Some musings on the rise of recommendation engines within different end user devices and websites. We all know that these engines or collaborative filtering applications are spreading. Examples are: TiVo boxes/PVRs and mobile phones with Crunkie-like functionality (mobile social networking and its next versions). The question is whether the value of these automated recommendations in TiVos or Amazon.com have equal value relative to the personalized recommendations from personal friends within or without a social software environment, both within the mobile (like Crunkie) as well as fixed Internet. I believe the automated recommendations have the same drawbacks as CRM systems and these are mainly due to sudden lifestyle, interest, value system, career and family situation changes. This is about dealing with the changing personal contexts of recommendations, something Amazon.com is not dealing with effectively in my opinion. However, our (close) friends know more about our personal and evolving context in a much broader sense and as a result are able to suggest more relevant recommendations to us.
My ideas or assumptions can be structured as follows:
- Personal recommendations from friends are more relevant than automated recommendations from software engines.
- Personal recommendations in real life (word of mouth) have more value for and impact on the receiver due to the richer and more detailed experience surrounding the recommended item.
- Personal recommendations from mobile social software have more value to the receiver relative to online social software as a result of the lower threshold for acting upon the recommendation in the physical space (seeing a recommended movie for example).
- Personal recommendations from close friends are better than personal recommendations from more far-off friends or contacts. Closeness is related to the intensity/frequency of online and offline contact moments as modeled by online social software analytics vendors like Visible Path and Spoke Software. I guess this functionality will soon be integrated within Crunkie-like applications in the mobile space.
- Role of viral marketing (SMS, e-mail, IM) and recommendations. Due to the rise of different digital media, there are more recommendations flowing around than ever before. Viral marketing in a sense presumes that the sender knows the receiver on a certain personal level. As such, viral marketing supports more effective recommendations relative to automated recommendation engines.
- Recommendations alone will not do the trick the coming years. We can already witness the rise of new sharing/lending/transaction models in MediaChest. In short, recommendations will move from informational to even more transactional models. Eg., a recommended item is sent with a product description and with different options for getting it (lending within MediaChest, auctioning within eBay, fixed buying at Amazon.com etc.) and even experiencing it directly in the mobile/physical space.
- Role of The Long Tail and recommendations. Works in two ways. My friends will have more Long Tail items over time and thus will provide me with more Long Tail recommendations over time. In a sense, personal recommendations reinforce the concept of The Long Tail. On top of that, automated recommendation engines also have a place for less popular content these days as can be seen in this post by Yme Bosma.
Putting it all together I believe recommendations within the mobile/physical space will be most relevant the upcoming years. The increasing cluttering of recommendations via offline and online/mobile contacts will lead to a situation in which the immediate experiental value of recommendations will be key. This will be a Crunkie-like application on your GPS-enabled mobile phone combining analytical intelligence from Visible Path for your social network and integrating MediaChest lending fucntionality for particular cases.
Yme Bosma has some interesting thoughts on the value of recommendations by our friends and their limitations. I disagree with him as shown above.
Feedback on this one is very welcome.....thanks.