Much of the publicity around Big Data has boomed its power to capture large amounts of data and aggregate the results to make operational decisions. Admittedly there is value in this and many organizations are already realizing that value. However, these are mainly operational applications that need specialized computer systems and well-trained data scientists. Their decision makers must know ahead of time which operational variables to collect data about. Their data scientists must know where that data can be found and how to visualize it so meaningful decisions can be made.
At CIMS we concentrate on a different approach, one that we expect will have a much greater payoff for industrial organizations. We are developing tools and methods to use Big Data for making strategic and innovative business decisions in what we see as the next-generation decision: the data-driven decision.
In data-driven organizations, decision makers will formulate strategic questions such as which markets to enter, which products to develop, how to position the company, and whom to partner with. Unlike operational decisions where data type and location are known, strategic decisions cannot assume ahead of time what information is important or where it is located.
Staffing Decisions for Kelly Services
Kelly Services is one of several industrial organizations we have been working with on this approach. The global staffing services firm wanted to know if flexible staffing in healthcare was a commercially viable new service offering.
To answer Kelly’s key question we had to begin by looking broadly to determine what information was critical even if it was not presently known. Variables such as state-by-state payer regulations, the supply of healthcare professionals qualified to offer services in a flexible and/or distributed fashion, the demand for flexible healthcare workers, healthcare trends, competitor actions and their level of market penetration must all be assessed.
If you want to determine whether or not a new line of business is viable you cannot simply add up existing operational information and visualize it in some kind of dashboard.
Gather Unrelated Information
Making strategic and innovative decisions requires gathering seemingly unrelated information from disparate sources. Unlike operational data, critical information for strategic decisions is often qualitative in nature. For example, knowing that state legislatures are considering bills for telemedicine might impact flexible healthcare where staffing is critical, but you can’t add, subtract, multiply, or divide that information or render it on a traditional dashboard in any meaningful way.
Strategic and innovation decisions require information from a variety of sources not always knowable at an early stage of decision making. These types of decisions require a variety of inputs such as competitive environment, market attractiveness, trends, customer drivers, patents, regulations, competitor actions, etc. The inputs are usually a mix of information that may or may not be able to be mathematically aggregated.
The CIMS Big Data Process
Working with Kelly and other CIMS member companies, we developed and tested a process to use Big Data to help make strategic and innovative decisions (Figure 1). The process still requires specialized computer systems to handle large data sets as well as conduct natural language processing. But rather than just aggregating large amounts of information, we search through large amounts of data to isolate the critical information.
To do this we must define the question and develop rules the computer can use to search through vast amounts of data. For each step in the process we developed tools to guide and focus decision makers on activities that result in making truly data-based decisions.
Step 1. Select the Project. — This may be difficult if managers do not understand how to use Big Data for industrial innovation and strategy decisions. A fundamental understanding of how to use Big Data is necessary for this step. CIMS has assembled materials to efficiently bring all the decision makers and subject matter experts up to speed on how to use Big Data and what roles they need to play.
Step 2. Define the Question.— This is surprisingly difficult to do. Big Data offers opportunities many decisions makers are not aware of. Applying the logic of being able to ask previously unknowable information takes some orientation. We provide tools to help gather thoughts with numerous examples and a template for forming Big Data questions (Figure 2).
Step 3. Key Words and Dictionaries. — Rather than searching a handful of key words at a time, natural language processing allows you to search all possible synonyms as well as phrases at the same time to produce a tight set of results rather than millions of useless pieces of information. You will likely be interested in multiple sets of words called dictionaries. For example, you may be interested in healthcare workers and flexible workers. Therefore you would create a dictionary of all healthcare worker terms and a separate dictionary of terms that describe flexible work. Again we have a template to help identify key terms and phrases. As the project progresses, Big Data will be used to identify words and phrases decision makers did not think of initially.
Step 4. Sources of Information. — Where one gathers information can be as important as the information itself. For example, if you want to find which states allow paying for flexible healthcare workers, you would want information directly from each state government and not news articles or social media. If you want to know if doctors or patients object to flexible healthcare you might want use social media rather than government or academic sources of information. Our Sources Worksheet helps you identify likely places critical information can be found. In this way Big Data is used to find many previously unknown sources of information.
Step 5. Design Rules. — Rules tell the computer how to use the data to isolate critical information. For example, tell the computer to return results for flexible workers and healthcare and you get a list of flexible healthcare workers. Add in which states allow payment for flexible healthcare and you get which states pay for flexible healthcare workers. You then add in dictionaries for each healthcare specialty, salary ranges, number of healthcare providers that offer flexible services, the supply of flexibly trained healthcare workers, and you begin to see the real business opportunity. We have templates that guide you through the rule-making step.
Step 6. Big Data Workflow. — This is the work conducted by the data scientists that combine your project definition, specific questions, key terms, data sources and rules to return results critical to your decision. This work requires specialized computer hardware and software as well as specially trained personnel familiar with using Big Data to isolate critical information rather than just aggregate data.
Step 7. Data Assessment.— Before the information is used to make decisions your subject matter experts must assess the adequacy of the data. Content experts need to assess if the information is related to your company and specific questions being asked; they must also determine if there is enough information, and they need to ascertain if the sources of information are credible. We provide a Data Assessment worksheet to guide your experts through assessing the data.
Step 8. Results Scorecard. — After the data are found to be adequate for decision-making, your subject matter experts and decision makers are able to score the data on the critical dimensions for your company, such as: strategic fit, customer needs, strength of your offering in the market, competitor actions, business environment trends, expected financial and other types of impact. Again, we provide scorecard templates and processes to incorporate assessments and inputs from all affected parties.
The process described here will help you use Big Data to make industrial strategy and innovation decisions. While it does required specialized equipment and skills, it does not require purchasing everything without trying it first. CIMS will help you perform a demonstration and a pilot on your questions. You may contact us at the addresses below.
As for the original Kelly Services query, we found an un-served market space, willing partners, complex regulations, and a management team eager to use Big Data again.
Stephen K. Markham, Research Associate, NC State Center for Innovation Management Studies (CIMS); Stephen_Markham@ncsu.edu,
Michael Kowolenko, CIMS Industrial Fellow, NC State University Virtual Cloud Lab, firstname.lastname@example.org