First, let’s consider a social network. In the old days before hyper-personalised news feeds, we imagined social networks as websites enabling you to see what your friends and family have been up to. In this sense, it’s hard to imagine why you would use this utility unless your friends were active users too. Moreover, as more and more friends join, the website becomes more and more engaging, just like how a party (remember those?) becomes more stimulating as more of your friends walk through the door.
Next, take a search engine. The suggested websites that pop up after you type in a query might appeal to some users around the world, but might not be particularly appealing to people who live in England, in the 30-40 age bracket and recently viewed car selling websites. As more and more users in this demographic engage with the search engine, the designers of the algorithm can begin to increase the relevance of the search results to the user.
Now consider an online video streaming website. This website has a lot of movies, more than you could search through entirely. But the website knows what movies you watched until completion in the last year, and also which movies you quit before the end. The website also knows other people who have similar watching patterns as you do. Next time you log onto the website, you find two or three well-placed recommended movies, that just seem right and kill the need for a 10-minute scroll through the catalogue.
What runs as a common theme through these three examples is that we can guess the companies who provide these services: there aren’t many. Social networks, search engines, video streaming, e-commerce, instant messaging, computer operating systems, instant payment, etc., all seem to be dominated by one, two or three companies. This has led to gargantuan stock market valuations for a small string of California tech companies, such that the FAANGs (Facebook, Amazon, Apple, Netflix, Google) together make up around 15% of S&P 500: a group of the top 500 listed companies in the United States.
Compare this to car manufacturing, and the concept of the network effect becomes more intriguing. Car makers are competing with other car makers to make high quality, safe and attractive automobiles. Once they sell one of their cars to a buyer, it is gone: the product they designed and manufactured has been converted into cash on their balance sheet, and they must make a new car entirely before they can sell another. Furthermore, the buyer will use the car after purchase, and will find that the quality of the car decreases with time and use. According to Hertz, a car rental service, the average lifespan of a car is between 100,000 and 150,000 miles. So the buyer might eventually be forced to discard the product she bought, and buy another car down the line.
A network-effect-driven product like a search engine, however, runs differently. My use of the web tool improves its capabilities for you. The product hasn’t gone anywhere; I can use it again and again, and it will most likely improve with every use. This will probably still be true in 10 years’ time. And the product in 10 years’ time will probably be better too.
What does this mean for future companies?
What we must ask is: is this network effect potentially problematic? While it enables these services to run very cheaply, if not free at the point of use, it also creates barriers for smaller competitors who want to grow. Making an e-commerce site that can rival Amazon is tricky: the network effect means it benefits from the large number of retailers listed on the site, shoppers regularly visiting it and historical data to optimise the logistics and website functionality. A new entrant would need to convince a large swathe of sellers and buyers to start selling and buying on their online platform instead: a hard ask when the retailers want to sell where there are many shoppers, and shoppers want to shop where there are many retailers.
Disrupting the disrupters is not trivial. Start-ups will need to rethink how we use data and build tools that are less data dependent. It is nearly impossible to match Google for their web search data, or Facebook on their personal data. Instead, a competitor to firms which maintain strong barriers through the network effect needs to create a product or service that relies on an ingenious invention, and not an enormous library.
Amongst other reasons, the network effect has empowered huge monopolisation of the data-driven industries, as well as others. It creates large barriers to entry for potential competitors. But it also means that the services we use are cheap. If we care about maintaining competition amongst our corporations, we will need to begin designing tools that won't require the huge data sets that the large tech companies hold for themselves - because competing with the giants on their own turf is not going to suffice.