In an industry built around investing large sums in uncertain ventures, the best companies seek out the best investors, and gains accumulate at the top. Part 3 in a series.

Originally Published on NewCo Shift.

Investing in software companies is inherently an uncertain activity. It’s called high risk, but highly uncertain is a better label. Yes, you’re taking a risk with money, but the real problem is the widely variant potential outcomes. If you invest in a restaurant, you are taking a risk but you will pretty much end up with a profitable restaurant, or lose your money. If you invest in a software company, you can go bankrupt, have a small but profitable company, sell for five times the money in, or end up with a world-spanning multi-billion dollar behemoth that turns everyone it touches into a millionaire. That dramatic range is why you can get a bank loan to start a restaurant but not a software company. It’s exactly why people invest in software, but also why it’s so difficult to do well.

There’s no proven method for managing that uncertainty. The most successful investors frequently get it massively wrong, and a playbook that worked perfectly in one circumstance falls flat on its face in so many others. Yet, even when they frequently make monumentally bad investments, the best investors keep delivering the best outcomes. If no one knows what separates the best from the rest, how can some firms or individuals keep winning?

Of course, many would disagree with my claim, they would say the best keep winning because they can tell a great company from a bad one, but if you look at the trends in venture capital you can see the industry as a whole has given up on a clear system, even if individuals still cling to deserving their greatness. Fantastically successful investor Paul Graham once said that he can be tricked by anyone who looks like Mark Zuckerberg. He’s since claimed that was a joke, but he built his empire by making more bets than anyone else, which is a strategy explicitly designed around the fact that he doesn’t actually know why some things succeed and others don’t. Disgraced investor Dave McClure started 500 Startups with the similar goal of just making lots of bets, rather than making any attempt at making “good” bets.

Even those who invest in venture capital firms have given up on knowing who’s best. Given a pot of money allocated to VC, limited partners will distribute it across many firms, knowing that they have to play many hands to get a winner. After all, the industry average return for venture funds is, ah, absolutely nothing. The winners win big, and the rest balance that out, so LPs need to put money in enough places to be confident they end up with the winners.

If no one knows the difference between the best and the worst, why do the winners usually keep winning?

Access.

Venture capital is all about access: Founders having access to capital, and investors having access to the best deals. If you’re a founder today and you have a choice between taking money from a top-tier firm that keeps delivering hits, or another firm you don’t know and who hasn’t done well, which do you take? Of course you take the best firm with the biggest network and most well-known brand name.

Similarly, if you’re an investor who’s helped take lots of companies public, how does your deal flow compare to those who are just starting out and who don’t have a reputation for building big companies? Of course the best companies come to you.

In other words, there’s an implicit matching algorithm, where companies that are obviously doing really well are able to work with what look like the best firms, and as a result they are able to reinforce each other’s success. The best firms look better because the best companies seek them out, and the best companies do better because they’re getting the chance to work with the best brands. (For all that I am skeptical of repeatable investment skill, I am a deep believer in the value of brands.)

Venture capital is defined by the asymmetric stresses pressed on investors and founders by the need for access; every entrepreneur stresses over how they’ll get access to capital, and every investor’s business model is built around managing deal flow. Entrepreneurs who already had a great outcome magically have no trouble raising huge amounts of money, and yesterday’s great investors have no trouble convincing today’s great companies to work with them.

This focus on access also helps to explain some of the churn the system experiences. If no one knows what makes a great company, how can the best investors always get access? They can’t. There are plenty of great companies that fail to get first-tier support early on. If they do raise money, then those who backed them end up looking like tomorrow’s geniuses, and the cycle starts over with them closer to the top.

This access-based sorting also helps to explain how the VC industry is so discriminatory. Less than 5% of investments go to women-led companies, and just having a woman founder ruins a team’s chance of getting funding , and the numbers are as bad for firms led by African Americans, for example. If we believed investors actually knew what they were doing, then we could only conclude that they were correct to exclude women and minorities from investments, that these founders just couldn’t build great companies.

Of course, the data clearly says otherwise: Founding teams with women on them significantly outperform male-only teams. Because investors don’t know how to pick a good company from a bad one, they are relying on access and reputation, and because they’ve never let women or minorities in before, they can’t now. Their “pattern matching” doesn’t hit here.

This matching algorithm that runs our industry is reliant on privilege and luck. Venture firms and founders are almost exclusively white men from expensive schools (with a huge proportion from just Stanford and Harvard), and if you were lucky enough to be an early employee at Facebook or Google (who have historically used the same sourcing requirements), then that’s a big leg up, too.

To be clear, I think some investors are much better than others, and entrepreneurs haven’t built huge, successful businesses out of sheer coincidence. It’s not that there’s no skill involved, or that the people who get so rich instead deserve nothing. It’s that skill is an over appreciated (and often small) part of what determined their success.

You will rarely find communities admitting that privilege and luck are what determine outcomes. Human nature itself has a deep aversion to accepting this. Instead, we do what humans have done forever: We develop myths.

Humans deeply believe that people get what they deserve, and deserve what they get, despite the evidence to the contrary. So many of our cultural biases are a story created to justify a reality we would like to perpetuate. For millennia we’ve been told that royalty was special, and that’s why they were in charge, when it was patently obvious that their ancestors were just the best and most ruthless at organizing enough troops to control a chunk of land. Genghis Khan was history’s greatest murderer, which enabled him to spawn kings and kingdoms that lasted for seven centuries, but you can bet his descendants didn’t use his skills at genocide as justification for their lofty positions.

Similarly, myths have grown around venture capital exist to explain the winners and losers. Somewhat like royalty, these myths help convince us that VC is more than privileged people using their positions to make lots of money. They must be winning because they deserve to win. Equivalently, people lose because they didn’t deserve to win. You could waste your life reading about how this founder got rich because they were smart and worked hard, or that investor succeeded because of their investment strategy, but you couldn’t consume a whole morning with the stories of equally smart founders who worked just as hard but went broke, or investors who applied that same strategy but somehow didn’t make it on the Midas list.

Thankfully, we’ve seen some really interesting experiments focused on eliminating access as a criteria for investing. Social Capital recently launched a programmatic investment algorithm, and Village Capital uses peer decision making between entrepreneurs. Backstage Capital was founded explicitly to invest in those who can’t get capital from the system as it exists today.

With these and related efforts, I’m optimistic that we can begin to peel back the myths about what makes a great investor, entrepreneur, or company, and instead begin building a more open market around investment and company creation. Only then can we hope to see venture capital include, enrich, and benefit all parts of the economy.

It’s time to develop new sources of capital for founders, to help them generate wealth through solving their customer problems without the massive failure rate. Part 7 in a series.

Originally published on NewCo Shift.

The venture capital world that funds the technology ecosystem appears to be specially designed to back the best founders working on the economy’s most important problems. This series has shown that it has instead evolved over time, with no higher purpose in mind than any other financial instrument.

This evolution is in many ways a strength, as by definition it is built on the successes of the past, but it leaves our ecosystem more blind than we realize. We can fear the fragility this engenders, but should instead see it as an opportunity to reach beyond its artificial limitations, to solve hidden or devalued problems. Technology funding’s demonstrated ability to change should give us confidence we can stretch it further, clearing new paths to success.

Wikipedia covers the history of venture capital better than I could, but it’s worth highlighting key epochs. The system as we know it was birthed by the windfalls from early funding wins, including DEC and Fairchild Semiconductor, so by definition there was no technology funding system in place at that point. Every deal involved people flying around the US collecting enough money to back a new venture.

These early big successes motivated a few people in the west to set up firms dedicated to funding technology companies — prior to this, the vast majority of American capital was in New York. Within a couple of decades, partially enabled by some regulatory changes in the US, there were enough firms around (including modern heavy hitters like Sequoia and Kleiner Perkins) that we had what felt like the first stable system, which of course led to over-investment and the first pull-back in the late 1980s.

What survived went on to fund the internet boom in the 90s, when a huge amount of wealth was created (and then destroyed) and this new ecosystem first made it into the public consciousness. Much of what we believe about venture capital comes from those days, but it was still changing quickly, with no seed funds, relatively small amounts of funding for software companies, and no obvious pattern of success.

Right now, the system looks dangerously stable. There are hundreds of seed and venture funds, all following the same playbook: Try to get their investments to the magic number of $1m in annual recurring revenue (ARR), raise an A round of funding, and keep on the funding train until you go public or go bust. There’s so much pattern matching going on that founders are contorting their companies to fit the funding schedule rather than discovering their destiny.

It’s important to recognize that this appearance of stability is a recent arrival. We might tell a story of how it’s a natural consequence of previous eras of success, but much of current best practice is cargo culting, copying the behaviors of the successful rather than understanding what made them work. If you step back even a little to gain perspective on the industry, you quickly see how much the system is still changing, and still needs to.

Don’t get me wrong: the system we have works. It is, essentially, functioning as intended, and any ideas or recommendations need to take into account not just what we dislike, but what makes it work. I hope this series has educated you somewhat both on how VC works, and why it works that way. As usual, when we dig deeper we find no villain at the heart of a web; we might not love venture capital, but it makes sense, and it works this way for good reason. And indeed, the system is working very well for a few people, and in the process is driving huge change in our economy and lives.

As much as the system of venture capital makes sense, we must ask: What sits outside? The industry generates money through positive feedback loops, but absence from the industry is merely an indication that something hasn’t worked, not that it can’t. What are we missing by doubling down on what we know, instead of exploring the unknown?

Investors are reliant on people near them, who resemble them, and who can absorb the weighty downsides of entrepreneurship. We’ve seen that investors don’t really know what separates great companies from bad in the early days, so they don’t strive to create the conditions necessary for gestation, and once a company is started, they do little for the winners and even less for those who fail. But don’t worry, all of this is hidden by the massive profits that the biggest winners generate for the top-performing investors, and the rest of the industry (while failing to meet its investment return goals) glides along in the afterglow.

(To think I was asked recently if I had become cynical about venture capital.)

It’s a funny thing. I grew up a communist (literally, on a commune) but have become a pretty big fan of well-regulated open markets (although they seem to exist only in theory; in practice we have lost the taste for effective regulation). A self-respecting capitalist can and should argue that this is a market, and it’s performing exactly as it should. I can hear it now: “Capitalism is inherently Darwinian, where evolution gives all prizes to the winners and the losers don’t live long enough to make it into the archeological records.”

It’s a fair point. Humanity can afford stretch goals like less collateral damage than the battle for life and death on the savannah, but we could ask for better even without that ideal. It took millions of years for nature to come up with the Dodo, only for it to promptly die off once it encountered outside species. How convinced are we that our apparently stable system is any more safe from an outside force?

The ultimate weakness in the Capitalist defense of venture capital is that for all the apparent competition we have a homogenous system. Shouldn’t we have multiple types of funding competing for the best companies and the best outcomes? That is, not competition between VCs who all work the same, but competition between different funding strategies?

Because there is no open market here. At best we have a dysfunctional oligopoly (is there any other kind?) with some churn at the top. For all the talk of disruption, everyone is trying to win by copying the winners, rather than seeking to disrupt them. The only people willing to step outside the current system are those who don’t have a choice because they aren’t allowed to succeed within it. Unsurprisingly, they find it challenging to compete not with another investor but with a whole system of funding.

As just one example, the most common barrier I hear to starting a new kind of venture capital is that the limited partners — that is, those who invest in the venture capital funds — would not be willing to support a new kind of capital. This is a perfect example of an ecosystem limitation, rather than a problem with individual players. I hear no argument that founders, employees, and customers don’t want competitive models; only that the source of capital would need to be educated, and that’s just too hard. Except… this whole industry is only a few decades old, and its creation required that same kind of education. Why should we expect a new kind of financing to be any easier to start, require any less systemic change, than the one we’re fighting against? And isn’t it ironic that an industry built on stories of disruption finds the idea just too hard for its own work?

That competition will show up eventually, though. We need it. There are too many software markets lying fallow, unfundable in the current model and thus deemed to be of no value. Someone will figure out how to finance those companies. And just as the first winners in venture were big winners indeed, the first few investors to step out of this world into a new one should make out like the oligarchs who laid the groundwork for our current world.

I’ve said before I don’t have the solution, but there are some market truths give me confidence there are better answers available:

• The best way to make money is to hold high quality assets for a long time. If nothing else, Warren Buffett has demonstrated this is both the best way to make money and indefinitely scalable.
• The majority of employment and wealth generation is provided by companies too small or too closely held to be public.
• The steady state of good companies is cash-flow generation managed by long-term teams who take pride in their work. This is literally the entire history of for-profit enterprises. Any other solution must either fail or revert to this at some point. None of these realities show up in modern venture capital. Companies can’t run on venture capital forever (although try telling that to Uber), and do usually need to show a profit to be sustainable (I expect Amazon begs to differ), but the companies that do either of these are explicitly leaving the world of venture capital.

It’s unquestionable that the financing structure of venture capital is tied in to this separation from market principles. The risky software companies we build today are funded via a structure invented to support the risky ventures of the 19th century: whaling. Suddenly the term ‘venture’ in venture capital makes more sense, doesn’t it? (Tragically, even though it was the days of slavery, those whaling fleets had better representation in some ways than current tech companies, with up to 20% of their employees being African American. Wow.)

We’re using an incentive structure that works perfectly to support individual voyages that might last a couple of years. Is it any surprise it is not great at building companies that last for decades, or have a high survival rate? In fact, whaling had a better survival rate than current venture capital, with more than 80% of the fleet surviving, and delivered better returns (14% IRR on average, and 60% IRR for the best). The funding perfectly matched the ventures.

I should not need to say this, but whaling is unlike company building. It’s unrelated to developing a product, it has nothing to do with creating a new market. It’s inanity to expect a funding mechanism built for one would work as well for the others. The fact that it’s making some people rich, and it hits a jackpot once in a while, should not confuse us.

Venture capital’s apparent stability convinces me it’s at its most vulnerable. Instead of continuing to fund disruptors, I think it will itself be disrupted.

If you’re a founder given a choice between a firm that kills most of its customers and one with demonstrated success at creating long-running companies that generate wealth for everyone involved, why would you pick venture capital? The only reason you do today is because it’s your only option.

Founders want this competition right now. Some want to build Facebook, but most want to build a great company, help their customers by solving a critical problem, and hope to get rich along the way. They don’t want a lottery ticket; they want upward mobility, entrepreneurial fulfillment, and to feel like they made a difference. Unfortunately, low-probability gambling is all the venture world sells.

The new models will start at companies run by women and people of color, because they’re the ones shut out of the current system, but as they start to succeed, they will start to pressure to rest of venture capital, and we will see just how stable the system really is.

I have tried in this series to help you understand not just what venture capital is, but that what you love and hate about it are intrinsic to how it works. I hope this deeper knowledge will help you make higher quality decisions about how to involve yourself in this world. Even more so, I hope it convinces you to seek out, or even create, other ways of funding companies, other ways of building them.

It’s time for founders to have truly competitive options for funding. Let’s go make it happen.

Software has the potential to increase productivity as much as electrification or steam power did, but its impact is stunted by its reliance on random interactions. Part 6 of a series.

Originally published on NewCo Shift.

The venture capital ecosystem bills itself as a meritocratic miasma of genius, with smart founders getting smart money from smart investors. In reality, there is an overwhelming reliance on privileged people bumping into each other at just the right time. This serendipity has spawned some great companies:

• Warby Parker was started because someone in an elite graduate program lost an expensive pair of glasses.
• Apple was started by a couple of guys who met at a hobbyist group in the computer heartland.
• Google’s founders met when one of them gave a tour to the other when he arrived at Stanford for a CS graduate program.

But how many great problems are being ignored because we didn’t get that lucky alignment of particles?

The remodeling industry is a perfect example. It’s an 83 billion dollar market, yet it’s only now starting to see software solutions. The industry itself bemoans neglect by the software industry. The article linked above has some impressive stats about how much waste they experience:

“…studies suggest 30 percent of the construction process is re-work, 60 percent of labor is wasted, and only ten percent of losses are due to wasted materials”

Shouldn’t there be companies fighting tooth and nail over that market? Shouldn’t there be tons of solutions out there, spending money like Uber and Blue Apron are to acquire new customers and take a cut of the productivity gains?

Yet I’m in the late-stages of having a garage built at my house, and as far as I can tell software was only used during design, not actual production. One of the contractors we considered seemed to be an Excel wiz, but wasn’t using off-the-shelf software. How many months of productivity could have been added back into these teams’ lives if they had better tools? How much less disruption could I have experienced, and even, how much less could I have paid if my contractor could get three jobs done in this time because she was so much more productive?

(Did you notice that even I’m relying on the serendipity of my building a garage to illustrate my point that VC relies too much on it?)

In a rational world, every reasonably sized market would have a well-funded ecosystem of software companies vying to take it into the information age. When I got my home equity loan for the garage, I should have been inundated with offers from software companies to help improve the project. Heck, someone should have offered me the loan interest-free if only I required my contractor use their software. Instead, my project is late, costs me more, and makes less money for all the workers because it’s left out of the information technology revolution.

And that’s just one industry, chosen at coughrandomcough. What about all of the other industries the software kings have not yet anointed as worthy, filled with deeply skilled and energetic experts who aren’t lucky enough to run in the right circles, or live in the right zip codes?

Venture investors famously want passion for the problem they’re investing in solving, so much so that the companies also then demand that any employees also be passionate in turn. And we want our software companies started by developers, by product people, not by business analysts. Or carpenters.

So now to start a company you’ve got to have a software developer thrilled about and experienced in a problem, able to accept the risks that come with starting a company (e.g., health insurance and wage loss), who is living in or can move to San Francisco, and hopefully is a white dude who went to Harvard or Stanford. One way to look at that is how discriminatory it is, but another way is just how much you’re relying on everything lining up just right. It might be that you’ll find a Stanford-educated software developer who deeply cares about building houses and can take the leap into entrepreneurship. But what are the odds that that person has the right insight at the right time, and then can find the right people to partner with?

Twenty years in I’m still awed by the opportunity for software to connect, educate, and empower people, but I’m incredibly disappointed by how little of that opportunity we’re progressing against. I think our inappropriately slow revolution is in large part thanks to this reliance on randomness. We have got to get past this if we truly want to get the most out of software before the heat death of the universe (coming more quickly now with all the power being consumed to mine bitcoin). If we can build an environment that does not use serendipity as a crutch, I am convinced we can generate more great companies, and importantly these companies can cover a broader swathe of the economy, and be run by a more representative sample of the market.

Let’s look to biology to see how much of a difference shifting to a constructed environment can make. Living creatures are full of enzymes, which are basically proteins that speed up the rate of a reaction. These reactions are critical to the function of the organism, and without the enzymes speeding them up, life as we know it could not exist. (Conveniently, I did my senior thesis at Reed College on protein structure, so I’ve got some knowledge here.)

In most cases, the reaction that they catalyze (that is, cause to happen) would happen without the enzyme, but it would do so at a far slower rate. For instance, mammalian milk contains the sugar lactose. This sugar will break down in water into glucose and galactose of its own accord, but not quickly enough to digest all the lactose in milk you drink. Mammals have evolved the enzyme lactase, which causes this splitting of lactose into simpler sugars to happen much faster.

Enzymes are incredibly complex — lactase has 1927 amino acids in five separate groups, arranged in an amazing 3D structure:  A rendering of the structure of lactase This huge structure is all necessary to enable the protein to place a lactose molecule near a water molecule in exactly the right arrangement to ensure the reaction happens immediately, every time, instead of eventually, sometimes. For all this structure, the site where the reaction takes place is quite small, just big enough for the two target molecules. Those 1927 amino acids mean the protein is about 37,000 atoms. Lactose is 35 atoms, and water is, ah, 3.

That’s a lot like designing a building the size of a sports stadium just to catalyze a meeting of two people.

How much quicker does the enzyme work? About 75% the world’s human population is lactose intolerant, meaning that if they drink milk as an adult, the lactose will cause adverse reactions instead of safely being broken down in the intestines. The rest express enough lactase that they are able to comfortably metabolize lactose, and thus can drink as much milk as they want. Again, remember that lactose breaks down in water on its own, just too slowly to be useful.

So here we have a situation where one of the major sources of calories around the world — cow’s milk — is enabled by this enzyme dramatically speeding up reaction rate.

What does this have to do with venture capital?

Again, venture today is heavily reliant on serendipity; that is, the right people bumping into each other at the right time in the right context. This is exactly how chemical reactions happen normally: Two molecules (e.g., lactose and water) live near each other, and every so often they bump into each other in a way that enables the reaction to happen. Most of the time, however, they fail to hit exactly the right setup, and nothing happens.

When the enzyme is present, though, its unbelievably complex structure ensures that the water and lactose molecules are placed into exactly the right orientation every time, and bam, magic happens.

The probability of a great company getting founded today is a lot like the probability of lactose degrading naturally: It happens, but slowly and infrequently.

I smile at the idea of complexes the size of sports stadiums built for speed-dating founding teams, but that’s not necessarily what I’m recommending here (although if that’s your plan, I’d love to consult on the project).

Even if we wanted to, I don’t think we could build a structure (physical or otherwise) like this, because we don’t yet understand yet what it takes to build a great software company, which means we can’t construct or evolve a perfect environment in which to make it happen.

All I really know is that what we’re doing now isn’t working. We’re not attacking the right markets, we’re not including enough people, and we’re not having a big enough impact on the economy.

For our ecosystem to be healthy, for it to be effective at transforming the industries that need it most, it has to do something differently. We can really only increase the rate of great company creation by increasing the rate of experimentation, or increasing the rate of success. Incubators and early stage investors are doing what they can to run more attempts in parallel, somewhat like a generative algorithm, but this is bound to have little impact because the goals — “be worth a billion dollars” — are so separate from the founding event. Investors are starting to figure this out and pull efforts back accordingly.

That leaves us the challenge of finding ways to increase the rate of success.

Of course, I have my own ideas for doing so, but I was always told as a leader my job was to present the challenge to the team and leave the problem of solving it to them.

Consider yourself challenged.

Giving something away for free can be a fantastic way to achieve adoption, but it raises a permanent conflict. The conflict is even greater when the free products are open source.

Driving adoption is a key requirement for any product business, and a popular tactic for doing so is hooking people with something free in order to sell them products once they’re invested. The most famous example is Gilette “giving away” razors, but this model is now widely adopted by software companies striving to grow quickly but cheaply. Splunk and New Relic are examples of fantastic businesses who allow their products to be used for free in some circumstances in order to drive long-term adoption and revenue.

Success with this tactic requires mastering the tension that it introduces: The thing you give away must be valuable enough that it drives adoption, but not so valuable that people don’t eventually need to upgrade.

If it is insufficiently valuable on its own, no one will use it, but if it is so valuable that people can succeed without ever paying, then you have users but not customers.

The goal of a freemium business is to separate the commitment from the buying decision. I want you to become so successful with my product that you become committed to it, which means when I finally ask you to pay the sale is both easier and more valuable.

For you to become committed, you must be successful. You must achieve whatever goal caused you to download my software in the first place. However, in your success, you must also hit some wall that requires you to upgrade.

Managing that balance is crucial to success with freemium. Give too little away and you don’t get any adoption, but give too much away and you have a passionate community but no money to pay for building the software.

This balance is complicated by the fact that the best communities are built around authentic identities, so changing what you give away involves changing who identifies with your company and why, which is much more fraught than tuning your business model. It’s a sufficiently complex operation that it’s one of the key dials that a freemium business will manage throughout its existence.

As difficult as this balance is in a proprietary software business, it’s much harder when dealing with open source. After all, when you decide to open source something, you’re not just choosing to make something free, you’re giving your community permanent rights to something you own. While you technically have the right to keep private any further updates, in practice once you open something up you have to either keep that project free forever or let it die.

Your community considers the opening of your code a permanent commitment to maintain this software for free forever, and anything less than that will generally be considered betrayal. They didn’t adopt you, or your revenue streams; they adopted this free thing that you promised would work well for them, and they’re going to raise hell when it doesn’t. They don’t care that your business model requires that you charge for analytics, or for clustering. They need those features to be successful, you implicitly promised them success, and now you’ve got blood coming out of your ears.

In some sense, successful open source software companies require lightning to strike twice: They need to free something that it is so great you’ll adopt it without paying, then they need something else that’s also so great you’ll pay them. Most OSS companies try to provide this extra value at the margins, because getting that second lightning strike is so hard. Proprietary software companies can provide hard limits on how you take advantage of their most useful features, but if those features are open source, then of course you can’t limit them in any way. You’ve got to develop additional features instead.

It’s not a coincidence that we’ve seen so few open source software companies build large revenue streams. Most end up shifting their focus from the open source that started everything to proprietary software that drives their business. The best companies remain community-oriented, but that is by no means a given.

If you’re considering building a freemium business, you must treat the success of this balance as a key business metric. It underpins the unit economics of your business, determining how big the funnel is based on adoption and how many convert based on the willingness to pay.

Getting it right is worth it because it enables a fantastic company with low marketing and sales costs, an engaged community, and easy upgrade paths for your best users.

Don’t get stuck on a solution until you know what problem you’re trying to solve.

I often get called in to help companies make decisions about their open source strategy. They want to release key parts of their software as open source, but they need some help figuring out the best way to make it happen. I always ask them the same question:

Why? Why are you planning to open any of your code?

They rarely have a good answer. They’ve already decided that this is the right decision, because a board member, founder, or customer has said it’s necessary, and they are just trying to figure out how to do it. But it’s impossible to build a strategy to accomplish your goals if you’re unsure what they are.

Are you trying to build a community? To get public review of core functionality? To grow adoption? Something else entirely? By now most people have realized that open sourcing software isn’t a route to magically get free contributions so you don’t have to write your own software, but there are plenty of other myths around it.

I’m a huge fan of open source. I’m an even bigger fan of the companies behind it building a sustainable business so they can stick around for the long term. Open source is a tactic, taken to accomplish (hopefully) a key goal for the business. It has upsides and downsides, like most tactics, and it has to fit your goals.

There are damn good reasons to open source software, either as a business or as an individual. Most motivations I’ve heard from founders don’t stand up, though. Companies have built healthy communities (Github, Atlassian, VMware, Microsoft) and scalable freemium models (Github, Splunk, Slack, Trello), all with no notable reliance in publishing open source software.

The fact that these businesses have been built with proprietary software doesn’t mean yours should be. But it at least shows that your goals have multiple routes to accomplish them, and you can’t just say you need to open source your product because you want the same adoption path as MongoDB.

Many open source software companies don’t have the luxury of starting with goals and working backwards. If your project started first, and you only built a company once your community made it clear you should, then many hard questions are already decided, and you have to work around the reality of your situation. Docker is a great example of a company being built in response to community demand, because an existing project got adopted broadly enough that the opportunity arose to build a business (or in their case, shift the focus of an existing business), and some of their struggles can be attributed to the path they took.

I open sourced Puppet through a combination of ignorance and confidence. It honestly never occurred to me that I could build a company around commercial software, since I really only knew the OSS world well, and when I started Puppet, around 2005, there seemed to be plenty of great examples of successful business models around open source so I assumed I’d figure out which of those made the most sense when it came time. In the mean time, I did exactly what needed doing: I focused on building software that people loved, and then convincing people to use it. By the time I realized how hard it really was, we were too far down the path to dramatically change our strategy.

If, however, you are early enough that all options lie before you, you need to take the time to ensure you have the highest probability of success. It’s hard enough to build a company from scratch; don’t make it harder. Your team should actually view open source as a liability, in the financial sense: You’ve made a commitment (usually implicit) to your community that you’re going to spend money indefinitely to make sure this software gets better and better. Just like other financial liabilities, like office leases, employee contracts, and partnership commitments, these can be fantastic successes. But like any liability, they can just as easily be a painful drag on your work if done poorly.

I think open source can be a huge help in building community, widening your marketing funnel, increasing the quality of your code, and building software people love. But you have to do it with clear ideas of what you’re trying to accomplish and what you’re willing to do to make it happen.