Five Types of Network Effect

I’m currently reading a book called Blitzscaling by Reid Hoffman and Chris Yeh. “Blitzscaling” is a technique they have observed over several years and applied themselves. They define it as prioritising speed over efficiency in the face of uncertainty. So not only are you jumping off a cliff and building a plane on the way down – the metaphor many use for founding a startup – you’re also attaching jet-powered rockets at the same time. You will probably fail.

But if, against all odds, you don’t fail, you may have cornered a new market before someone can beat you to it. This means you will be able to begin all your motivational speeches and TED talks with, “Everyone said we were crazy. But we believed in our idea and overcame the haters! [Pregnant pause awaiting rapturous applause.]”

Of secondary importance: you’ll also potentially have established a business position that no other company can attack. Whereas in previous times, first-mover advantage was of arguable merit, today, it can be a first class ticket to a land of unicorns.

This is due to network effects.

Network Effects

The authors’ definition of a network effect is as follows:

A product or service is subject to positive network effects when increased usage [sic] by any user increases the value of the product or service for other users.

Hoffman, R, Yeh, Chris (2018) “Blitzscaling: The Lightning-Fast Path to Building Massively Valuable Companies”, p. 66

This, they say, results in superlinear growth – i.e. growth that exceeds a simple linear trajectory – and causes lock-in effects. Think of the telephone network as an early example: it’s not much use having one if no-one else does. If a friend gets one, then its value increases infinitely. For every user that joins the network, you can contact more people, thus increasing the value.

If a rival technology comes along, its technical merits may be far superior but it would be a hard sell to get people from the original network to migrate because no-one is there. And until anyone is there, no-one will go there. Hence the lock-in effect available to the first mover.

With the rise of the internet, the opportunities for pursuing these lock-in effects are increasing. But if you want to tap into them, you have to do so before the competition and grow fast enough that, by the time they find out about it, they can’t beat you to the necessary scale.

Five types of network effect

So, as promised, here are the five categories the authors separate them into. They cite Arun Sundararajan as the originator of the categorisation. I’d like to thank Arun before shamelessly ripping off his work (in my own words) and having the cheek to criticise it:

Direct Network Effects

This applies to social networks, the telephone, fax machines, and messaging apps such as WhatsApp. The crucial thing here is that each user adds to the value of the network simply by being there, and they seek the same type of value as everyone else. It’s not eBay where you need to match up sellers and buyers, who are seeking different benefits.

Indirect Network Effects

In this case, we see a virtuous circle: the more people use the platform, the more incentive there is to create complimentary goods such as apps or other digital products. The more digital goods are available, the more attractive the platform becomes for users. Examples would be Android or iOS.

Two-sided Network Effects

This is eBay, AirBnB and so on – where an increase in one category of users increases the value to the other. You have that awkward balancing act of getting enough people listing their home for sale, but in order to get them to do so they need to be convinced that there is a big enough audience. And who would search for homes on a property sales site where there aren’t any listings? But who would search for them if they’re gone in twenty seconds because the demand outstrips supply?

I have to say that I’m not sure there is a great difference between this and “indirect network effects” – the same issue applies to iOS and Android. The only difference is that the platform developer also adds their own apps, and perhaps the problem of supply and demand is cushioned somewhat by the fact that digital goods can be used by an infinite number of people at once unlike, say, an AirBnB property.

Local Network Effects

A small subset of users increase the value for specific other users. The example here is where users of telecom networks were allowed to nominate “friends”, to whom their calls were free. In this case, the value was only redeemable with a certain number of specific people but presumably incentivised groups to choose the same network.

Compatibility and Standards

The example of this is MS Word: because MS Office products have become standard over the years, people just expect you to have that product. Not necessarily because of technological superiority, but because it’s a standard. I know from experience that many advertising agencies use Apple products, which is a reason I have them.

One person’s lock-in is another’s monopoly

Lock-in effects are great for those who achieve them, but they have the potential to be anti-competitive, and ultimately lead to monopolies. From the point of view of a successful business, this is of little interest. However, the very idea of a lock-in is that you shield yourself from competition. It also requires a hell of a lot of investment to get there in the first place, which is why Blitzscaling is required.

Uber is cited quite often in the book: not yet having turned a profit, it is already starting to scale back the financial incentives intended to increase its user-base and establish a defensible position. From the European perspective this just looks like a targeted attack on established industries not by means of innovation or user-friendliness but  market distortion.

An important take-away from the book so far is to be realistic about the chances of businesses based on two-sided business models: you need to get very big, very fast to succeed.