Day 1

I'm a bit overwhelmed and dismayed at the workshop choices here at the Gartner BI Summit. There are so many sessions I'd love to attend, but time is at a premium. Data Discovery/Self-Service BI or Introduction to Hadoop? Data Warehouse Best Practices or Healthcare Analytics? They're all worthwhile, and applicable to many current CapTech clients. But it looks like I can't go wrong no matter which ones I attend.

The day started with the Keynote Address, with Bill Hostmann, Ian Bertram and Ted Friedman joining forces to discuss the current state of BI, and the new direction its heading in the next few years. Key points included:

  • By 2020 everyone and everything will be connected to the internet (and generating data)
  • Lots of noise, but important signals – the better BI Tools will differentiate themselves by helping separate the signal from the noise
  • New skills will be required to manage the new data
  • Very few companies use predictive (13%) or prescriptive (3%) analytics. The evolution of BI will be in this direction –" don't just tell me what happened and why, tell me what do to next"
  • Big Data is the next wave, and will involve many different skills and perspectives

My next session was on Self-Service BI, hosted by Dan Sommer. Clearly there are numerous issues with Self-Service, not the least is the political issue of data and tool ownership and governance. Ownership and management of Self-Service is often at the Department level, and begins to grow out to Business Units, not from the Enterprise down to the Business Units. The key takeaways include:

  • IT must partner with the Departments to ensure a common set of Metadata and data quality
  • Power Users should be allowed more access to data (within legal/ethical/business guidelines). These users are the key to effective data discovery.
  • Pricing for Data Discovery tools seems to increase as the number of seats increase, the opposite of traditional BI Tools.
  • Most traditional BI Tools are behind pure Data Discovery tools, so press your current Enterprise BI vendor on their Data Discovery plans
  • The "Shadow IT" group will not go away - don't build a wall around your information, but instead work with them to ensure that the data they need is provided.

Rita Sallam presented a workshop on Deriving Value from Content and Social Analytics. The most interested takeaway here was the diverse skills needed to extract value from the fire-hose of social information. Some of these skills include understanding unstructured data, extracting information from video, interpreting sentiment, email and text analysis, context analysis, and semantic modeling. The best example of this was interpreting the real meaning of "Apple killed it with the new iPhone5". What does that really mean?

My next session was on Data Warehousing Best Practices, led by Roxanne Edjali. She addressed concerns from her clients that traditional Data Warehouses were becoming obsolete – not true. If anything, they are maintaining their relevance, and becoming more critical. Her main points were:

  • Well-managed Metadata is becoming more and more critical in order to manage the constant flow of new information
  • Data loads are becoming more frequent and irregular, so adjustments to capacity and architectures need to be managed in an ongoing way
  • Clear roles and responsibilities need to be defined in order to better manage the information. Include business users in these groups.
  • Big Data experts must join the team, not create their own.
  • Avoid alienation of the various "experts" in the new data – recognize the unexpected skills and make good use of them.
  • Make mixing the business side and the IT side mandatory

Last was a discussion of current trends in "Big Data" with HP. Highlights include:

  • Big data used to be a boutique area, now the big players are moving in and it's more mainstream.
  • There is a platform shift from generic hardware to more specialty hardware like in-memory modules and system-on-a-chip.
  • There are huge and wide reaching impacts to BI from "Big Data". These include new ways of fraud detection, medical diagnosis and public section situational awareness.
  • There is a merger between business strategy and technology strategy to provide new tools for marketing – these provide almost near-real time information .
  • A few statistics:
  • 90% of all data will be of a mixed type by 2015 (no definition of "mixed" was given, but we can draw a few conclusions)
  • 75% of current data warehouses will not scale to meet the new velocity and complexity of data demands
  • There will be a 48% growth rate in digital content
  • 86% of companies cannot deliver the right information at the right time

That's it for Day 1. Check back here for a summary of Day 2!