Making decisions is one of the core competencies of entrepreneurship. When you’re in the thick of things, you may be making multiple decisions per hour. And even when they don’t seem like big decisions, the impact of each decision – however small – compounds rapidly. Let’s say that improving your decision making allows you to increase the utility function of each decision by 0.5%. But after 1,000 such good decisions, that same function will have increased 147X.
Despite this large impact, the techniques of decision-making are often overlooked.
I sat down with my Blitzscaling co-author Chris Yeh to discuss this in depth on the most recent Greymatter podcast, which you can listen to here:
What is Good Decision-Making?
Entrepreneurs and executives will do all sorts of things to improve themselves, from taking a class on a particular technology, to reading a book about venture deals, to practicing intermittent fasting. Yet very few of them stop to ask the simple yet powerful question, “How can I make my decision-making better?”
And by the way, “better” doesn’t just mean “more accurate.” It also means “faster.” Many of the decisions facing entrepreneurs seem to have high stakes; they can literally mean the difference between the life or death of the company. When faced with such weighty decisions, the normal human instinct is to slow down, collect all the available information, and get emotionally comfortable with the possibilities before making a decision. That instinct is almost 100% wrong for entrepreneurial decision-making.
I describe entrepreneurial-decision making as confronting a rocky minefield (at night, inside a dense fog bank…) and deciding to sprint across. The central skill is learning how to be quick and decisive, and to adjust your course to avoid the worst of the mines.
When I evaluate entrepreneurs, one of the key factors I analyze is their decision-making calculus. This is especially true for repeat entrepreneurs. If they’re the kind of infinite learner who uses their past experiences to make their decision-making better, they have a huge advantage when it comes to everything from hiring to strategy to sales, making them an even better bet the second time around. But if they don’t have the drive and self-awareness to improve their decision-making, it’s probably not worth paying the experience premium to invest in them.
Making Decisions While Blitzscaling
If entrepreneurship requires a lot of decision-making, blitzscaling increases the pace and the stakes of that decision-making, which means an even greater premium on learning to make good decisions quickly under challenging circumstances.
I learned this from my own entrepreneurial experiences. Before becoming an entrepreneur, I thought I was really good at decision-making, in part because I considered myself intelligent, and in part because I thought about decision-making as a sport to master.
Then when I started my first company, SocialNet, I made bad decisions.
Sometimes I made decisions too slowly, sometimes I made decisions badly, and sometimes I did both. Those experiences led me to refine my decision-making framework.
It might have been tempting to blame those failures on others and say, “I trusted so-and-so, and he was wrong.” But I knew that I needed to take responsibility for every decision I made. When you find yourself in that situation, you need to ask yourself, “What should I do differently going forward to improve the way I play the game?” This self-reflection was a key reason that I was a much better decision maker during my time at PayPal.
Blitzscaling compresses this learning process, making it more challenging, but also more potentially productive. When you’re racing your competitors to achieve scale, you don’t have a lot of extra time to consult every stakeholder and reach a consensus. You have to make decisions, take responsibility for them, and learn from them.
This is one area where I recommend leveraging existing decision-making tools like the RACI (Responsible, Accountable, Consulted, Informed) and DACI responsibility assignment matrices. These proven tools can help structure and accelerate your decision-making.
“What should I do differently going forward to improve the way I play the game?”
Inspiration to Data
One of the key transitions of blitzscaling is the need to go from inspiration to data. This key transition is all about improving your decision-making. When scaling, you must be data-driven. There is no universe in which you don’t start harmonizing and operationalizing your entire company with data.
The drumbeat of a data-driven company is the company dashboard, and the meetings it holds around that dashboard. The metrics from the dashboard are the key inputs for collective decisions. They’re how you understand if you’re making progress or not. For every decision or action the company makes or takes, the dashboard helps answer the question, “How did that move affect our OKRs (Objectives and Key Results), and what can we conclude as a result?”
The challenge is finding the right time to move from inspiration to data. You can always instrument your business and gather metrics, but that doesn’t mean the data you gather will be useful.
For example, if you haven’t achieved product-market fit, instrumenting every aspect of customer acquisition is overkill. Just track metrics that help you improve the product. On the other hand, if the company has started scaling and you’re spending money to fuel growth, being late to instrumenting and analyzing your data can prove a costly mistake. The key is to be thoughtful, rather than formulaic.
I remember an early Airbnb board meeting where one of the other board members said, “It’s critical that we need to measure and track our operating margins.” Quickly, I spoke up to disagree. “Don’t do that. This is a digital marketplace. At scale, the margins will be fine, which means it doesn’t really matter what they are right now.” That was the right answer for Airbnb. Yet for a different company, like a hardware business, understanding operating margins might be essential.
One of the most important reasons to instrument the business is to help you drive growth. It’s hard to acquire more customers and drive more engagement if you can’t measure what is driving customer growth or where they’re spending their time in the product. And don’t forget to measure the opposite as well. I’ve seen entrepreneurs make the mistake of focusing all their efforts on customer acquisition when they also needed to fix a leaky funnel or customer churn. Growth may mask such problems temporarily, but they can prove fatal if left unaddressed.
In addition, even after you have made the transition to data and built a dashboard, you need to iterate and keep that dashboard up to date. As you review your numbers and reach conclusions, you need to be thinking about not just: “How true are these conclusions now?” but also, “How true will they be three months from now? 12 months from now? 36 months from now? And further, what are the things that will change that truth?”
The business intelligence you develop isn’t a fundamental natural law, like Planck’s constant or E = MC squared. The market can change in a way that renders your previous heuristics false. Or, you might proactively change your business in a way that requires you to rebuild your dashboard to account for an additional revenue stream.
Your collection and use of data also needs to take into account the complexity of the business, including the ability to use a specific subset of data to make decisions. Many times, a global view of data is sufficient, like the virality curve at PayPal. But other times, you need the ability to make micro as well as macro decisions. At Uber, for example, the relevant unit of analysis wasn’t the entire business, but the individual city. Without the ability to consider city-level data, Uber would have been flying blind when trying to write a playbook for launching into new markets.
A final word of advice: No matter how comprehensive your instrumentation, or how rigorous your analysis, you should retain some humility when it comes to the power of data. Data is so absolutely essential, but you have to make sure you’re getting the right data. The data that is relevant may change over time, and you will likely need to make important decisions when the data are incomplete and insufficient. This is why data is essential, but so are judgment and instinct. When you can’t use data to validate or invalidate your investment thesis, one of my favorite techniques is to talk with my smart friends and get their feedback. It’s not as good as dispositive data, but it offers a quick way to make decisions in its absence. You have to ask the right questions. “Will this work?” is a bad question that will likely result in bad data. Instead, ask for specifics like, “If this ends up failing, what are the most likely factors that will have led to that failure?” Or, “Here is my plan; what could be a better way of accomplishing the same objective?” Even if the answers aren’t definitive, they can help you distinguish between a helpful gut instinct and feel-good wishful thinking.
Common Decision-Making Pitfalls
Another way to improve your decision making is to develop your ability to avoid common decision-making pitfalls. The good news is there are many resources that can help you do this. One of the resources that helped me was the Harvard Decision Science Laboratory, which Professor Jennifer Lerner ran for many years. The researchers there worked to catalog the various decision-making biases that can trip you up, like the sunk cost bias or confirmation bias. When you’re considering going by your gut instinct, it’s a good idea to crosscheck your feelings against the most common biases.
For example, there is a concept from science and psychology of “Type 1” and “Type 2” errors. Type 1 errors are false positives, Type 2 errors are false negatives. I have some personal experience with Type 1 errors from Greylock’s investment in Groupon. At the time, the company was the fastest growing business of all time, and I projected out the growth curve and thought that we were headed for an all-time positive outcome. The problem was that Groupon’s revenue growth was a false positive. Neither the merchants who offered Groupons or the consumers who redeemed them were satisfied with the core coupon product, and quickly abandoned the service. But the company had such a massive and effective sales engine that it was covering up the fact that the business was a leaky bucket that would decline when the sales engine ran out of new prospects to convert.
Something similar happened with Dropbox, albeit in a more benign way. One of the key data techniques that investors rely on is cohort analysis. When we looked at the Dropbox business through that lens, it looked amazing. Each cohort of Dropbox users grew more valuable over time, and newer cohorts improved on older cohorts. This is incredibly rare; most businesses have cohorts that decline over time, or at best, reach an asymptotic maximum. Dropbox’s cohorts were increasing. Our Type 1 error was to overly generalize from this data and assume that the same dynamic would apply to any additional products that Dropbox introduced. Alas, the company’s product additions tended to follow the traditional value curve.
Fortunately, Greylock still made money on both investments because we invested early enough in the compounding cycle, but had we been smarter about avoiding Type 1 errors, we might have helped Groupon address its leaky bucket, and Dropbox reinvent its additional products more aggressively, producing a better outcome for everyone.
In each of these cases, there was data – a metric that appeared to confirm what we wanted to believe – that we had made a great investment that would be a massive success. But because we didn’t look beneath the surface and ask, “What’s behind this data? What individual components and underlying forces are at work?” we allowed ourselves to be complacent. The lesson here is that the data doesn’t have to be wrong in order for it to mislead you. Data doesn’t speak for itself; its meaning almost always depends on context and interpretation. It’s up to you as the decision-maker to take the necessary steps to accurately assess the situation.
One of my favorite techniques for interpreting data is to make sure I always have an active theory of the data. Back at LinkedIn, I saw Jeff Weiner apply this technique in a stunning and amazing way. Jeff calls himself an infovore, which means he’s also a datavore. Part of how he managed LinkedIn so effectively was the way he used different dashboards to make data-driven decisions. Jeff was such a maestro of these dashboards, that I can recall multiple times when he’d be looking at a dashboard, look at a number, and realize something was off. “Something’s wrong,” he would say. “That number should be X, and instead it is Y. Call the product manager in charge of that number.” He had an active theory for how the data should look, and this allowed him to be much more sensitive to deviations from that theory.
Each time this happened, we would check with the product manager and discover that the email notification was down, or the instrumentation had been misconfigured, or the log file was delayed, and so on. Jeff explained, “This is my active theory about how these numbers work. If the numbers are off, I pursue the issue. And afterwards, based on what I find, I update my model to account for any changes or refinements.”
Unusual Data Sources
A final way to become a better decision maker is to tap into sources of data that others overlook. Correctly applying the data you obtain from “unusual” sources can confer a significant competitive advantage. While I can’t very well share these kinds of sources in a public essay like this and expect them to remain unique, I hope that the examples I provide can inspire you in your own search for unusual data.
For example, one interesting way to acquire such data is to synthesize data from multiple discrete sources. You can buy credit card data (anonymized, of course) from a number of sources and use that data to identify buying patterns and trends that you can leverage to grow your business.
Another source of data that we made extensive use of at LinkedIn are job postings. Of course, that was easier for LinkedIn, but anyone can obtain this data. Job postings can give you a sense of what key companies are working on and what skills and products are in demand.
And of course, there are any number of analytics companies like App Annie that sell or even give away data on trends in the mobile space.
All of these data can be combined together to generate unique insights into your company’s particular situation, making you a better decision maker, and giving you a competitive edge over the other players in your space.
Data and VC Investing
One of common misconceptions outsiders have about venture capital investing is overestimating the importance of data in the decision-making process. One of the fundamental principles of venture capital is that size does matter. If you don’t have a sizable market, you won’t achieve a venture return. Time and time again, I see pitch decks that are packed with statistics from various market research firms in an attempt to precisely qualify the Total Addressable Market (TAM), Serviceable Available Market (SAM), and Serviceable Obtainable Market (SOM). There may be investors who are comforted by seeing this thicket of acronyms, but I am not one of them. I don’t believe that having an investment banker or industry analyst call a 100-person sample tells you much about the market.
I ignore these numbers and focus instead on the essential question, “Do I think this will grow to be an interesting market?” In the kind of investing we do at Greylock, which is primarily early stage (Seed, Series A, and Series B), we try to answer this question by asking, “What could happen to increase the size of this market?
The canonical case for this approach is Uber. Wall Street’s “Dean of Valuation,” Professor Aswath Damodaran of NYU, painstakingly calculated the possible value of the company by assuming a scenario in which Uber successfully captured 10% of the global taxi market. As a result, he concluded that Uber would never be worth more than $1 billion. What this analysis missed is how Uber’s market is far broader than serving as a simple taxi replacement. The lower cost and greater convenience of ride hailing convinced consumers to substitute Uber for many other goods and services, ranging from airport parking lots (remember those?) to daily commutes. And that doesn’t even account for expansions like Uber Eats.
Making better venture capital decisions requires imagination and judgment, not just a calculator.
Mastering the techniques of good decision-making is one of the most impactful things entrepreneurs can do. And making correct decisions more quickly, often by developing the ability to act on incomplete and uncertain information, can help blitzscaling entrepreneurs beat their competition.
The challenge of good decision-making is you can’t simply follow a simple set of rules. While you can learn techniques and frameworks, so much depends on collecting and analyzing data, applying human judgment, avoiding common pitfalls, and updating your models based on outcomes. But those who master this process, like Jeff Weiner, can produce outstanding results and have a greater impact on the world.