To remind everyone, I provided the following definition of Dark Data in my post
Dark data is data and content that exists and is stored, but is not leveraged and analyzed for intelligence or used in forward looking decisions - Isaac Sacolick, see full definition
Since then, the subject of dark data has been covered by CIO.com (The Dangers of Dark Data), Forbes (Factories of the Future), and VentureBeat (Are you afraid of the dark data?). VentureBeat points to the cause of the problem, "
As storage has become cheaper, those who generate data have grown used to hanging onto it... When data is “dark,” it’s often because the organizations that own it lack the tools, infrastructure, or skills to effectively leverage it - John JosephInternal Dark Data, that is, dark data that is already captured and stored by the enterprise but not leveraged to drive insights or decision making represents a threat and a possible missed opportunity. The threat is if storing the data impacts the performance of a key business operation or contains sensitive information that should be better secured. The opportunity is to really figure out the value in retaining and processing this data.
Determining the Business Value of Stored, Dark Data
One of the themes we discussed during the podcast is developing the discipline on identifying the business value in dark data. To do this, use basic data visualization, analytics, and quality tools to identify the substance of the data and look to answer some basic questions:
- How to catalog the data so that business users can learn about its existence? Can the data be broken down into basic entities, dimensions, metrics and volumes to provide more details to business users looking for data sources?
- Identify 3-5 potential questions, insights, decisions, or activities that can be researched using this data should someone commission a data scientist to investigate.
- Also identify "know issues" with the data source. This can be measures of data quality, information on how the data is sourced, and other feedback that might undermine any analysis of the data.
- Have a Data Governance board "score" this data set based on its potential vs. known issues. Absent of any easy way to quantify value, scoring by a voting committee can at least rank what data sets look attractive for further analysis.
- Commission time-boxed studies (aka, agile sprints!) on data sets that have the highest scores. Review results and re-rank based on findings. (Note: See my post, Best Data Visualization Practices in Self Service BI Programs for some ideas on how to implement.)
- Make sure that you have disciplined agile data scientists. Have them demo their findings to the Data Governance board and adjust the Score based on what was discovered.
For additional insights,see my post 10 Attributes of Data Driven Organizations.