The Cold Start Problem: How to Start and Scale Network Effects by Andrew Chen Part IV — The Moat (Book Notes)
Chapter 29 — Wimdu VS Airbnb
In 2011 Airbnb encountered its first competitor. Wimdu was a direct copy of Airbnb, focused initially on the European Market. It launched with 90 million dollars in funding and in less than 100 days it hired over 400 people and had thousands of properties on its marketplace. Airbnb was only 2 and a half years old, raised 7 million dollars, and had 40 employees. Wimdu replicated Airbnb's efforts. They built bots that would scrape listings, descriptions, photos, and availability. They also booked Aribnbs and convinced people to also put their listings on Wimdu. The company built over 50k listings and was on its way to 130 million in gross revenue in its first year of operations. After this fast start, something crazy happened. Wimdu went to zero. The experience for customers was terrible on Wimdu. Airbnb made sure they set expectations low and then blew expectations out of the water ensuring they had a high NPS so people would tell their friends. Airbnbs competitors who took shortcuts couldn’t deliver in the experience. The hard side of the network wasn’t fully formed or curated to reflect quality.
Since Wimdu was looking to sell the company, Brian Chesky knew Wimdu’s founders could move faster than him for a year but he wasn't going to keep doing it. Airbnb’s strategy around beating Wimdu was simple. Build Airbnb for long-term success and ultimately build a better community and product. Airbnb quickly moved to improve support for international regions. Their first move was to internationalize the product by translating it into all major languages. Airbnb’s next move was to support adding multiple currencies in the app from 1 to 32. They then bought all the local domains. Next, they scaled up paid marketing using Facebook, Google, and other channels. Most importantly they finally pulled the trigger in putting boots on the ground. They would launch each region with a PR blitz.
Chapter 30 — Vicious Cycle, Virtuous Cycle
The products or services that have wide sustainable moats around them are the ones that deliver rewards to investors. A strong moat means that it takes a lot of effort, capital, and time to replicate a product's features and capital. It’s the difficulty of cloning products like Slack that makes their products highly defensible. As Airbnb described, the cold start problem of launching in a new city would need to reach a tipping point of over 300 listings with 100 reviews. This requires real effort because the networks are quite large. This is the moat.
For Uber, being dominant in NY did not help the company succeed in San Diego as its network effects were localized primarily to each individual city. This was always the critique of Uber’s business and lead to trench wars that needed to be fought city by city.
The Battle of Networks
In the battle of networks, the steaks are high. Looking at Airbnb and Wimdu but also Slack and hipchat or Uber and sidecar. Of each pair, one is a decillion dollar company and the other is ancient history. This happens because network products can lean towards a winner take all. When one product emerges as a winner in an atomic network. It’s just that group choosing their favorite app. If repeated enough times you create a winning player in the market. On a team or entire company. Sometimes they’ll converge on the same set of products. They will use the same collaboration tools to send important docs/ edit tools. A single app in each category tends to share a lion's share of engagement so that the team that uses Slack will not spend an equal amount on Microsoft teams as well. It’s usually one or the other.
It isn’t true, simply having network effects is not enough to keep people from copying your idea. Because if your product has them it’s likely that your competitors have them as well. Whether you’re a marketplace social network, workplace collaboration tool, or app store. You are in a networked category in which the network connects people and is governed under the dynamics of cold start theory. If every product in a category can rely on its network then it’s not about who is the largest. instead, the question is, who is doing the best job amplifying and scaling their acquisition engagement and economic effects. It’s what we see repeatedly over time. Myspace was the biggest social network in the mid-2000s and lost to Facebook with a focus on college networks with a focused product execution. Hipchat was already ahead but was upended by Slack. GrubHub created a successful/profitable multibillion-dollar food ordering company but rapidly lost ground to uber eats and door dash. Working in the categories of marketplaces, messaging apps, social networks, collaboration tools, or otherwise the good news is your product has network effects, the bad news so does your competition. it’s how you grow and scale that network that matters.
Chapter 31— Cherry Picking — Craigslist
Before Airbnb and Wimdu, there was Airbnb and Craigslist. The founders are billionaires. Today it is a massive horizontal network for many local categories. with 80 million listings per. month and 20 billion page views. Top 100 website on the internet and amazingly it operates with a staff of a few dozen people. A long line of startups had cherry-picked craigslist’s categories. This is famously called the unbundling of craigslist. For example Indeed from the job section, stub hub for tickets, Etsy for selling arts and crafts, and so on. When these subnetworks splinter off, it provides an opportunity to hit a tipping point in one fell swoop. Every dominant network might seem invincible.
Only one entry point is needed for an upstart to build an atomic network whereas an incumbent has to protect all its entry points. When the incumbent network does a bad job of this a new competitor can waltz right into the market. This is the core asymmetry of network-based competition. Airbnb provides the best example of this. On Craigslist there was a smaller category for renting out rooms. However, the experience was terrible. Sometimes there was accurate pricing and photos. But for often than not there was no way to easily check if there were certain dates available. Nor were there standard features like ratings and reviews it just didn’t work well. Airbnb started with a significantly better experience aiming to solve all of that. Just like Craigslist Airbnb had listings with maps, descriptions, and pricing but also extended the functionality with galleries of photos, reviews and ratings, integrated payments, reservations, the profile of the hosts, and much more. They threw up a series of listings with prices and ways to contact the host. in retrospect,
Finding The Soft Spot
New players in a market start with seemingly undesirable niche segments. Which are ignored by incumbents because they are focused on the most profitable segments of the company. initially craigslist seems huge. Network density beats size. Which atomic network do you pick? Why was room rental such a strong starting point? The initial starting point matters because some are more easily able to access network effects. In Airbnb’s case, the high value of every transaction and user stems from a shared room ultimately adjacent to travel in an industry where stays are often measured in the thousands of dollars in a single trip. This high economic value meant that Airbnb could quickly scale with the economic network effect. The high average order value for Airbnb meant that it could then use this revenue to power the rest of its business.
Switching over entire networks
Part of why cherry-picking can be dangerous for the incumbent. The up-start networks can reach over and directly acquire an entire set of users who have been conveniently aggregated on your network. Airbnb is an example again. It’s just software after all. The company not only includes craigslist but also turned the idea of the shared rooms into an entire product. They actually used craigslist users to advertise Airbnb to other users. How? Early on they added functionally to Airbnb where when they were done setting up their account on Airbnb they can post it on craigslist with photos, and details, are you interested, and got a question? Contact me here link. That drove craigslist users back to Airbnb. These features were accomplished not by using APIs provided by craigslist. but by reverse engineering a platform ad creating a bot to do it automatically. Any product that wants to cherry-pick needs to make its own destination and scale it.
Chapter 32 — Big Bang Failures- Google+
The Big Bang launch is often the strategy of the larger player in the market. It uses its advantages in size and scale to quickly overwhelm an opponent.
The Big Bang launch looks something like this. In Jan. 2007. Steve Jobs stood in front of a crowd of thousands at the Moscone Center in San Fransico and announced a new device; The iPhone, to the world. It was incredibly well-received.
Startups and teams working on new network products often look at this type of launch and work backward to emulate it. Sometimes this looks like a more wide launch across press, social media, and paid marketing. Maybe it’s accompanied by a big push from companies big product or perhaps a key partner sending over a tone of users all at once. A big email marketing campaign might go out as well. The intent is the same launch big with the best product. Get in front of as many people as possible. Get the press, influencers, partners, and key users excited and the network will be built from these important nodes down to the individual users.
For network products, this is often a trap. It’s exactly the wrong way to build a network because a wide launch creates many many weak networks that arent stable on their own. when companies don’t understand these nuances it leads to disaster.
Anti-Network Effects Hit The Google Plus Launch
June 2011 at the Web 2.0 Summit. Google VP describes the future of networking and launches Google Plus. They lead with aggressive upsells from their core product the Google.com homepage linked to google plus. It was also integrated widely within Youtube, Photos, and the rest of the product ecosystem. Within months, they had brought on 90 million users. Unfortunately, this consisted of many weak networks that weren’t engaged due to new users showing up to try out the product as they read about it in the press rather than hearing from their friends. The high churn was covered up by incredible amounts of traffic that the rest of Google generated. Even though it wasn’t working the numbers keep going up.
When unengaged users interact with a network product that hasn’t yet gelled into a stable atomic network. They don’t end up pulling other users into the product. Google Plus was a ghost town compared to Facebook which at the time was gearing up for IPO. Google Plus had on average 3 minutes between Sep. — Jan. a month opposed to Facebook which had 6 hours in the same period. The Fate of Google Plus was sealed in their GTM. By launching big and not focusing on small atomic networks they fell victim to vanity metrics.
The Problem with the Big Bang — The problem is twofold.
First It’s built on broadcast channels. The weakness of media coverage, conferences, or advertising. It might generate a large spike of users when successful. It is necessarily untargeted. Instead, you can get users across many networks which will turn users off if the network isn’t built.
The second issue is that it takes time for a product to have the right features. But also to have enough built out to have viral growth such as sharing, invites, and collaboration.
You’d rather have a smaller network that is denser and more engaged than a larger number of networks that aren’t there. It’s better to ignore the top line aggregate numbers when a network product requires other people to be useful. Instead, the quality of the traction can only be seen when you are all the way into an individual user within the network. Does a new person on the network see value based on how many other people are on it? Ignore the spike you may see in a new product in its first days. These are vanity metrics. They may make you feel good but it doesn't matter if you have 100 million users if they're churning out at a high rate due to the lack of other users engaging.
When you build a network from the bottom up they’re more likely to be more densely connected. Thus healthier and more engaged. A new product is often incubated within a sub-community. Whether it’s a college campus, SF techies, gamers, or freelancers.
Chapter 33- Competing for the Hard Side — Uber
If network effects are so powerful, why are the larger networks so vulnerable?
Uber’s competitive battles around the world offer a clue. Uber’s competitive tacts were fierce and interdisciplinary. In the early years of Lyft and Uber, Uber had a strategy around flipping people from Lyft to Uber. Uber knew drivers would be going to Lyfts headquarters for customer support. They arranged trucks with mobile billboards to tell Lyft drivers to “Shave the stash”.
Finding the competitive levers
When there’s a battle between two networks. There are competitive levers that shift users from one side to the other. What are they? In Uber’s case, it would involve financial incentives, paying up for more sign-ups, also product improvements that drove acquisition, engagement, and economic forces. Drawing in more diversity through product improvements is straightforward. The better the experience of picking up riders and taking them to their destination the more the app would be used. Uber focused its efforts on targeted bonuses for drivers. They were targeted at flipping Lyft drivers or Dual Drivers. They were offered large compelling bonuses that compelled them to stick to Uber.
Chapter 34 — Bundling- Microsoft
Bigger networks are fearsome. They can quickly solve the cold start problem for new products. This is often called bundling. Multiple products for one price.
The importance of a killer product
Microsoft Office is an example of bundling. Word and Excel were originally made for DOS and were keyboard only and textbase. When it came to word processing and spreadsheets, Microsoft was losing. For Microsoft’s productivity applications, the break came when the world changed from text-based DOS applications to graphical UI in the mid-1980s. As the industry shifted this created an opening as every application had to be rewritten to support the new paradigm of toolbars icons and dropdown menus etc. Much effort was put into Microsoft office suite’s tools to work with each other. An excel chart can be embedded in a Microsoft word document which made the combination of the other products more powerful.
Competing with the network, not just features
Facebook does this at scale. Take Instagram as an example, in the early days the core product tapped into the Facebook network by making it easy to share photos from one product to the other. This creates a viral loop that drives new users and engagement too. Being able to sign up to Facebook using your Instagram account also increases the conversion rate which creates frictionless experiences while simultaneously setting up integrations for later experiences. Instagram is a great example of bundling done well.
Locking in the Hard Side
Microsoft had a huge effort to attract and retain developers. Microsofts dev tooling launched on its earliest operating systems. Visual basic was a key part of the flywheel for windows. For every copy of Visual Basic, we sell there are 10 copies that go with it. Microsoft ensured that new releases of DOS and windows wouldn’t break the code past developers wrote making it possible that every new device can be backward compatible. By supporting old legacy applications they were able to use its ecosystem of developers.
