KEMBAR78
Buisness Intelligence and Web Analytics | PPT
Business intelligence, web analytics and FOSS Using the Mail & Guardian Online and Amatomu.com as examples
Background 0.5 million readers, 100 000 visits  a day, various sources of BI, vendor lock-in, new territories that lack BI/web analytics
Background Publishing business models dependent on analytics, changes in the Web: Web 2, social media and Semantic Web
Reality The business of a media company is to create sell-able statistics, whereas the public mandate is to inform and encourage democratic governance
Rethinking Watershed moment, great risk and great opportunity
Catalyst Rebuilding our publishing system from the ground up + experience with Amatomu.com
More goals than Bafana - A comprehensive series of dashboards - Real-time reports of user activity - Rich visualisation - A product responsive to key metrics
Specific Challenges Internally, data is not centralised, standardised or aligned to our knowledge management goals
Specific Challenges   Web 2,  server logging and auditing
 
Amatomu Growing blogging community, social media
Amatomu Inspired by a lack of information and the potential to facilitate growth of the SA blogosphere
Amatomu Aggregates blog content, gathers analytics and presents personalised statistics, global statistics and rankings
Amatomu Collect statistics for 5.6 million PI per month, 2.5k active bloggers producing 1 05 million db transactions per month
Lessons learned Real-time analytics systems are difficult to scale, data increases in value as it ages, FOSS can do the job
FOSS architecture LAMP stack
The great debate Packaged software or create custom code?
Packaged software Easy to evaluate, low risk, ongoing development and active community, medium reward, hard to tailor to specific business requirements
Custom code Hard to evaluate, high risk, high reward because closely aligned to business processes
BI Packages 2 biggest are Pentaho and JasperSoft, both very powerful solutions for out-the-box BI, both offer FOSS servers with paid support
The thorny path We have chosen to develop custom code because we’re confident we can do it successfully and no packaged software does what we want it to
External factors Systems need to be overhauled every 2 years and rebuilt every 5 years
Guiding principles Data must be reusable, interoperability via web services and XML, scalability via the cloud  - Amazon EC2, SimpleDB, S3 storage
Insurance policy Development using frameworks for PHP, Ajax and CSS: CodeIgniter, JQuery/ExtJS, Google Blueprint
Insurance policy Protects us against change in the development team and common mistakes
Emphasis Rich media graphing, real-time visualisation, enabling knowledge creation through discovery
Inspiration Edward R Tufte Digg Labs
Helpful tools FusionCharts  - visually attractive graphing, mapping, widgets
Thank You Demo and discussion

Buisness Intelligence and Web Analytics