29 Apr 2019

Uber's IPO and Local Network Effects

Uber Ridesharing Revenue

My writing has been inconsistent (at best) lately, but Uber's IPO seems as good a time as any to come out of hibernation. I have written about Uber's business model numerous times in the past and more often than not, I have defended its financial performance. My argument was that Uber's losses were caused by large investments into logistics infrastructure (largely fixed) that would then result in long-term revenue growth (and overtake costs). Uber seems to be using that exact same argument to position itself to investors. Uber's IPO prospectus finally shed some more light on its progress, but I was concerned by what I saw. It showed the scale of investments, but it also showed that Uber is no longer a high growth company.

The chart above shows Uber's adjusted net revenue from ridesharing (which deducts excess driver incentives and revenue from divested units). Ridesharing gross bookings are now growing by single digits sequentially, and net revenue has stopped growing entirely. As we can see below, the picture for Uber Eats isn't dramatically different.

Uber Eats Revenue

Asymptotic Network Effects

The fact that gross bookings are growing faster that net revenue suggests this is being caused by excess spend on driver and restaurant incentives, i.e. supply acquisition. But shouldn't Uber's scale and the network effect between users and suppliers lead to lower acquisition costs over time? That should lead to net revenue growth outstripping growth in gross bookings. What's going on?

Uber's network effects have always been more complicated than those for other marketplaces like Amazon. When we were categorizing network effect-based businesses on Platform Hunt, we were forced to carve on-demand services into a separate category from marketplaces. Since on-demand services typically focused on a single, specialized supplier (drivers or restaurants), the addition of suppliers increased the value of the product only to a certain point. The team over at NfX have done an even better job explaining this and have called this pattern asymptotic network effects. Notably, this affects both the ridesharing and food delivery businesses. For example, when wait time for a ride hits 4 minutes, user experience becomes "good enough" and further supply acquisition does not meaningfully improve it. Similarly, once a food delivery service offers enough choices of available cuisines, it becomes "god enough". In other words, the defensibility provided by asymptotic network effects only lasts until the service (or competitors) achieve critical mass. This makes these network effects weaker, and hence the market more competitive. 

That said, this only explains why Uber still faces competition once they achieve scale within a market. But Uber’s bigger problems are related to the challenges it has faced in expanding internationally, especially when compared to companies like Airbnb and Amazon. The explanation here is also linked to the nature of its network effects.

Local Network Effects and Regional Fragmentation

Airbnb and Amazon's Network Effects

Unlike Airbnb and Amazon, Uber's network effects exist purely within a tight geographical radius (within a few miles). Both Amazon and Airbnb could scale up a supply network in one location, leverage that to grow demand in another which would then attract more supply in that location and so on. However, Uber needs to scale up a supply network in one location and then start from scratch all over again at the next one. In other words, when Uber expands into a new market, its only advantage is capital. This is especially troublesome when first movers in local markets (e.g. Grab in Southeast Asia, Didi in China, Yandex in Russia, Ola in India etc.), have already established local supply networks, which makes competition even more of an uphill climb. 

Uber's Local Network Effects

Notably, the pattern of local network effects isn't limited to the ridesharing business. It also affects food delivery, grocery delivery, classifieds, C2C marketplaces or any service that needs to be delivered locally (and in-person). One common theme among these industries is that tend to be regionally fragmented. Apart from Uber, can you think of a single, standalone and global player in ridesharing, food delivery or classifieds? The very nature of local network effects makes it nearly impossible (or in Uber's case, prohibitively expensive) for these businesses to expand to multiple markets.

Uber has been attempting to divest local units and find other avenues of growth to make up for their network effect handicap. Micromobility is one that Uber seems particularly bullish about. The fact that nearly 50% of vehicle trips are under 3 miles clearly shows that there is latent demand for scooter and bike rental services. But the complete lack of network effects strains pricing power and unit economics even further. At best, I can see these services being used as loss leaders to ensure users do not stay away from Uber's transportation ecosystem. Apart from that, I fear scaling back its global ambitions may be the only way for Uber to achieve sustainability.