We have always known that culture describes groups who identify with each other and follow certain norms. Social networks reinforce these norms. Social media, whose explosion we have all witnessed in our personal lives, can amplify these norms and further reinforce the beliefs and social bonds of such groups.
An important question is whether personal relationships and social networks should affect an organization’s data culture and if so, how?
The relationship between social networks and productivity is complex and context-dependent. Also, when we search for academic research we find little on the use of social networks specifically focused around data or data governance. One study that comes close was conducted by the McKinsey Global Institute; it found that social technologies have the potential to increase productivity by 20-25% in knowledge-intensive industries. The study highlights that social technologies can help organizations capture and share knowledge, facilitate communication, and enable collaboration.
This reinforces my own position, that people and relationships are already the foundation of an unintentional data subculture that is not formal or led. We don’t have to look very far to see it.
For example, when someone is searching for specific data or has a question about it, they seek out expertise. They do this by using messaging applications, emails, hallway conversations, et cetera, to ‘ask around’ and see who may know something or know someone who does. This is not an approach based on any formal knowledge of peoples’ expertise. In other words, there is no written account of who to ask for what or how to collect knowledge from learned experience. This example will sound familiar and ring true for most. And what are people trying to do? They are trying to be productive by learning from others.
This unintentional data subculture is not sustainable or efficient; worse, it breaks down as people retire, data volumes grow, and the data landscape becomes more decentralized. We need to flip the script and intentionally foster and maintain a data culture that places people at its core.
I call an intentionally managed, people-centric data culture a social data fabric. It compliments Gartners' definition of a data fabric which is primarily focused on technology. Also, full disclosure, the term is already being used to mean a framework for managing social media data, but that use is relatively new and evolving, so I’ve commandeered it for data culture and governance purposes.
My definition of a social data fabric is the relationship between people, the data they use, and their common interest in specific outcomes.
Definitions are easy, right? The hard part and the key question is how do we actually foster and manage a social data fabric?
To begin with, we need to collect ‘creatorship’ and usage data. Creatorship is a funny word I made up, but I am trying to distinguish it from ownership which is a loaded term for governance people. Creatorship simply designates a person as the original creator of an asset, such as a dashboard, spreadsheet, query, report, table, term, metric definition, and so on.
Usage data is probably more obvious. It's the record of anyone using an asset. The word ‘using’ is nebulous. Does it mean accessed, read, copied/consumed, or changed? The specific definition will depend on each organization’s goals and the availability of that detail. The basic requirement is that we can collect a record of someone having accessed an asset.
Next, we need to be able to infer expertise. Inferred expertise is a process of taking the collected knowledge above and assigning a rating of expertise to a person for an asset. A vast array of statistical calculations could be applied to computing this, and it would be expected that human confirmation and feedback would inform the calculations and improve the accuracy of a given expertise rating over time. For instance, a person might be rated an 8 on a 1-10 scale of expertise for a report.
Finally, we must be able to cluster people into interest and knowledge areas based on their activity and role; these include the assets they use, their role or department, and the inferred expertise derived from their usage data. This is not rocket science. We know how to cluster. We just need to put it to work using this input.
The astute reader will notice that there is nothing very social about this, so far. It’s more of a ‘collect and compute’ process. But this is where we turn the corner and use what we have learned.
We should treat the clusters and groups as mini-communities and social networks. This involves driving communication that is specific to their interests so they are kept up to date on changes and offered ways to interact with and learn from each other. This also means providing them with mechanisms for easily establishing their own peer-to-peer connections. This includes things that are the equivalent of typical social media capabilities such as ‘likes’, comments, posts, and following.
What the social data fabric needs to end up looking like is ‘Facebook for data’ where each of our streams should be intelligently curated for us based on our data asset usage and who we decided to follow and connect with.
This will form the core of a data-driven culture and provide the foundation for shared learning and efficiency needed to unlock human potential.
McKinsey Global Institute. (2012). The social economy: Unlocking value and productivity through social technologies.