The Cold Start Problem 7/7
29/ Wimdu vs Airbnb
Wimdu did an impressive job copying Airbnb but didn’t create the expectations gap – the positive difference between the user’s experience and their initial expectations; the wider this gap is – the higher word of mouth distribution there is. This often is driven by curating the hard side, not simply growing the numbers for the sake of the numbers.
30/ Vicious Cycle, Virtuous Cycle
The moat for software products is the complexity around time, capital and capabilities to replicate the product and the network. Copying features and entire products is not either hard or expensive.
Once the company reaches a tipping point in a certain market/network, their competitors have to replicate the same tipping point, starting from a disadvantaged position. Competitors’ offers need to deliver better value in order to stand out, otherwise why would a user bother?
The book makes a correct distinction between the geography of demand and the strength of associated moats. Uber: local supply, local demand, weak moat. Airbnb: local supply, global demand, strong moat. Slack: used mostly within the company; Zoom: very often used between companies.
Networked products are more likely to fight the “winner takes all” battles, with losers being wiped off the market. It doesn’t make sense for users in a certain circle (company, family, club, etc.) to use several competing products to solve a single problem.
An accumulating advantage occurs when one product adds networks faster than its competitor(s). Success breeds success, and cost structures reflect that. And it’s not about the size per se, it’s about the quality of networks. All large networks have weak parts that are lacking attention for whatever reason; emerging players targeting these parts can disrupt the larger businesses as a result.
Network effects themselves are not a moat, because competitors in the category also have them. In mature markets competition looks like a zero-sum game where the same network effects help the winner and hurt the loser. Losers will be happy to shrink to a stable user base instead of completely dying.
In the market competition usually occurs between companies at different stages of their growth: the problems an incumbent is facing are very much different from the problems of an upstart. Shift to profitability cripples the entrepreneurial culture.
31/ Cherry Picking
As mentioned before (the book is getting repetitive at this point), networks of networks are not universally dense. They’re prone to unbundling when upstarts target certain segments and service both sides of the marketplace better. The incumbent (think Craigslist) can’t counter this, rather focusing on features serving a diverse group of segments, and such unbundling resembles death by a thousand cuts.
Upstarts can build denser networks than incumbents can (regardless of their size) – thanks to the focus. Using other networks as locomotives (Airbnb hosts publishing their listings on Craigslist with a link to Airbnb) used to be a powerful growth hack (mostly impossible now).
Once a network is lost to a competitor – it’s almost impossible to bring it back as well as to raise external financing to fund a comeback. Loss of market share is a pill that’s very hard to swallow.
Dependence on the donor platform can be dangerous if most of the company’s growth depends on the donor’s good will. If a product becomes too important and too successful – it often makes sense for the incumbent to copy its functionality rather than buying it.
32/ Big Bang Failures
There are two types of product launches: top-down and bottom-up. Top-down goes with fanfare and huge media attention; bottom-up is suitable for partisan-style launch, building dense networks and scaling them up. Using the top-down approach for networked products achieves only one thing: a bunch of weak networks that can’t stand on their own.
Launching networked products top-down requires a traffic donor to overwhelm users with the torrent of touch points between the core product and the new product. Unfortunately, a more likely outcome is not increased engagement, but rather skyrocketing annoyance and user blindness.
Top-down launch (aka Big Bang) relies on broadcast channels, which are by definition poorly targeted. Hence, this approach (even if successful) attracts users from all walks of life, not necessarily the desired ones.
Network growth numbers are very confusing if the launch is top-down. The necessary tools and metrics haven’t been built yet, so it’s lots of noise and little signal. Vanity metrics cloud the big picture.
Big companies focus on big markets and consider small ones as not worth focussed attention. This creates opportunities for smaller players to dominate these smaller markets and then attempt to do more serious heavy lifting alongside the bigger guys.
It’s very unlikely that unless an initial set of networks is successful that scaling up is a good idea. At the same time, it’s also unlikely to foresee the trajectory of the platform’s growth past the initial set of networks, too. The investment case is very unclear at this stage, and the typical “total addressable market” analysis is misleading as the end product might deviate a lot from what’s being presented at the moment.
Big Bang launches are almost inevitable because new products are expected to be very successful (and jump the shark – the Cold Start Problem – just like that) in the limited top management’s time span – usually a few Board meeting cycles. Smaller companies are blessed not to have this sort of external expectations.
33/ Competing Over the Hard Side
MK: I remember the early days of Aviasales when we were just one travel startup without VC funding fighting an uphill battle against a bunch of loud, noisy and ambitious startups. Our goals were (in this particular order) to survive, show positive user acquisition and engagement dynamics, and make life very hard for competitors. This behaviour is common among the companies taking competition seriously.
Building a better product is one important lever for bringing and retaining the hard side, but nothing beats good old cash (the monetary incentives). When the hard side is limited, and the competition for it becomes a zero-sum game, even losing money on subsidies to a member means this member is not going to create value for competitors. It’s a very expensive game that can’t be played forever, but still.
For social networks or video platforms making content creators shift would require additional economic benefits and extra distribution; for B2B it’s the feature set and custom pricing.
Competition monitoring is essential (even via reverse-engineering APIs or making test orders to figure out a sequential order number). It’s important not to scare oneself into believing that any misfortune is an outcome of competitive tension. Nonetheless, having a good idea what competitive products the hard side is using helps prioritise many activities.
Competitive moves are only meaningful if they deliver the value to both sides; adding new features that people don’t want will be a waste of focus and effort. When market shares are split 50/50, competition turns into trench warfare with uncertain outcomes.
Bundling is an approach of offering multiple products (each having distinct value to users) at one price or making a super app where the features complement each other logically and via the loyalty program. Regardless, if the feature set is not up to the user expectations, bundling becomes more of a liability than the benefit.
Nowadays bundling is not about embedding Excel spreadsheets into Word documents but about generating clicks in one app leading to another app of the same or a friendly software vendor. (It’s not to say that the Cold Start Problem will be immediately or subsequently solved, but all things being equal, a bunch of clicks is better than no clicks.) At the very least, increasing the number of touch points between the bundled products is likely to lead to success.
Bundling more and more apps into a super app inevitably leads to clutter and the loss of appeal to users – with the exception of certain markets where users expect precisely the cluttered interfaces.