Data Supply Chain Management
Data has become the biggest new frontier in business in the last 20-30 years, but the amazing thing is that data has always been there. It’s been the artifact of people, process and technology for decades—as we’ve invested into technology, data was always the byproduct. It’s only recently that data has become en vogue due to the introduction of multi-parallel processing that has taken the cap off of compute and storage.
Now, analytics can be applied to larger data sets in timeframes that are sub-second or even real-time. In addition to technological advancements, we’ve seen rapid improvements in organizational maturity through the introduction of new roles devoted to leveraging the technology advancements — roles such as data scientists and data engineers to name a few. If you look closely at these two components — people and technology — it really does represent a tsunami of change. Organizations must put business rigor and management protocols into change management to ensure that investment into these new and complex areas yield returns that measurably enable the business strategy. I know, I just strung a bunch of corporate buzz words in that last sentence, but if you follow along in this journey, over the next few paragraphs I’ll provide some context into exactly what that sentence means and how your organization might think about data, not just as a corporate competency, but as a strategic weapon.
We approach organizational data challenges through a methodology that was created to manage strategic investments into all aspects of data, and to ensure that financial benefits of these strategic investments are realized. Our unique methodology is called Data Supply Chain Management. It is designed to start with raw data (commodities), then, through metadata harvesting and creation, we enrich and transform the raw data into sub-assemblies (views and tables) and ultimately into finished goods (insights and visualizations). The finished data ‘products’ are highly search-able and usable for various data science and advanced analytical activities. This unique approach to managing data provides the ability to manage performance metrics and KPIs that will ultimately speed up the critical measurement of analytical productivity and competitive advantage through analytics: Time-to-Insights.
The Time-to-Insights measurement system above illustrates how data can be managed at both the micro and macro process-levels to optimize business outcomes. By using various forms of metadata and data quality analytical techniques, you can measure and manage the movement of data from raw materials (i.e. data generated by applications, machines, mobile platforms/apps, people, etc.) through to analytical insights while maintain integrity, continuity and visibility of everything that happened to the data as it moved through the supply chain. Treating data management like other business functions by having management metrics and process in place is what will unleash competitive advantage. Below is a comparative analysis among companies within an industry that are differentiating through data versus those that aren’t.
As an organization begins to manage data more like a supply chain, three things occur:
(1) the critical measure of Time to Insights begins to shrink.
(2) if investment allocation into the Data Supply Chain doesn’t change, the number of Insights produced in a reporting period increases
(3) the feature engineering and volume of features implemented into advanced models will increase.
As these variables begin to change, there are secondary outcomes that occur as the broader business begins to absorb the supplied data artifacts. As more insights are created, the business is able to test outcomes and decide which insights are of use in both short-term and long-term plans. Next, as more features are added to advanced models, the models become more accurate, precise and robust. This is when an organization’s data truly transforms into a competitive advantage. Organizations grow over time, and most of the time, their data grows with them. As the company pivots and changes, so does the data; therefore, data is often unique to that company; Though might share the same structure as competitors from a macro perspective (i.e. Insurance companies have very similar data models), each organization will vary greatly at the data value level due to the variations of business architecture. As companies add features to models to increase accuracy, precision and robustness they are also increasing the uniqueness of the model itself. Over time, they start to become “one of kind” within their respective industry. We have found this pattern to be true in Insurance, Banking, Telecom, Utilities, Energy, Pharmaceutical, Asset Management and Medical Device industries. So, what can organizations be doing to put themselves on the path of competitive advantage with data? In future posts, we’ll cover some additional patterns that we’ve seen related to how investment and capital allocation is occurring today versus where the focus needs to be to yield the three outcomes described above that lead to competitive advantage.
We have found that most organizations focus investment and capital allocation towards the book-ends of the supply chain – data supply (the blues above) and/or data demand (the greens above). In most instances, they neglect the middle stages of the Data Supply Chain that enrich the data. That approach leads to wasted time and failed data initiatives. For example, some organizations take advantage of the cost savings of big data platforms in order to hoard vast datasets in unmanaged Data Lakes in the hope that the collected data will somehow add value to the organization. Unfortunately, this approach leads to vast quantities of unknown and unfindable data that isn’t readily available or usable for analytical activities.
Organizations that focus on hiring Data Scientists without robust data enrichment procedures, and more specifically, a data supply chain management approach, wind up having highly valued resources who spend their time combing through datasets in search of something useful. If they find something promising, they spend weeks if not months, wrangling the data into shape so that they can begin applying their high-priced expertise in advanced algorithms. At organizations that have vast data lakes, the time spent on this process of identifying the data only multiplies. This situation can lead to false positives of high demand on resources that are perceived to be constrained, so the feedback loop results in the hiring of more data scientists!
In these scenarios, the organization has put significant investment into big data without a clear path to ROI. We know that a more holistic and pragmatic approach is necessary – one that effectively manages data throughout the Data Supply Chain, and all the benefits and efficiency KPIs required to guide investments into data.