Systems thinking

Data leaders can get the most out of their efforts by embracing the fact that large organisations are complex adaptive systems. To influence the outcome of the system, we must understand the agents (people) and their interactions (processes, systems, policies, and other unspoken norms). Implementing new technologies and systems of work often produce underwhelming results - despite some successes (we always shout about those!), the experience from the trenches is somewhat painful. The data leader has other options that don't look like either a "test lab" or a "multi year transformation". 

Image by marian anbu juwan from Pixabay

Data leaders have a wide range of remits in the financial sector, from owning group-level, global data policies to delivering cutting-edge data analytics products in experimental ways. Among some common challenges for all these roles, data minimisation stands out as an important one. It impacts compliance with regulation, such as privacy and security, as well as cost management and sustainability. Not to mention, of course, it is a foundational data management principle. In practice, it’s an uphill struggle. It appears to me that far more effort in an organisation goes into the creation of data – more products, more tools, more business activities – than into the management of its lifecycle until eventual disposal. It is time to look at this phenomenon in a different way and give systems thinking proper attention. 

Large organisations are complex adaptive systems. They are networks of people inside and outside the company, who change their behaviour and processes in unpredictable ways – sometimes in unconscious ways too – to achieve their goals. Thanks to that, most companies have become resilient and innovative. On the other hand, these same characteristics make businesses increasingly challenging to control. Leaders that understand this dynamic nature of the organisation can get the most of their efforts and resources.  

Data leaders are not immune to the accepted path for getting work done in a large company. They have to put forward business cases for “data products” or good-old-fashioned projects as per the business rituals and traditions. However, those units of work (products and projects) are suitable for building or changing “parts” of the system – such as processes, user journeys, IT applications. In complex systems, those parts are not static.

Systems thinking offers a good approach to influencing the “whole”, such as macro-functions, departments, regions, product lines, rather than its parts. Using systems thinking, the focus of change efforts moves to the behaviours and interactions of people that lead to the outcome we see in the “whole”.  

Image by Chen from Pixabay

Given that financial services are highly (if not entirely) dependent on digital systems, a way to see the “whole” is to observe data propagation at a macro-level - note that I’m using the term digital here to encompass all forms of IT solutions, not only the apps produced by the “digital” teams. Whether by real-time applications, batch jobs, or manual processes, data is silently copied from department to department, from system to system, from databases to email inboxes, collaboration servers, and executive dashboards. The dynamics of digital data is a reflection of how your business operates, more so than documented org charts and process flow diagrams.

Back to the data minimisation challenge: Attempting to put a policy into practice by changing every process and getting people to evidence they are doing it is, at best, ineffective. Accepted controls, such as attestation statements promising that their project has ticked the “Have you considered data minimisation?” box is well intended, but borderline meaningless. Most people don’t go to work thinking “I’m going to make copies of this data even though I can easily access it at the golden source!” 

Instead, a data leader must bring the real question to the fore and tell a compelling story to the executive team: Why do people feel that copying is easier than collaborating? Is it because individual performance is so tightly managed that they need to be in control of “their” data? Is it because digital data is not accounted for? Is it because data “space” and cost have become someone else’s problem, deep in the IT or procurement function dealing with “cloud bill” or “enterprise licence”? Is it because there was another project to delete and archive data, hence “it’s being fixed”? Is it because the tens of thousands of employees that make up these large organisations feel they only need to pass the mandatory e-learning training to be compliant? Is it because leaders feel overwhelmed by the size of the data problem? 

I grant that the bottomless pit of legacy systems has a part to play, and I don’t wish to downplay the technology and processes issues here. But consider that unstructured data in emails, personal network areas, and collaboration platforms contain data that is just as valuable – especially as we go through the AI hype. Once again, this data reflects how your business trully functions.  

In my next budget discussion, I’ll advocate for getting answers to those behavioural questions before jumping into building or rolling out more controls and processes to manage data. And I will focus on learning more about effective education and communication before assuming that a not-so-well defined data literacy training is the answer.