By Richard Kessler
I love cooking and working with ingredients to get just the right flavor. I once created a spicy dish with eggs and vegetables, but when I added the jalapeños toward the end of the process, they overwhelmed the dish. By experimenting, I found that if I added the peppers early in the process, I achieved a far more subtle but pervasive flavor by baking it in.
A small change to the beginning of the process made all the difference. I propose that an equally small change in how we manage information can dramatically reduce business costs and risk, while improving business decision making.
Enterprise data, whether stored in-house or with third parties, is as varied and disparate as it is growing in complexity and scale: electronic communications, such as email and text messages; files, such as documents, presentations and spreadsheets; and structured databases, such as an organization’s general ledger. Obtaining true insights that lead to improved business decisions, improved risk and compliance management and monitoring, and better investigations requires applying analyses across these disparate and diverse data sets.
A key challenge is the lack of a common paradigm for aligning information. Each data source has its own set of defined attributes, such as “person, product or account,” “from, to and subject,” or even the content tags we make up and apply as we create and save new data—such as with social media.
How does one create order and align the data in this universe of data? Attempting to align the data at the end of the process requires significant time, effort and resources, and the cost of achieving the desired result will often outweigh the benefit. But if we start at the beginning of the data life cycle and apply simple, standard setup attributes when we create or receive any type of data, we can dramatically improve our ability to align data later on.
I propose that we apply standard attributes to all new business systems, so we can gradually move all enterprise data into alignment. That involves:
· Developing a standard set of attributes via a cross-domain analysis;
· Selecting a small number of high-value attributes, say 10 to 15;
· Applying them consistently to all new data sources, with as few exceptions as possible; and
· Ensuring that the attributes are introduced when a system is materially updated or replaced.
If an organization looked across its structured, unstructured and communication data stores, it could identify a handful of common, high-value attributes that could be applied across all data. Aligning the data this way would deliver benefits across the enterprise.
For example, having all data tagged with the ID and role of the person who last changed or accessed it (e.g., employee ID, accountant) would help with early detection of unwanted behavior, with compliance monitoring and—if something untoward occurred—with investigative use cases such as discovery. Data would be pre-aligned to the individual.
Similarly, tagging data with attributes that connect it with clients or products would allow new insights into what clients want and need, what products rely heavily on the data quality of a particular data set, and where interest is growing in a new marketing campaign. Augmenting data as early as possible in its life cycle with standard attributes would create the desired alignment—often difficult and expensive later on—by baking it in at the beginning.
Typical high-value attributes might include the person who created, received or last touched the data; the reason it was created or touched; the person’s organization; the jurisdiction the person was in at the time; the relevant product or service; the relevant client; and whether the information is subject to regulatory requirements. Looking at each data discipline or use case, organizations could determine which attributes will most often help for these uses and then standardize on the best possible set—a select few, though enough to have a significant impact.
Ensuring Data Quality
Applying these standard attributes to data will also support efforts to ensure data quality. The importance in the modern enterprise of understanding and managing the accuracy, validity, reliability, timeliness, relevance and completeness of data cannot be overstated. Early alignment of data would facilitate more targeted investment to improve data quality where its improvement is needed most.
For example, if profitability depends on the speed of a certain process associated with a client, then having all the relevant data (emails, documents, customer profiles, etc.) tagged in a standard way could provide significant insight into how to accelerate the process to keep your organization making money, now and in the future.
Perhaps you’re thinking that this process would be far too expensive and complex. But it is not expensive to bake data-quality planning into the overall planning process, and enterprises should be thinking strategically if they are to compete and evolve to keep up with a rapidly changing landscape. If enterprises apply these attributes universally to systems during their normal change and replacement cycles, this might be a relatively small cost for a huge impact— and even make what once seemed impossible eventually achievable.
Often, the changes with the greatest impact start in small or subtle ways and become big over time—whether it’s carefully introducing a subtle flavor, transforming a business or finding all of the data associated with a particular person. If we make simple changes, apply them consistently, and introduce them gradually with few or no exceptions, we can enhance our ability to use the universe of data. Attaining new business insights, detecting and preventing unwanted behavior, and rapidly responding to events and incidents will all be significantly easier.
Disagree with this proposal? Think there’s a better and more cost-effective way to reach the same goal? I’d like to hear it.
To effectively use all the data they are collecting, companies must solve this problem. And they must start doing so today.
Richard Kessler is the executive director, head of Group Information Governance at UBS, and a faculty member of the Compliance, Governance and Oversight Council (CGOC). He has served more than 25 years in various IT and legal roles, including application development and design, trade automation, data architecture, and as a discovery consultant.