In 2013, Gartner conducted a survey on Big Data Adoption in Supply Chain Industries and found that adoption has been flat and is lagging behind the overall adoption rate of other industries such as banking, insurance, and retail to name a few. Gartner ascertained that these characteristics pertaining to the Supply Chain industry are attributable to an inherent lack of understanding of what Big Data truly is and a fundamental lack of the required skill sets. This, in essence, is the challenge facing the Supply Chain industry.
In parts one, two and three of this four part series on Big Data, we looked at what makes data “big”, how it can benefit organizations that apply the right analytics, and the implications of doing so, respectively. In the closing segment of this series, we focus our attention on examples of how elements of Big Data can be leveraged. The challenges identified by Gartner translate to opportunities regardless of your organization’s analytical maturity level.
Big data is about a better perspective, first and foremost. It is a resource that when refined can highlight opportunities and threats facing your organization. What get’s lost in translation is this notion that in order to get results, it demands the neurological capacity of a team of rocket scientists hammering away on an infrastructure that puts the CERN to shame. Fear of the unknown is an incredible impediment, you see. On a daily basis, more public data sets are published, more industry data aggregators emerge, and more data is generated by your organization. Insight lurks inside this ocean of data and it only surfaces when it is summoned.
To emphasize the concept of “big data surprise”, DC Velocity published an article recently on how an unlikely data set provided the insight that a distribution center needed to understand why there was an increase in forklift related damage. This discovery begins with asking the right questions, and ends with validating them using the necessary data and straight forward analytical techniques. MIT grads need not apply (sorry).
Let’s look at a freely available external data set called the American Customer Satisfaction Index, and how it is a macroeconomic leading indicator for consumer spending and economic growth.
See anything interesting in the diagram above? The consumer spending metric has been lagged one quarter to emphasize that, generally, when the consumer satisfaction index falls, people spend less money. The ACSI is a leading indicator. In fact, the ACSI can be used to predict Gross Domestic Product performance and has been demonstrated to foretell the financial performance of an organization. If your demand is consumer driven, try comparing your performance by overlaying your financials with the ACSI trend and see what you get. What do you see? Big data is not just about the data you generate and complexity of the approach used to produce insight. It’s the delicate interaction of the vast, growing ocean of data that surrounds you, and your approach towards harnessing it and proving (or disproving) an educated guess.
Oftentimes, the legwork to produce correlations has already been done, ACSI being one example. Another example is your FICO, or credit score. It is not only a leading indicator of your economic responsibility, but also indicates whether you will speed, get into accidents, commit a crime, or commit fraud. It is the definitive Key Performance Indicator for measuring risk. In fact FICO has re-branded itself into a big data player by providing data, and analytical products and services. Ever try comparing your vendors’ FICO or credit scores with their overall performance? Wouldn’t that be an interesting exercise?
Sometimes it is about the data you are not capturing. How would you know how to pinpoint the characteristics of a customer who is about to churn when you don’t actually keep a historical record of inactive customers? Doing so and analyzing transactional data would allow you to proactively address these scenarios. Every bit of data helps. If you don’t have it, take the steps to ensure that you do.
With a keen understanding of your market forces, the wealth of data is only a few keystrokes away. Let’s look at Brenda, a buyer at a healthcare organization who is tasked with planning next year’s demand for medications. Planning demand should factor in external data sets that influence your industry. Brenda is undoubtedly interested in the relationship between medications and demographics, such as age. If she knows that the general population is estimated to get older over time, she would also know that sales growth in medication aimed towards the elderly will likely follow suit. That’s census data. That’s World Health Organization data. What about disease incidence trend data from the Center for Disease Control? Some flu seasons are worse than others and flu seasons are starting earlier and finishing later. Maybe Brenda can take a look at Google Trends and Trends Graphs? Some say this is the best place to capture leading indicators on almost anything, simply based on search data. Brenda may even want to track when certain drugs are scheduled to go generic because that will tell her how her purchasing power will change over time (i.e. a certain drug could become cheaper when it goes generic). Where does she get that? The Orange Book Files. That’s the data that tells you when drug patents will expire. Maybe she wants to buy medications with a higher efficacy rate? The Adverse Event Reporting System, all collected and served by the great folks at the Food and Drug Administration. Or, she could measure efficacy by fetching data from her hospital Electronic Health Record. That could be a straightforward cost-to-outcome ratio. If Brenda wants a ‘one stop’ shop for data, she could visit her neighborhood ‘data aggregator’. There are plenty of these players who serve both free and fee based data. In more advanced cases, some organizations let others figure it all out. You can provide your anonymous data to ‘crowd-think’ sites like at Kaggle where you can run a competition to produce predictive models for your company. Netflix offered $1 million dollars if someone could improve their movie recommendation engine. Many organizations have followed in Netflix’s footsteps.
Brenda is one example of a specific function in a specific supply chain industry. Think about the data you can refine for your particular supply chain industry. If you are already doing this type of analysis then congratulations; you are already scratching the surface of big data. Don’t let them make you think otherwise! Now think about taking it up a notch.
Reigning in big data doesn’t require ‘experimental tech’ to collect and house the data and esoteric algorithms to make sense out of it and predict outcomes (as wonderful as that is if you have the means!). It is simply about discovering the unknown through the justification of hypotheses. It could be as simple as this: think about forces that influence your performance and grab the relevant data. Next, create a simple line chart to juxtapose your performance data relative to market force data across time periods. Now, add and remove layers of insight and see what you find. Rinse and repeat. This is how the picture becomes richer and more vivid. This is how simple it is to find strong correlations or to prove a hypothesis. Take these results and apply sets of data on a plot diagram and produce a trend line to see how powerful the correlation is (for those of you taking score, all of this is called multivariate analysis). Think outside the box by truly understanding the forces that facilitate or obstruct you as you navigate towards your corporate mission and vision. With a sharper view you can anticipate the trickle down effect that these forces will have on your ability to handle changes that occur downstream, like your inventory levels and service volumes, and look upstream to identify leading indicators of certain forces, such as an ACSI or FICO, to improve your outlook. As you mature and begin to handle more complex and high speed data sets, your clarity along with your analytical maturity will evolve. When you get to that pinnacle where you are analyzing the sensors of your conveyor system or air control systems to predict failure, or mining social networks and predicting imminent and negative customer sentiment then you would have mastered the art of data science. Your journey needs to start somewhere.
Data is the new oil. Right now, there is a silent war of titanic proportions being played out and it is about who can reach as many people as possible in order to generate and collect as much data as possible. Just look at some of the recent acquisitions by Google and Facebook, and then think about what it means to them in terms of reach. How else would Facebook predict when someone will fall in love or Google predict the onset of the flu season?
The point is that you need to firmly understand the landscape of forces that affect your business and to be open to learning about forces that you had no idea even existed. Data science is about constant analysis of disparate data sets to identify those golden nuggets that can take your organization to the next level. This is where the supply chain industry needs to focus in order to reap the benefits and shuffle off the ‘laggard’ label bestowed by Gartner. No matter how you look at it big data has always been, and will always be, around. The opportunity facing the supply chain industry is to develop the necessary skills and awareness of the vast ocean of data through which it navigates.
So what do you do? Besides becoming intimately familiar with your data ecosystem, invest in developing or acquiring talent who can quickly understand your market forces and, in turn, who can fetch data, mash it together and identify opportunities and threats. The most successful data scientists come from completely unrelated fields. They bring to the table fresh perspectives and, most importantly, creativity. If you have challenges, partner with those who can help. We are there. Start small based on your maturity level and grow from there. Get those much needed quick wins and build on them. And don’t forget about that time tested challenge; change management will also be a significant hurdle that you must overcome. Organizationally, you will be faced with push back by those self proclaimed data architects with a false sense of entitlement over your corporate data assets, trying to sway you into the belief that ‘it can’t be done’, because they fear change and are absolutely terrified of losing control. The new world is about social collaboration and sharing, not about data fiefdoms. This paradigm shift begins with you.