With 90 or so Swedish Internet entrepreneurs about to meet for 24 hours to create new web services, I'm going to write a note on why I think invite-only is a bad marketing strategy for web services.
The semi-public, invite-only service is an interesting child of Web 2.0. The classic example obviously being Gmail. A more current example would be the free Spotify accounts.
With the success of Gmail, one might be fooled into thinking that the buzz created by the invite-only aspect was a major reason for the strong word-of-mouth. I know I certainly thought so when being part of two community projects at Universum four years ago. (If the results of those projects are anything to go by, I was wrong. Buzz is created by amazing products.)
Part of the invite-only thinking is summarized in the fourth question Jaiku-founder Jyri Engeström says creators of social web services should be asking themselves: "What is the gift in the invitation?"
Creating a reason for people to invite friends is a good thing. If it is possible to make the invitation into a gift it is even better. But fake scarcity rarely creates a valuable gift. As a creator, one has to come up with a better way to make a gift out of the invitation than to restrict access. (An exception would obviously be made for sites like Asmallworld where not being open to everyone is a core part of the service and not a marketing gimmick.)
Andrew Chen's post What's your viral loop? Understanding the engine of adoption explains how to design in order to make users invite other users. Read it and think about how invite-only affects the viral funnel. Conversion rates goes down across the board, which is exactly what one wants to avoid.
Jyri's often-used example of an invitation-as-a-gift is the $10 gift Paypal gave new members when they they accepted an invitation from a friend. That is a gift that fuels growth, not restricts it. Even if giving $10 away might be too much for many services, it is a good benchmark when thinking about invitations-as-gifts for marketing purposes.
The two reasons one should ever use an invite-only system are 1) to slowly let people on to the service to find bugs and issues and 2) to control growth ("make sure we don't grow too fast") for technical/operational reasons. Both are fair reasons, but from a marketing perspective they should be recognized for what they bring: slower growth than a site open for everyone would have.