Steve Jurvetson

Managing Director of Draper Fisher Jurvetson

Steve began his talk by pointing out that viral marketing is different from regular word of mouth in that it has an involuntary component. Ideally, you share it even if you hate it. As an example, he offered Hotmail viral strategy. Hotmail apparently began as a database company, and their mail client was supposed to be a proof of concept, a demo. They decided to add a footer to every email sent offering users to “Get your free private email at hotmail.com”.

The growth was exponential – at a certain point they were adding 250k new users per day. Students were the best vectors, since they had a need to communicate over long distances when they were away from home. After only 4 months, they were the largest email provider in Sweden without any advertising there whatsoever. There were 4 Hotmail accounts for every modem in India!

Two aspects of the Hotmail technique were particularly clever. First of all, the footer was an implied endorsement by a friend: the email must have been sent by someone who used the service. Second, it was a proof point for the new product category of free email: receiving the email actually meant that the service work, which reduced perceived risk.

The net result was a spread so quick that no other company could catch up for the next few years. The nature of exponentials meant that a company starting a few months later might grow at the same rate but still see the user gap increasingly widening – which is why Microsoft ended up buying them. There was simply no way to compete. Interestingly, you can see the same exact trends in other products: ICQ, Kazaa, and Skype.

Steve also spoke about the two laws that govern networks: Metcalfe’s Law (the value of a network is proportional to the number of connections in the network – think Hotmail, Skype) and the more powerful Reed’s Law (the value is proportional to the number of groups in a network – think MySpace, Flickr). In an attempt to model the effect of viral marketing, he proposed a simple formula of:

USERS = (1 + fanout) ^ time

which he then expanded into:

USERS = [(1+fanout*conversion)*retention]^(frequency*time)

Counteracting this effect are the decay effects of novelty loss and saturation, which slow the process. Of course these sorts of formulas are oversimplifications, but they can help us think of the different factors of a good viral marketing campaign.

In closing, Steve gave a number of recommendations for a product hoping to spread virally:

a) internationalize quickly so you can spread globally

b) ensure platform homogeneity, so users can participate regardless of their device

c) tie your product to social activities, preferably targetting weak social ties (acquaintances rather than friends)

d) lower the friction to adoption (for example by making the signup process easy… or even non-existent)

e) create retention hooks that make people want to come back

f) leave visible traces

g) add communication

h) focus on involved non-purchasers: the people who participate but not yet fully