Data is being gathered at an accelerated pace, by everyone from governments to businesses to individuals. For businesses especially this is resulting in huge potential for growth and innovation. However, this is affecting organisations’ ability to strategise as managing the vast volumes of data has become a strain on infrastructure. Nimbus Ninety members gathered at the Andaz Hotel in Liverpool Street to discuss how to overcome the problems of an increasingly datafied world.
BACK-BREAKING DATA VOLUMES
Data’s presence has been ubiquitous and constant for centuries (Domesday Book was published in the late eleventh century). What has changed the game is its scale. It has become a truth self-evident that data is growing exponentially. A now commonly known fact is that more data has been created in the past 2 years than in the whole of human history. In 2013, there were 4.4 zettabytes of data. By 2025, projections are estimating a grand total of 44 zettabytes of data will have been created.
Many businesses are looking to become data-driven but only 0.5% of data available to them is actually used — unsurprising when there’s only a finite amount of data which is actually useful. But how does one find out what data is the right data?
A good case study to look at is that of CERN in Switzerland. They use on average 0.1% of their available data. So why is it that they only use a fifth of what everyone else uses? Two reasons. The first is that due to the amount of data they have, they simply can’t store it. The other factor is simply that they can afford to ignore the majority of it. Why? They know what they want to do with it; they have specific use cases for scientific discovery which means they can easily identify what data they can keep and what they can discard.
So, the first step to developing a great data strategy is working out the business strategy. To illustrate this problem Paul Brooks, CTO of Dell, posed this question. “Do you want to grow or do you want to consolidate?” For example, which is more important to you: making money or saving money? Brooks stated that doing both simultaneously and at scale is almost impossible so making that decision is a vital move: “if you don’t know your business problem, your vast amount of data isn’t worth anything”.
Once your business strategy is finalised, an important next step is developing your infrastructure. Infrastructure is becoming more vital as it tries to cope with the pressure of increased volumes of data and businesses are starting to recognise its importance (a survey of Nimbus Ninety members revealed that infrastructure would be their highest expense this year).1 Whilst tooling is critical to this process, another key issue is having the personnel with the right skills to use the technology and thus deliver the projects that align with your company’s goals.
MORE DATA, BETTER SKILLS
There are, unfortunately, a number of issues in this area. The skills that data scientists possess are becoming even more of a premium. There are so many different approaches to data science projects (regression, clustering, visualization, etc.) and the average tenure of a data scientist is also much shorter in comparison to other roles: the job website, Indeed, claims that the average tenure for a data scientist is less than a year.2
One solution posited was to create “feature teams”. This involves bringing together different skill sets into a team that works on a project together, led by a product owner who can oversee delivery. Bringing together skill sets is indeed useful for project delivery but simultaneously bringing together disparate data sets is a fantastic way of improving things.
However, this type of change needs top-down promotion. What became clear over the course of the evening is that leadership behaviour needs to evolve. NTT’s Wayne Speechly noted: “strategies are driven by behaviours. Behaviours are driven by belief systems which are, in turn, driven by leadership.” If, for example, the business strategy is to grow, then leaders need to instill behaviours in workers which discover what data is going to help achieve that end. The feature teams can then use that data to provide insight into how best to approach your business problems.
The idea that brought murmurs of assent among the Nimbus membership was the concept of companies becoming more collaborative with their data. We learned that competitors are starting to work together for a greater good of solving the data issue. Further to this was the notion of being more open with data sets.
This, however, brings up problems surrounding regulation and governance. When your business has access to the data of millions of customers, sharing that data can be somewhat of a minefield. During the panel discussion, one of our members stated that they could readily accept their data being accessible to multiple organisations as long as it wasn’t being used for “nefarious purposes”. The nods that greeted this comment were tempered by the assertion that regulatory institutions are, as they always have, lagging behind innovation.
On this note, our panel agreed that there needs to be a global set of standards to facilitate collaboration between businesses. Changes such as GDPR regulations are only the beginning. Indeed, this movement is barely in its infancy. But with regulators stepping up to the plate, in conjunction with changes to team structure, we could start to see the elimination of data silos (or “people silos” as one of our members called them) within and between companies as well.
1. Nimbus Ninety, Digital Trends Survey 2020.
2. Indeed, Accessed 30 January 2020, https://www.indeed.com/career/data-scientist/salaries
This event was held in partnership with NTT, a technology services company that provides intelligent technology solutions across the globe.