Wednesday, October 20, 2010

facebook and twitter analytics

I wonder why neither Facebook nor Twitter is sharing analytics data for a Facebook Page or Twitter Page - they have become destinations in themselves and brands would surely be willing to pay to see user behavior on these pages.

One reason to not share this data could be that brands would start asking for more freedom with page layout and content. A way around this would be define and provide new "social media metrics" centered around the structure of the Facebook or Twitter page rather than providing either basic clickstream data or standard web analytics metrics.

Some data I would like to see:

  1. Max/Average reach of my tweets - number of hops in my network the tweet percolates to.

  2. Content (tweets or facebook posts) that leads to people following me or liking my page.

  3. Content that gets re-tweeted, replied to, liked or re-shared the most.

  4. What %-age and section of my network interacts with me the most. Which of my immediate network connects me to the rest of them? How does this change over time?

  5. Interest churn period: average length of time people follow me for.

  6. Interest churn: ratio of number of new followers to number of lost followers in a given period.

  7. Conversion data for the new funnel where you are trying to convert: people outside your network to followers/fans to clicking on your links to "liking" your content to replying/commenting to re-tweeting/re-sharing.

  8. Distribution of number of people followed by those who follow you - the ideal would be a bell curve - users following too few people or too many people are either not going to be listening at all or listening to too much noise.
Next up might be allowing A/B testing on FB/TW pages and updates. But that seems a little distant at the moment. :)

Sunday, October 10, 2010

where you are

Some food for thought about ad targeting:


I'm vising MissionLocal.org, which is a local website about events/food/news in the Mission district in San Francisco, from Seattle. Groupon shows me a Seattle ad.


In this case, should the ad have been targeted based on where I'm coming from, or based on the page context, given the strong location signal present in the context?

Targeting systems do wrong in not considering geo-location as a signal derived from different sources of information present across behavior and context (and device GPS), but only using it as a simple IP-based filter.