January 15, 2009

Multichannel Forensics For Social Networks

99% of the time, the Multichannel Forensics examples I offer are related to what you might loosely call "retail" --- physical stores, online businesses, catalog brands.

When we deal with retail, we deal with long periods of time. We measure repurchase rates over the course of a year, or in some brands, four or five years.

In subscription-based businesses, we often convert churn rates into the Multichannel Forensics framework --- a 4% monthly churn rate yields a 60% percent annual retention rate, while a 7% monthly churn rate generates a 40% annual retention rate.

But what about something like Facebook, an application where half of users visit every single day?

Social Networks are best measured on a daily or weekly basis. All the laws of Multichannel Forensics still apply, but the timeframe is seriously compressed.

This becomes important for the Social Network, because you can quickly identify the amount of time that passes before the "user" is about to defect --- you can build "KPIs" around a "time to defection" metric generated by the Multichannel Forensics framework.

Also important is the type of activity the user participates in --- you want to measure which activities are in "Equilibrium Mode" or "Transfer Mode" with each other. For instance, are a ton of folks in Friendfeed following Robert Scoble, only to slowly defect to another social media evangelist? If this happens, you'll notice because Robert Scoble is a "micro-channel" that is transferring users to a new individual. You'll be able to react to the fluid dynamics within your social network.

So, what are the key takeaways?
  1. For most social networks, reduce the timeframe to a daily (Facebook, Twitter) timeframe, or a weekly timeframe, measuring retention and user acquisition over short intervals.
  2. The definition of a "channel" changes. Micro-channels in a social networking framework include various activities (e-mail, search), widgets, or even key individuals (Robert Scoble) that attract or cultivate users.
  3. These activities can be simulated over time, to understand the long-term trajectory of your social network.

2 comments:

  1. Can you elaborate on the "time to defection" for a social network? Why is this important to identify for users? How would you go about calculating this?

    Are there any case studies/references for such analysis on communities?

    ReplyDelete
  2. I'm not aware of any case studies related to measuring user defection in a social network.

    That being said, it is the exact same concept as measuring the merchandise ecosystem at a retailer. At Nordstrom, Shoes are the "glue" that keep customers coming back ... if shoes aren't sold, the apparel ecosystem falls apart.

    On Twitter, if your favorite users are no longer using the service, the system falls apart.

    So if you are Twitter, you simply create a massive "spreadsheet", if you will ... each row is a user (me), each column is a super-user (Robert Scoble). You use the Multichannel Forensics framework to calculate what happens when Robert Scoble disappears.

    It is a massive report ... but very easy and straightforward using the Multichannel Forensics framework.

    ReplyDelete

Note: Only a member of this blog may post a comment.

Items That Appear In Multi-Item Orders

In a typical Life Stage Analysis within a Merchandise Dynamics project, it is common to see exaggerated trends when comparing first-time buy...