The data collected includes individual subject results (across multiple criteria), external testing results (eg, learning style analysiss, NAPLAN scores), behavioural information (as recorded by teachers, plus any 'escalations' that were deemed appropriate), involvement in co-curricular and external activities, notes made by the counselling department, account balance and parents' marital status (the latter two being used as in indication of family stress).
Being able to analyse these sorts of data may provide insights such as realising that it is counterproductive to nag certain students when they fail to submit drafts on schedule, or alerting teachers of the need to adapt to the learning style of a particular cluster of students.
Clearly, some of the data stored by the school is sensitive, so "the whole space we've designed is bound by security," said Carpenter. For example, teachers cannot access information collected by the counselling department, but the head of school can. However, there may be some benefit in alerting teachers to an issues in one of their student's lives before "I left my homework at Dad's" is tendered as an excuse.
"When you find something that can make a difference, it's exciting," said Carpenter.
Carpenter noted that there was little pressure to deliver immediate results. This allowed the project to develop steadily, addressing what was actually needed in a way that allowed for future flexibility rather than choosing whatever gave quick and easy results.
Donald Farmer, from Microsoft's SQL Server team, noted that Microsoft's approach means it is possible to experiment with the tools that already exist within a school. Teachers can do their own exploration, and the most successful analyses can be put into production.
Farmer noted that the alignment between features of SQL Server 2008 R2, SharePoint 2010 and Office 2010 provides in 'in' to millions of potential BI users around the world. Microsoft is ready to put BI into the hands of users, he said, noting that IT departments are suffering cutbacks despite growing data volumes. If users can't do BI, it probably won't get done at all.