’s a bit more nuanced. Each vertical, each sub-vertical, each vintage – they all look different. Additionally, Sapphire’s Beezer recently said that it’s normal to expect a 20-30% loss ratio in the first five years of your fund. Not all your companies will make it, but that’s the game we play.

On a similar note, institutional LPs often plan to build a multi-fund, multi-decade relationship with their GPs. If they invest in a Fund I, they also expect to be there by Fund III.

How much greater is the next round’s valuation in comparison to the one in which you invested? Twice as high? Thrice? By definition, if you double down on the same company, rather than allocate to a net new company, you’re decreasing your TVPI. And as valuations grow, the cost of doubling down may be too much for your portfolio construction model to handle, especially if you’re a smaller sub-$100M fund.

It’s for the same reason that in the world of professional sports, there are salary caps. In fact, most leagues have them. And only the teams who:

  • Have a real chance at the championship title.
  • Have a lot in their coffers. This comes down to the composition of the ownership group, and their willingness to pay that tax.
  • And/or have a city who’s willing to pay the premium.

… can pay the luxury tax. Not to be too much of a homer, but the Golden State Warriors have a phenomenal team and are well-positioned to win again (at least at the time of this blogpost going out). So the Warriors can afford to pay the luxury tax, but smaller teams or teams focused on rebuilding can’t.

The Bulls didn’t re-sign the legendary Michael Jordan because they needed to rebuild. Indianapolis didn’t extend Peyton Manning’s contract ‘cause they didn’t have the team that would support Peyton’s talents. So, they needed to rebuild with a new cast of players.

Similarly, Sequoia and a16z might be able to afford to pay the “luxury tax” when betting on the world’s greatest AI talent and for them to acquire the best generative AI talent. Those who have a real chance to grow to $100M ARR, given adoption rates, retention rates, and customer demand. But as a smaller fund or a fund that has a new cast of GPs (where the old guard retired)… can you?

If a star player is prone to injury or can only play 60 minutes of a game (rather than 90 minutes), a team needs to re-evaluate the value of said player, no matter how talented they are. How much of a player’s health, motivation, and/or collaborativeness – harkening back to the anecdote of Nomar Garciaparra at the beginning – will affect their ability to perform in the coming season?

Take, for instance, the durability of a player. If there ‘s a 60% chance of a player getting injured if he/she plays longer than 60 minutes in a game and a 50% of tearing their ACL, while they may your highest scorer this season, they’re not very durable. If that player missed 25% of practices and 30% of games, they just don’t have it in them to see the season through. And you can also benchmark that player against the rest of the team. How’s that compared with the team’s average?

Of course, there’s a parallel here to also say, every decision you make should be relative to industry and portfolio benchmarks.

How great of a percentage are you getting diluted with the next round if you don’t maintain your ownership? This is the true value of your stake in the company as the company grows.

E(v) = (graduation rate) X (valuation step up from last round) X (dilution)

If the expected value is greater than one, the company is probably not worth re-upping. And that probably means the company is overhyped, or that that market is seeing extremely deflated loss ratios. In other words, more companies than should be, are graduating to the next stage; when in reality, the market is either a winner-take-all or a few-take-all market. If it is less than or equal to one, then it’s ripe to double down on. In other words, the company may be undervalued.

And to understand the above equation or for it to be actually useful (outside of an abstract concept), you need market data. Specifically, around valuation step ups as a function of industry and vertical.

If you happen to have internal data across decades and hundreds of companies, then it’s worth plugging in your own dataset as well. It’s the closest you can get to the efficient market frontier.

But if you lack a large enough sample size, I’d recommend the below model constructed from data pulled from Carta, Pitchbook, and Preqin and came from the minds of Arkady Kulik and Dave McClure.

The purpose of this model is to help your team filter what portfolio companies are worth diving deeper into and which ones you may not have to (because they didn’t pass the litmus test) BEFORE you evaluate additional growth metrics.

It is also important to note that the data we’ve used is bucketed by industry. And in doing so, assumptions were made in broad strokes. For example, deep tech is broad by design but includes niche-er markets that have their own fair share of pricing nuances in battery or longevity biotech or energy or AI/ML. Or B2B which include subsectors in cybersecurity or infrastructure or PLG growth.

Take for instance…

Energy sector appreciation values and follow-on recommendations

The energy sector sees a large drop in appreciation value at the seed stage, where all three factors contribute to such an output. Valuation step-up is just 1.71X, graduation rates are less than 50% and dilution is 38% on average.  

Second phase where re-upping might be a good idea is Series B. Main drivers as to such a decision are that dilution hovers around 35% and about 50% of companies graduate from Series A to Series B. Mark ups are less significant where we generally see only an increase in valuation at about 2.5X, which sits around the middle of the pack.

Biotech sector appreciation values and follow-on recommendations

The biotech sector sees a large drop in appreciation value at the Seed stage. This time, whereas dilution seems to match the pace of the rest of the pack (at an average of 25%), the two other factors shine greater in making a follow-on decision. Valuation step up are rather low, sitting at 1.5X. And less than 50% graduate to the next stage.

In the late 2023 market, one might also consider re-upping at the Series C round. Main driver is the unexpectedly low step-up function of 1.5X, which matches the slow pace of deployment for growth and late stage VCs. On the flip side, a dilution of 17% and graduation rate of 60% are quite the norm at this stage.

All in all, the same exercise is useful in evaluating two scenarios – either as an LP or as a GP:

  1. Is your entry point a good entry point?
  2. Between two stages, where should you deploy more capital?

For the former, too often, emerging GPs take the stance of the earlier, the better. Almost as if it’s a biblical line. It’s not. Or at least not always, as a blanket statement. The point of the above exercise is also to evaluate, what is the average value of a company if you were to jump in at the pre-seed? Do enough graduate and at a high enough price for it to make sense? While earlier may be true for many industries, it isn’t true for all, and the model above can serve as your litmus test for it. You may be better off entering at a stage with a higher scoring entry point.

For the latter, this is where the discussion of follow on strategies and if you should have reserves come into play. If you’re a seed stage firm, say for biotech, using the above example, by the A, your asset might have appreciated too much for you to double down. In that case, as a fund manager, you may not need to deploy reserves into the current market. Or you may not need as large of a reserve pool as you might suspect. It’s for this reason that many fund managers often underallocate because they overestimate how much in reserves they need.

If you’re curious to play around with the model yourself, ping Arkady at ak@rpv.global, and you can mention you found out about it through here. 😉

Photo by Gene Gallin on Unsplash


Stay up to date with the weekly cup of cognitive adventures inside venture capital and startups, as well as cataloging the history of tomorrow through the bookmarks of yesterday!


The views expressed on this blogpost are for informational purposes only. None of the views expressed herein constitute legal, investment, business, or tax advice. Any allusions or references to funds or companies are for illustrative purposes only, and should not be relied upon as investment recommendations. Consult a professional investment advisor prior to making any investment decisions.

#unfiltered #75 Why I Write Long Form Blogposts

typewriter, blog, write

This past Wednesday, I was having lunch with an artist-turned-VC. And as you might imagine, we had to cover every topic at the intersection of art and startup investing. But of all the ground we covered, one stood out — content creation.

She’s on the Gram, LinkedIn, and everything in between. (Although surprisingly not on Twitter.) But to help her focus, she uninstalls those apps on her phone. Otherwise, she says she’ll end up “doomscrolling.” I get it. In fact, many of my friends and colleagues have shared similar things as well. But…

I’m weird. At least among my friend group, I’m really weird. I’m terrible at social media. I’m an 80-year old stuck in a 27-year old body. At least on the social media front. I find it so hard to keep my attention on social. In fact, I schedule ten minutes three times a week to hold myself accountable to be on LinkedIn and Twitter.

So, when it came to sharing my thoughts and learnings publicly, it was a pretty easy decision. Of course, I eventually came to self-rationalize it as the ability to own my own piece of virtual real estate, but there are three more reasons I chose blogging rather than tweeting or social posting.

1. I write to think

I’ve written about this before so I won’t elaborate in length on my own rationale here, but share a few examples of others also holding it in high regard.

There’s only so much you can flush out in just 280 characters, or over any short post. And while some of my thoughts fully flushed out may only be that long or less, not having that restriction gives me peace of mind to not hold back.

One of my favorite George Orwell lines happens to be: “If people cannot write well, they cannot think well. And if they cannot think well, others will do their thinking for them.”

On that same wavelength on writing, Jeff Bezos makes Amazon execs write six-page memos. In most companies, team members often resort to PowerPoint presentations. Take anywhere between five and ten slides. Maybe less, maybe more. It’s much less thought out than a six-page dissertation. As Bezos says, “The reason writing a ‘good’ four page memo is harder than ‘writing’ a 20-page PowerPoint is because the narrative structure of a good memo forces better thought and better understanding of what’s more important than what.”

Equally so, it’s the same reason the best investors write memos for their investment decisions. My favorite public ones are Bessemer’s, which encapsulates much of their thinking at the time in amber. Turner Novak also turned his ability to write great memos to eventually raising his fund, Banana Capital. And the great Brian Rumao writes memos not just pre-investment, but also in his post-mortems where he gathers his learnings.

While I won’t go as far as to comparing myself to the afore-mentioned, I do find great pleasure and great learning from putting words on paper.

2. Longer feedback loop

My writing is more often a form of self-expression, self-curiosity, and self-discovery. So, unlike a product manager or founder who’s relentlessly testing and iterating on feedback, I enjoy longer feedback loops. I may start another content engine at some point that is for a particular audience, focused on feedback and iteration. But this humble piece of virtual estate will stay me. With no algorithm conditioning my attention span and yearn for external validation. That’s not to say I won’t ever (or have not ever) written things that you my awesome readers want, but it is only at the intersection of what you want and the what I enjoy writing about and asking others about that mint content here.

I also spend a lot of time thinking about audience capture, a term Gurwinder brought to my attention in an essay he wrote about Nikocado Avocado, which I also touched on in an essay I wrote near the end of last year.

I’m reminded by something Gurwinder wrote a few months ago about the perils of audience capture. In it, he shares the story of Nikocado Avocado, who lost himself to his audience, in a section of that essay he calls: The Man Who Ate Himself. He also shares one line that I find quite profound:

“We often talk of ‘captive audiences,’ regarding the performer as hypnotizing their viewers. But just as often, it’s the viewers hypnotizing the performer. This disease, of which Perry is but one victim of many, is known as audience capture, and it’s essential to understanding influencers in particular and the online ecosystem in general.”

3. The impermanence of social media

Most things on social media are ephemeral in nature. It’s designed to capture the moment, but not chronicle the moments. On Twitter, you can only pin one tweet. On Instagram, you can pin three. And on LinkedIn, only three are visible on the featured carousel, and include, five max before it takes you down another layer of friction to discover more.

There’s a level of impermanence which makes thoughts feel whimsical rather than evergreen. To use a phrase I recently heard Tim Ferriss use, the “durability of the signal seems to wane so quickly.” And that made my thoughts feel cheap.

That’s not to say every post I write has their weight in gold, but the searchability and the evergreen nature of my favorite blogposts (saved in my “About” tab) are the reasons I keep most of my thoughts here.

Photo by Fiona Murray on Unsplash


#unfiltered is a series where I share my raw thoughts and unfiltered commentary about anything and everything. It’s not designed to go down smoothly like the best cup of cappuccino you’ve ever had (although here‘s where I found mine), more like the lonely coffee bean still struggling to find its identity (which also may one day find its way into a more thesis-driven blogpost). Who knows? The possibilities are endless.


Stay up to date with the weekly cup of cognitive adventures inside venture capital and startups, as well as cataloging the history of tomorrow through the bookmarks of yesterday!


The views expressed on this blogpost are for informational purposes only. None of the views expressed herein constitute legal, investment, business, or tax advice. Any allusions or references to funds or companies are for illustrative purposes only, and should not be relied upon as investment recommendations. Consult a professional investment advisor prior to making any investment decisions.

The Superpower of Being Underestimated

underestimated, rejection, star

The Warriors went through one hell of a season. Even as someone who doesn’t live and breathe basketball, watching Stephen Curry this past season, especially during the finals with the Celtics was a thrill out of this world. He is undeniably one of the greats! Yet it’s fascinating to think that the world didn’t always see him as such. From being a 3-star recruit to the 256th-ranked player in 2006 to 7th pick in 2009, Curry’s gone a long way.

Though he recently won an Academy Award for Best Original Score for his music on Dune, Hans Zimmer‘s early music career was not easy. He had been thrown out of eight schools and only had two weeks of piano lessons. Yet today he is undeniably one of the greatest composers of our time.

Comment
byu/realhanszimmer from discussion
inIAmA
Source: Hans Zimmer’s Reddit AMA

When Stan Lee first pitched Spider-Man, his publisher thought it was “the worst idea I have ever heard.” The publisher himself told one of the greatest storytellers: “First of all, people hate spiders, so you can’t call a book Spider-Man. Secondly he can’t be a teenager—teenagers can only be sidekicks. And third, he can’t have personal problems if he’s supposed to be a superhero—don’t you know who a superhero is?'” The rest… is history.

In the making of Star Wars, George Lucas was rejected time and time again – from Disney to United Artists to Universal. And the one bet that 20th Century Fox took on him was for only a budget of $8M, that eventually became a $10M budget, when at the time, the best blockbuster films all had budgets of $20-30M. Yet, today Star Wars stands as one of the greatest cultural assets of the 20th and 21st century.

In the world of startups, the world’s most valuable companies are worth more than four times and raised half as much as the world’s most funded companies. Funding, in many ways, is a proxy for investor optimism in the early days that this company will be the next big thing. But investors, like any other person, can be wrong. In fact, startup investors are often wrong more often than they’re right. But it also goes to say the world’s best companies are non-obvious, in the non-consensus. In other words, underestimated.

Source: Founder Collective

As the above graphic shows, even if one picks right, we still grossly underestimate the potential of outliers. After all, humans are terrible at tracking nonlinearities:

  • In 2012, Canva was rejected by over 100 Silicon Valley investors. Now it is a growing $40 billion business of gargantuan proportions.
  • The Post-it note was an result of a failed experiment to create stronger adhesives. But Dr. Spencer Silver, its inventor, kept at it, which led to his nickname as “Mr. Persistent” because he wouldn’t give up. Today, Post-it notes are sold in more than 100 countries, and over 50 billion are produced every year.
  • Google, one of the most recognizable names today, struggled to raise capital and find customers in the early days. Who needed another search engine? For 1.5 years, every search company approached by Larry and Sergey to consider Google’s tech turned them down. The pair funded Google on their credit cards and couldn’t even afford to hire a designer so regressed to minimalism.
  • Tope Awotona, founder of Calendly, started three failed businesses and emptied his 401k to fund Calendly. Yet despite his hustle and persistence, most VCs he talked to turned him down. Despite starting in 2013, it wasn’t till 2021 that Calendly had their A-round. Calendly took much longer to get the attention of external funding than many of its counterparts. The company is now one of the most popular scheduling tools and worth $3B.

But even when people got it right, they still underestimated the upside.

  • Even when Kleiner eventually backed Google, legendary investor John Doerr couldn’t believe it when Larry Page believed that Google could get revenues of $10B.
  • When Bessemer invested in Shopify, Bessemer thought that the best possible outcome for Shopify was a 3% chance of the company exiting at $400M. As of the time of this essay, it’s worth over 100 times more with a market cap of $43B.
  • If you invested in Amazon on the first day in 1998 at $5, most people would have sold at $85 in 1999 – a 17x in less than two years. But if they held to today, they would have made a multiple north of 600x. That said, selling itself is more of an art than a science.

… And the list goes on.

As Warren Buffett says, “the rearview mirror is always clearer than the windshield.” Our fallacy with estimation is painfully obvious in hindsight, but dubitably unclear in foresight.

Early on in my venture career, an investor once told me a profound statement. One that I still remember to this day. The best ideas – and often the leaders of tomorrowoften seem crazy at first. And because they’re crazy, they’re nonobvious. They’re in the non-consensus.

As Steve Jobs says, “the ones who are crazy enough to think they can change the world, are the ones who do.” The world’s most transformative individuals and businesses take on many more headwinds than those optimizing for local maxima. But history shows us that those that dream big consistently outperform those optimizing for marginal improvement. While there is nothing wrong with the latter, I hope the above anecdotes serve as a reminder rejection is not a sign of failure. Rather, it’s a sign that most people have yet to see what you see.

Your job is to teach them to see what you see. After all, the only difference between a hallucination and a vision is that other people can see a vision.

Photo by Aziz Acharki on Unsplash


Edit: Added in Stan Lee’s story.


Stay up to date with the weekly cup of cognitive adventures inside venture capital and startups, as well as cataloging the history of tomorrow through the bookmarks of yesterday!


Any views expressed on this blog are mine and mine alone. They are not a representation of values held by On Deck, DECODE, or any other entity I am or have been associated with. They are for informational and entertainment purposes only. None of this is legal, investment, business, or tax advice. Please do your own diligence before investing in startups and consult your own adviser before making any investments.

#unfiltered #66 Humans and Nonlinear Thinking

Humans are terrible at understanding percentages. I’m one of them. An investor I had the opportunity to work with on multiple occasions once told me. People can’t tell better; people can only tell different. It’s something I wrestle with all the time when I hear founder pitches. Everyone claims they’re better than the incumbent solution. Whatever is on the market now. Then founders tell me they improve team efficiency by 30% or that their platform helps you close 20% more leads per month. And I know, I know… that they have numbers to back it up. Or at least the better founders do. But most investors and customers can’t tell. Everything looks great on paper, but what do they mean?

When the world’s wrapped in percentages, and 73.6% of all statistics are made up, you have to be magnitudes better than the competition, not just 10%, 20%, 30% better. In fact, as Sarah Tavel puts it, you have to be 10x better (and cheaper). And to be that much better, you have to be different.

And keep it simple. As Steve Jobs famously said that if the Mac needed an instruction manual, they would have failed in design. Your value-add should be simple. Concise. “We all have busy lives, we have jobs, we have interests, and some of us have children. Everyone’s lives are just getting busier, not less busy, in this busy society. You just don’t have time to learn this stuff, and everything’s getting more complicated… We both don’t have a lot of time to learn how to use a washing machine or a phone.”

If you need someone to learn and sit down – listen, read, or watch you do something, you’ve lost yourself in complexity.

“Big-check” sales is a game of telephone. For enterprise sales or if you’re working with healthcare providers, the sales cycle is long. Six to nine months, maybe a year. The person you end up convincing has to shop the deal with the management team, the finance team, and other constituents.

For most VCs writing checks north of a million, they need to bring it to the partnership meeting. Persuade the other partners on the product and the vision you sold them.

And so if your product isn’t different and simple, it’ll get lost in translation. Think of it this way. Every new person in the food chain who needs to be convinced will retain 90% of what the person before them told them. A 10% packet loss. The tighter you keep your value prop, the more effective it’ll be. The longer you need to spend explaining it with buzzwords and percentages, the more likely the final decision maker will have no idea why you’re better.


Humans are terrible at tracking nonlinearities. While we think we can, we never fully comprehend the power law. Equally so, sometimes I find it hard to wrap my hear around the fact that 20% of my work lead to 80% of the results. While oddly enough, 80% of my inputs will only account for 20% of my results. The latter often feels inefficient. Like wasted energy. Why bother with most work if it isn’t going to lead to a high return on investment.

Yet at the same time, it’s so far to tell what will go viral and what won’t. Time, energy, capital investments that we expect to perform end up not. While every once in a while, a small project will come out of left field and make all the work leading up to it worth it.

When I came out with my blogpost on the 99 pieces of unsolicited advice for founders last month, I had an assumption this would be a topic that my readers and the wider world would be interested in. At best, performing twice as well than my last “viral” blogpost.

Cup of Zhou readership as of April 2022

Needless to say, it blew my socks off and then some. My initial 99 “secrets”, as my friends would call it, accounted for 90% of the rightmost bar in the above graph. And the week after, I published my 99 “secrets” for investors. While it achieved some modest readership in the venture community and heartwarmingly enough was well-received by investors I respected, readership was within expectations of my previous blogposts.

My second piece wasn’t necessarily better or worse in the quality of its content, but it wasn’t different. While I wanted to leverage the momentum of the first, it just didn’t catch the wave like I expected it to.

Of course, as you might imagine, I’m not alone. Nikita Bier‘s tbh grew from zero to five million downloads in nine weeks. And sold to Facebook for $100 million. tbh literally seemed like an overnight success. Little do most of the public know that, Nikita and his team at Midnight Labs failed 14 times to create apps people wanted over seven years.

When Bessemer first invested in Shopify, they thought the best possible outcome for the company would be an exit value of $400 million. While not necessarily the best performing public stock, its market cap, as of the time I’m writing this blogpost, is still $42 billion. A 100 times bigger than the biggest possible outcome Bessemer could imagine.


Humans are terrible at committing to progress. The average person today is more likely to take one marshmallow now than two marshmallows later.

Between TikTok and a book, many will choose the former. Between a donut and a 30-minute HIT workout, the former is more likely to win again. Repeated offences of immediate gratification lead you down a path of short-term utility optimization. Simply put, between the option of improving 1% a day and regressing 1% a day, while not explicit, most will find more comfort in the latter alternative.

James Clear has this beautiful visualization of what it means to improve 1% every day for a year. If you focus on small improvements every day for a year, you’re going to be 37 times better than you were the day you started.

While the results of improving 1% aren’t apparent in close-up, they’re superhuman in long-shot.

Source: James Clear

Photo by Thomas Park on Unsplash


Stay up to date with the weekly cup of cognitive adventures inside venture capital and startups, as well as cataloging the history of tomorrow through the bookmarks of yesterday!


Any views expressed on this blog are mine and mine alone. They are not a representation of values held by On Deck, DECODE, or any other entity I am or have been associated with. They are for informational and entertainment purposes only. None of this is legal, investment, business, or tax advice. Please do your own diligence before investing in startups and consult your own adviser before making any investments.

DGQ 5: What startups would I love to have in my anti-portfolio?

ice cream, mistake, anti portfolio

In the venture world, there’s this concept of the anti-portfolio. A portfolio for incredible startups you had the chance to invest in, but chose to pass on. Usually the startups that qualify to be in this anti-portfolio have already reached mainstream – either having gone public and/or have reached unicorn status. For anti-portfolio references, I highly recommend checking out Bessemer‘s or tuning into Samir Kaji’s Venture Unlocked podcast, where he asks each guest about their anti-portfolio.

But having chatted with a number of incredible investors, what’s more important than names on an excel sheet is the lesson or lessons we take away from passing on the greats. Those lessons are the very answer to one of the most insightful questions an LP (limited partner) can ask. “How does your anti-portfolio advise your current investment thesis?”

In a similar way, life is a mixed bag of engineered serendipity and endured scar tissue. Our past mistakes inform our future decisions. You learn how to handle kitchen cutlery after cutting yourself a few times. You learn to walk after stumbling. And you learn to communicate after making a fool of yourself. We are a product of the scar tissue we’ve accumulated.

I’m in my first inning in the venture world, and admittedly, way too early to have any true hall-of-famers in my anti-portfolio. So rather than looking into the past from the present, I thought I’d look into the “past” from the future. A “past” that has yet to come, but will be defining of my future. Something Mike Maples Jr calls backcasting. Starting from the future and making my way back to today, along the way, figuring out what I need to do to get to that future. If you’ve been following this blog for a while, you know I’m a big fan of his mental model. “The future doesn’t happen to us; it happens because of us. […] Breakthrough builders are visitors from the future, telling us what’s coming.”

Rather than what startups are in my anti-portfolio, what startups would I love to have in my anti-portfolio?

On a similar note, for non-investors: Ten years from now, what are mistakes you’d want to have made that you tell yourself that it was a decade well-spent?

Photo by Sarah Kilian on Unsplash


The DGQ series is a series dedicated to my process of question discovery and execution. When curiosity is the why, DGQ is the how. It’s an inside scoop of what goes on in my noggin’. My hope is that it offers some illumination to you, my readers, so you can tackle the world and build relationships with my best tools at your disposal. It also happens to stand for damn good questions, or dumb and garbled questions. I’ll let you decide which it falls under.


Subscribe to more of my shenaniganery. Warning: Not all of it will be worth the subscription. But hey, it’s free. But even if you don’t, you can always come back at your own pace.

Expert + Reasonable + Crazy Idea = Crazy Good

The amazing Paul “PG” Graham came out with an essay this month on crazy new ideas. And the thing I’ve learned over the years, being in Silicon Valley, is if PG writes, you read. In it, one section in particular stood out:

“Most implausible-sounding ideas are in fact bad and could be safely dismissed. But not when they’re proposed by reasonable domain experts. If the person proposing the idea is reasonable, then they know how implausible it sounds. And yet they’re proposing it anyway. That suggests they know something you don’t. And if they have deep domain expertise, that’s probably the source of it.

“Such ideas are not merely unsafe to dismiss, but disproportionately likely to be interesting.”

I’ve written a number of essays about crazy ideas. Here. Also here. The last of which you’ll need to Ctrl F “crazy”, if you don’t want to read through all of it. And also, most recently, here. But that’s besides the point. The common theme between all of these is that crazy ideas are not hard to come by. Crazy good ideas are. Good implies that you’re right when everyone else thinks you’re crazy. When you’re in the minority. And the smaller of the minority you are in, the greater the margin on the upside. Potential upside, to be fair.

As investors, we hear crazy pitches every so often. David Cowan at Bessemer even wrote a satire on it all. For the crazy pitches, go to episode five. The question is: How do we differentiate the crazy ideas from the crazy good ideas? But as PG says, if it’s coming from someone we know is a subject-matter expert (SME) and they’re usually grounded on logic and reasoning, then we spend time listening. Asking questions. And listening. ‘Cause they most likely know something we don’t.

That was true for Brian Armstrong, who recently brought his company, Coinbase, public. He worked on fraud detection for Airbnb in its early days prior. And he knew he was getting into the deep end with crypto back in 2012. But he realized how unscalable crypto transactions were and how frustrated he was. Garry Tan, then at YC and part-time at Initialized, saw exactly that in him. A reasonable SME with a crazy idea. Garry just released an amazing interview between him and Brian too, if you want to tune into the full story.

What if some of the variables in the equation are missing?

But most of the time the founders you’re talking to aren’t subject-matter experts with deep domain expertise. Or at least, they haven’t left an online breadcrumb trail of whether they’re a thought leader or if they’re reasonable human beings. So subsequently, in the little time I have with founders in a first or second meeting, I look for proxies.

For proxies on domain expertise, I go back to first principles. What are the underlying assumptions you are making? Why are they true? How did you arrive at them? What are the growing trends (i.e. market, economic, social, tech, etc.) that have primed your startup to succeed in the market? Does timing work out?

To see if they’re “reasonable” under PG’s definition, I seek creative conflict. How do you disagree with people? If I brought in a contrarian opinion you don’t agree with, how do you enlighten me? How do you disagree with your co-founders?

In closing

To be fair, we’re not always right. In fact, we’re rarely right. On average, in a hypothetical portfolio of 10 startups, five to six go to zero. One to two break even. Another one to two make a 2-3x on investment. That is to say, they return to the investor $2-3 for every $1 invested. And hopefully, one, just one, kills it, and becomes that fund returner. Fund returner – what we call an investment that returns the whole fund and maybe more. Of course, every time a VC invests, they’re aiming for the fences every time. As a VC once told me, “it’s not about the batting average but the magnitude of the home runs you hit.” And even in those 10 investments, it’s a stretch to say that all of them are “crazy” ideas.

But the hope is that even if we’re wrong on the idea, we’re right on the people.

Photo by Àlex Rodriguez on Unsplash


Stay up to date with the weekly cup of cognitive adventures inside venture capital and startups, as well as cataloging the history of tomorrow through the bookmarks of yesterday!

Fantastic Unicorns and Where to Find Them

As a venture scout and as someone who loves helping pre-seed/seed startups before they get to the A, I get asked this one question more often than I expect. “David, do you think this is a good idea?” Most of the time, admittedly, I don’t know. Why? I’m not the core user. I wouldn’t count myself as an early adopter who could become a power user, outside of pure curiosity. I’m not their customer. To quote Michael Seibel of Y Combinator,

… “customers are the gatekeepers of the startups world.” Then comes the question, if customers are the gatekeepers to the venture world, how do you know if you’re on to something if you’re any one of the below:

  • Pre-product,
  • Pre-traction,
  • And/or pre-revenue?

This blog post isn’t designed to be the crystal ball to all your problems. I have to disappoint. I’m a Muggle without the power of Divination. But instead, let me share 3 mental models that might help a budding founder find idea-market fit. Let’s call it a tracker’s kit that may increase your chances at finding a unicorn.

  1. Frustration
  2. The highly fragmented industry with low NPS
  3. Right on non-consensus
Continue reading “Fantastic Unicorns and Where to Find Them”

#unfiltered #5 The Insider “Silicon Valley” TV Show – The Show, plus Thoughts on Eccentric Cold Emails and Crazy Startup Pitches

Tech satire.

I gotta say I love it! Memes. GIFS. YouTube vids. TikTok clips. The whole nine yards.

As a testament to how much I love satirical memes and GIFs, six years ago, when I was testing out “best” cold email methods, as a semi-random A/B test, I emailed half of the folks I reached out to, leading or ending with either a meme or GIF. The list ranged from authors to musicians to researchers to Fortune 500 executives to founders to professional stone skippers. And the results weren’t half bad. Out of 150 odd emails, about a 70% response rate. Half of which resulted in a follow-up exchange by email, call, or in-person. The other half were gracious enough to say time was not on their side.

So when I learned, from the most recent episode of Angel podcast, about David Cowan’s version, I just had to check it out. And I wish I had only discovered it sooner. Made by Director Martin Sweeney, and co-visionaries, Michael Fertik of Reputation.com and David Cowan of Bessemer Venture Partners, bubbleproof is tech hilarity… made by the folks who have tech day jobs. Though I still haven’t watched the 6 seasons and 53 episodes of the Silicon Valley TV series yet. Sorry, friends who keep recommending it.

I just finished episode 5, where they share a snapshot of comedic ideas and pitches – from lipid fuel technology to an Airbnb marketplace for prisoners. And not gonna lie, I had a good chuckle. But when the episode wrapped up and I finally had a chance to think in retrospect, those ideas could have been real pitches in some world out there. When I first started in venture, I met with my share of cancer cures predicated off of a happiness matrix and feces fuel and African gold brokers. In case you’re wondering, yes, I did get pitched those. The last one admittedly should have come through my spam folder.

In these next few weeks, while you’re WFH (work from home), if you’re curious about tech from the ironic perspective of those who live and breathe it every day, check the series out. Only 10 episodes. 7-15 minutes per. (And while you do that, maybe I’ll finally get around to watching Silicon Valley. But no promises.)

As a footnote, Bessemer also has a track record for being forthcoming and intellectually honest. I would highly recommend checking out their anti portfolio, that lists and explains not their biggest wins or losses, but their biggest ‘shoulda-coulda-woulda’s’.


#unfiltered is a series where I share my raw thoughts and unfiltered commentary about anything and everything. It’s not designed to go down smoothly like the best cup of cappuccino you’ve ever had (although here‘s where I found mine), more like the lonely coffee bean still struggling to find its identity (which also may one day find its way into a more thesis-driven blogpost). Who knows? The possibilities are endless.


Stay up to date with the weekly cup of cognitive adventures inside venture capital and startups!