A reason to have a freemium business model, which I didn't clearly point out in my last post, is to combine the network effects of a free service with the revenue-generation of a paid service.
Network effects are obviously core to social and communication services. Building a new service while being paid-only is difficult, as the initial utility is close to zero for users. Having free users make the service better for paid users.
Skype is an example of a freemium service that the majority of users doesn't pay for, while some users pay for premium features like international calling and group video chat.
That's not to say it is impossible to build a primarily paid-for communications service. WhatsApp has done just that. The reason likely being that its users see WhatsApp as a replacement to text messaging (which is more expensive than WhatsApp).
Generating advertising revenue in a communications online environment is tricky, even at massive scale (the history of ICQ, MSN Messenger and webmail). Successfully implemented premium features should generate high revenue per paying user, and with a decent conversion rate the average revenue per user (free and paying) should be quite good.
More posts on freemium strategy: