Social Engagement Index – An experiment in collecting social data & making sense of it

Background

Problem : To understand, evaluate and respond to social engagement activities by rival organizations to better engage with customer / potential customer base.

Who would want this? : All organizations with active social media plan.

How would it work? : The system would analyze open data in real-time and present info graphics to the marketing team.

Why would anyone want this? : To keep effective check measures against your competition and to actively improve your social base, and eventually your customer base.

 

Experiment

Phase 1 : Identify few interesting data sources

This was fairly easy because, almost all media companies have done intricate analysis on top Facebook pages. I have selected top few interesting pages and created simulated sample data to ensure that the experiment is headed in the right direction, for guidance.

Simulated Data

Simulated Data to understand and evaluate progress of the experiment

Phase 2 : Data collection

This is a more complex task, to collect data from a non homogenous source and make data organized for analysis.

Phase 3 : Understanding the data

Experimented with a whole variety of weighted values and sampling methods to make data more tangible for use, and the simplest was to use, weighted means of individual engagement levels / parameters.

Result

Phase 1 : Identifying what to collect

Cron Job result

Cron Job result

Created a CRON Job to crawl through pages, identify relevant data and insert them to a mysql db.

Phase 2 : The data

Data collected and organized into tables

Data collected and organized into tables

The data is approximately 1000 rows of data per page per day.

Phase 3 : Making sense

Created a simple sensible equation to digest the vast amount of information into few tangible data points and plotted them on a graph. (Table & Graph below )

Real pages, and their social engagement index plotted over time. Page Red vs Page Blue ( to keep identity anonymous )

Real pages, and their social engagement index plotted over time. Page Red vs Page Blue ( to keep identity anonymous )

Real data table showing the values of social engagement between two pages.

Real data table showing the values of social engagement between two pages.

Conclusions:

Real data gives us more surprises than simulated data ;)

This gives us evidence that this type of analysis can give us detectable information from the vast amount of data and possibly use this for tracking competition and to use this information to forward ones social engagement plan.

Mandatory Disclaimer

  • Data has been made anonymous.
  • This is a preliminary investigative experiment to look into the possibilities in this arena
  • Real data points are shown
  • These views are of the author alone, and not of Adobe Systems.

Comments are closed.