The Data Dilemma: Tips on Optimizing Data Strategies to Maximize Automation's Impact
If automation is the magic tool for shared services, right now – data is the power that runs it
While 'data' has been integral to linking applications and maximising the benefits of an ERP system, it has taken on an urgency more recently, given that automation’s value-add depends on the structure and accessibility of enterprise data. In a recent SSON survey of Nordics-based shared services executives, one out of four practitioners professed "data analytics" as the biggest "buzzword or opportunity" for shared services in the year ahead. And that may still be underestimating the real opportunity.
Given the region’s relatively high cost base, organizations have always been highly tuned to the opportunity afforded by "leverage.” Technology is proving highly interesting in this respect, but its returns depend heavily on the ability to tap into large amounts of relevant data.
This data, increasingly, has less to do with the ability to store and process or create and delete, and more to do with using information. Data supports better decision-making and enables value-adding outputs based on the vast data lakes and the data fabric that extend across an enterprise. But to be effective, this data needs to be clean, accessible, digestible, in context, and in the proper format for supporting decision-making and analysis.
And that is where the problems arise.
As the enterprise becomes increasingly digital, vast quantities of data exhaust are released across the enterprise ether. One of the by-products of automation is that we now have complete transparency across everything – presenting a much fuller picture of the user profile, content profile, and context profile. But the sheer quantity of this data can be overwhelming and organizations that have tackled the challenge head-on are already coming up against the upper limits of what can be managed.
At the same time, many of the more evolved intelligent automation tools, for example those integrating machine learning or cognitive capabilities, dependent on enormous amounts of data to "feed the beast" and mine and learn from patterns.
In order to manage this data, no organization should overlook the importance of a robust data strategy, which provides a framework for identifying and surfacing the data that's relevant for a given activity or process. Some organizations have designed "maps of relevancy" as a basis for collating or collecting the information that's needed.
The advantage of the digital environment means that enterprises can also tap into virtual data lakes that effectively bring pieces of a transaction to the data for action, and then return just the results. This massively simplifies the need to move large quantities of data around the enterprise and facilitates analytics and decision-making.
“Data fuels automation’s capabilities.”
Simen Munter, COO of Nordea’s Commercial and Business Banking division, acknowledges the all too real "gap" in data as workflow automation becomes more common across the enterprise. The limitations, he explains, extend from having the right data segments, to accessing data, to its quality. This is all the more significant as the opportunities afforded by intelligent automation and artificial intelligence are fundamentally driven, or enabled, by data. In other words: Data fuels automation’s capabilities.
This will only become more apparent as organizations embrace the cognitive capabilities associated with machine learning and prescriptive or predictive analytics. Machine learning requires not just large quantities of data sets but also large quantities of patterns for these machines to observe and learn from.
Shared services executives that have recognized this are quickly putting relevant strategies in place. Here are 10 steps to implement in order to optimize your data strategy.
- Communicate data management as a core skill and prioritize these roles – With much of the actual work being automated in future, the value-add of the human workforce will be its ability to drive, manage, and work with data.
- Reassess the role of humans – Instead of processing transactions, in future the human workforce needs to re-focus on creating the data sets that feed automation.
- Find, sort, and structure your data – Whether via mapping, identifying the source, structured or unstructured, real or virtual, your team's ability to identify relevant data, and sure it is accessible and workable, and feed it into the appropriate channels will be key.
- Identify the black holes in your data – With the vast majority of valuable data still unstructured, one of your key challenges will be to identify your gaps and fill them.
- Reconsider data flows – Traditional data flows are sequential, but in robotic or intelligent automation, systems can tap into multiple data sources at the same time. This means rethinking data flows from the beginning.
- Data-oriented processing creates an ever-wider circle of influence – More and more activities will be pulled into scope once you open the doors. Plan for this.
- Shifting into more sophisticated tool sets changes requirements – Do you have the data necessary for those new systems? If not, how quickly can you access it? Or does it make sense to reassess your tool implementation strategy.
- Plan for the future – Once you understand what data is available, design a plan for the data that isn’t available.
- New tool sets treat data as a virtual lake – Rethink "storage", access, and time.
- It’s not what you do, but how you do it that defines the future.