The Cold Start Problem
SYNOPSIS When a networked product launches, it faces a chicken-and-egg problem: people need to use it for it to be worth anything. Think of Facebook, Slack, or Airbnb. So how do you start the very first network without a basis to work from? Andrew Chen, General Partner at Andreessen Horowitz, calls this the Cold Start Problem.The Cold Start Problem is Chen’s attempt to help us better understand network effects: how to solve the Cold Start Problem, how to scale network effects, how to manage growth plateaus, and so on. Chen’s Cold Start Theory is broken down into 5 stages: 1. the cold start problem; 2. the tipping point; 3. escape velocity; 4. hitting the ceiling; 5. the moat. SUMMARY1. “The Cold Start Problem”To overcome the Cold Start Problem, businesses tend to start with a single network—what Chen calls an ‘atomic network’. This is perhaps the most crucial idea in the book. Networked products tend to start small, in a single city, college campus, or in small beta tests at individual companies—like when Facebook launched at Harvard University. “Only once they nail it in a smaller network do they build up over time to eventually conquer the world,” Chen writes. Moreover, appropriate atomic networks are often smaller than entrepreneurs think. Uber’s early atomic networks were not cities like San Francisco; ‘5pm at the Caltrain Station at 5th and King Street’ is more accurate.Networked products should focus on the smallest network needed to sustain the product. Different products require differently sized first networks. For Slack, a small team within a company is enough for the platform to work. But, when the credit card was first launched by Bank of America in 1958, it was done so across the whole of Fresno, California.BoA understood that for a credit card to work, a large enough pool of people must buy in—enough for merchants and consumers alike to derive value from the new system. Despite the difference in scale—Slack with 4 or 5 colleagues, BoA’s credit card with 60,000 Fresno residents—the principles of atomic networks are the same. Start as small as your product will allow. Once the first network has been nurtured, the process can be repeated (when a product reaches its “tipping point,” which will be discussed in the next section).Attract the hard sideChen also distinguishes between the easy and hard sides of a given network. To solve the Cold Start Problem, products must, above all else, attract the hard side—sellers on a marketplace, content creators on a video platform, or in the case of Tinder, attractive women. Tinder launched on the University of Southern California campus. The founders leveraged their popular friends to promote the app at parties. Students had to download Tinder to allow party access. The next day, hundreds of hungover, like-minded students had a second chance at love via Tinder.Network density is crucial. However small the first network, its nodes must find value in the product and one node’s engagement with the next must be high. Simplicity is also crucial to a product’s success. Zoom, now worth tens of billions, has eclipsed apps like Skype and Microsoft Teams. The product is intentionally bare. According to Chen, Zoom is the perfect storm of killer product and viral capability.“Zoom’s simplicity is a strength when it comes to the company’s ability to grow its network,” Chen writes. “When the product concept and value is simple to describe, it makes them easier to spread from user to user.” Zoom, and dozens of other networked products, ensure those first customers are acquired without friction by making the product free. “It’s hard enough to build an atomic network; why make it even harder by erecting barriers?”2. “Tipping Point”Tinder’s success among University of Southern California fraternities and sororities—executed using parties among popular college students—unlocked other colleges in America. Tinder had built a few different networks the right way: focus on the right audience (in this case young students looking for love). At a certain point, Tinder reached the tipping point for network effects: building networks of engaged users became easy. The company had discovered a repeatable strategy.LinkedIn, like many others, utilized an invite-only strategy, which was successful for one key, often overlooked reason: by targeting a small group first and allowing them to invite whomever they choose, network proliferation takes place by itself. It is a solution that solves the hardest problem of all, because mid-level professionals—those most likely to use and benefit from LinkedIn—will invite other, similar people. Thus, LinkedIn reached its tipping point after roughly a week. It engaged its users, and was valuable beyond the early-adopter tech community.Besides things like market subsidization and invite-only strategies, other methods, like bootstrapping a product, can ensure products that rely on communities don’t dry up, à la Reddit (the founders would post on the site’s front page manually with dozens of bot accounts). This was necessary for Reddit to build momentum and gain a core user base. Organic users soon began to post their own content, which rendered the founders’ bot accounts surplus to requirements. But that kickstart was crucial. 3. “Escape Velocity”After a startup solves the Cold Start Problem and reaches its tipping point (when the startup of new networks becomes replicable), the next stage, at least for successful products, is Escape Velocity. This is when products scale their growth.Chen breaks network effects down into 3 types: Engagement, Acquisition, and Economics.The Engagement Effect is what happens when a product gets stickier (and more engaging) as more users join. Companies like LinkedIn, Facebook and Slack have tapped into the Engagement network effect well which allows them to drive up retention over time. The best companies do this in 3 ways.First, successful networked products create new use cases as a network develops. For example, as Slack becomes more popular within a company, new chats are created, where colleagues discuss all sorts (work-related or otherwise), which drives engagement. Second, products reinforce the core ‘loop’ of a product, where users in a network interact (for Slack this might be a manager who shares a file with a direct report, who in turn ‘closes’ the loop with the competition of the task). Third, products reactivate churned users.The Acquisition Effect is essentially viral growth via organic use—the network effect that powers the acquisition of new customers. PayPal is a good example of this viral effect. Initially, a company that struggled to envision the ‘perfect customer’, it eventually latched onto eBay, where PayPal was already used by hundreds of sellers (unbeknownst to the PayPal team). PayPal went with this and created its own ‘pay with PayPal’ badges to place on eBay items. When a product has a built-in feature that encourages collaboration, it can spread on its own. “This is the Product/Network Duo at work again, where the product has features to attract people to the network, while the network brings more value to the product,” Chen writes. Finally, The “Economic Effect” is where network effects improve business models over time via improved feed algorithms, increased conversion rates, premium pricing, and more.4. “Hitting the Ceiling”After a period of viral growth (Escape Velocity), even the most formidable startups hit the ceiling. The growth chart turns from a hockey stick to a squiggly line (if the company does well), where products plateau then return to growth, over and over. To maintain growth, networked products must remain proactive. “Dealing with the ceiling is a never-ending battle,” Chen writes.SaturationThere are a few causes of slowdown from ‘rocketship growth’. One is saturation. This happens when a product grows to dominate its market and has no more worlds to conquer. At the same time, the marketing channels a company uses become less effective over time (as with banner ads and email marketing), which Chen calls “the law of shitty clickthroughs.”When the network revoltsThis is when the ‘hard side’ of the network—the minority of users that create disproportionate value and as a result have disproportionate power—recognize their own influence and demand better terms. This happened when the most valuable Uber drivers demanded better pay and benefits. As a company grows enormous, it becomes difficult to keep everyone happy.Eternal SeptemberWhile the hard side evolves, the rest of the network changes too. In what Chen calls the eternal September, as a mainstream audience is reached, what made a product’s initial community special is lost. Usage becomes less appealing as the network grows larger.OvercrowdingAnother common way to hit the ceiling is through overcrowding, where the discovery of relevant people and content becomes hard. This problem must be solved before users start to leave. Solutions often include things like search functionality, algorithmic feeds, or curation tools. Startups that focus on bottom-up distribution (i.e. target other small customers first), such as Slack, Dropbox or Zoom, will inevitably see their growth slow. The problem is that smaller customers churn more easily because, among other reasons, they are more price-sensitive than larger customers (they are more likely to run out of money or change their business model, for example). Therefore it is common for a networked product to hit a ceiling after it builds its first atomic networks. To solve this problem, a startup should remain proactive with the addition of new features (and in the case of B2B, focus on enterprise sales). 5. The MoatThe Moat is the final stage of Chen’s Cold Start Theory and is about a successful network that defends its turf with network effects. Warren Buffett popularized the concept of the competitive moat. He argued that to make good investments, one should determine the competitive advantage of a company, and above all, the durability of that advantage. For networked products like Slack or Airbnb, their software and functionality can be replicated fairly easily. Instead it is the difficulty of cloning their network that makes these types of products defensible.Vicious cycle, virtuous cycleAll companies in the same field have network effects—it’s how you scale and leverage them that matters. Small companies have some advantages—namely speed, and a lack of sacred cows. Bigger ones have established relationships, manpower, and product lines to lean on. Small companies usurp bigger ones frequently (Facebook blew MySpace out of the water); big companies bat small ones away often (Airbnb swatted away copycat firm Wimdu). For bosses of companies both large and small, there are ways to navigate competition with the other.Cherry-pickingThis is when a company, usually a smaller one, focuses its resources to acquire a small network from another company. An example is how Airbnb snatched Craigslist’s shared-rooms idea and made an entire product with it. In this instance David (Airbnb) was the cherry picker; Goliath (Craigslist) couldn’t defend all of his networks. By the time Craigslist stopped Airbnb from its ability to redirect its users, Airbnb had already built its atomic network.Big Bang launches to Big Bang failuresWhen a global brand launches a swanky new product, excitement builds. In the context of networked products, this type of launch often fails. Google+, launched in 2011, faceplanted because of its go-to-market strategy. While its user quantity (raw sign-ups and monthly active users) was predictably giant—within months, Google announced 90 million sign-ups—user quality sorely lacked.Users heard about Google+ in the press, not from friends. Because of this, engagement was poor. Users averaged 3 minutes on Google+ per month around launch; in the same period, Facebook users averaged 6-7 hours per month. The launch of Google+ was based on hype, and it never had the strength of small networks that successful products have.Compete over the hard sideWhen there’s a battle between networks, it is the networks themselves that are up for grabs. To compete over the hard side is when a network directs its resources towards the defense of (or attack of) the highest-value-additive part of the network. An example of this was when Uber entered a fierce competition over drivers with the likes of Lyft and Sidecar.BundlingBundling is when a bigger network uses its resources as a launchpad into another product domain. Companies of this size can solve the Cold Start Problem and establish traction—provided the product itself is good enough. In a “stroke of product marketing genius” according to Chen, Microsoft bundled Word and Excel together to make Microsoft Office. An effort was made to enable interoperability between Office apps. The rest is history. Provided the product is outstanding and advances the industry in some important way, bundling can be a powerful tool to accelerate success.BenefitsAndrew Chen’s The Cold Start Problem is a unique, ambitious book full of insights. For the first time, entrepreneurs of networked products, such as social media platforms or online marketplaces, have a step-by-step guide they can use to navigate product launches: how to get off the ground, traps to avoid, methods to scale, how to compete either as a minnow or market leader, mental shortcuts for complex ideas, and more. With recent case studies, some of which he experienced first-hand, Chen has created terms and frameworks for all stages of a business, for methods that have served the world’s most successful people.