A new field called decision science has emerged to help organizations make the right decision and put it into practice.
What is decision science?
Decision science is a new approach to data science that strives to improve the quality of decisions and implement them throughout the organization. It combines data-driven insights with other fields such as cognitive science and organizational psychology to overcome flaws in individual and collective decision making.
The chief responsibility of a leader comes down to making decisions. An executive’s daily tasks, such as directing operations, devising strategies, and defining culture, require evaluating background information and alternatives, then using these to form decisions in a complex environment. Operational managers likewise make decisions to create the highest possible efficiency in an organization.
Data is a powerful tool for all levels of decision makers, from high-level business strategy to daily operations. The explosion of data in market intelligence has resulted in a proliferation of data scientists that derive insights from data, along with entirely new industries built on data science.
Yet many firms still base their decision-making processes on methods that predate data science. These organizations need a way to leverage data insights to facilitate decision-making processes and solve actual business problems. They also need to confront flaws in individual and organizational decision-making to ensure leaders make the right decisions and the organization puts those decisions into action.
This conundrum resulted in the rise of a new field called decision science, which combines data-driven insights with other fields such as cognitive science and organizational psychology to improve decision-making processes.
The challenge: making the right decision and getting people to carry it out
Modern markets are evolving at a faster pace than ever before, and ambitious organizations are striving to adapt. Today’s decision makers need to make quick, accurate decisions based on reliable data. Yet data science itself often focuses on delivering insights to answer business queries, without taking into account the process that’s required to make the right decision and gets people to carry it out.
The decision-making process is fraught with numerous forms of human fallibility, on both the individual and organizational level:
Individuals have cognitive biases
Our brains evolved to use a system of shortcuts or rules of thumb to allow rapid, nearly unconscious decisions. But these decisions aren’t always accurate or optimal: they’re laden with biases. These include:
- Anchoring: we tend to rely too heavily on the first piece of information we read or hear;
- Survivorship bias: we concentrate on successful programs, people or things while ignoring those that failed;
- Sunk cost bias: we continue programs or endeavors as a result of previously invested resources.
In terms of data science, a human interpreting data may classify it into geographic regions or racial groups when there’s no discernible behavior difference between those classifications. And when it comes to interpreting millions or billions of data points, we’re able to understand basic aggregates, but struggle to think about relationships between variables — information that’s critical to decision making.
Most businesspeople are aware of cognitive biases, but this awareness has done little to counteract the effect of biases in individual decision making.
People don’t act consistently with the best, rational decisions
The way people actually behave doesn’t mesh with the way best practices predict they behave. We’re predictably irrational, as Dan Ariely’s book title puts it. People involved in decision making nearly always have some combination of self-interest, overconfidence, and attachment to the past that clouds their decision.
Decisions involve multiple parties with competing interests
Parties with competing interests often settle on a decision that satisfices the interests of each party, rather than optimizes the best solution while accounting for each party’s parameters. Hence, competition and negotiation strategies have to take a role in the decision-making process.
Conflict after the decision is poisonous
Conflict before the decision is normal and healthy, but parties need to commit to the decision so it can be properly implemented. If they don’t commit to the decision, they’ll either fail to put it into practice, or actively sabotage it if the decision conflicts with their self interest. Decision makers and operations often don’t talk to each other, so even if the organization makes the best decision, implementing it in daily operations is another challenge.
Traditional data science doesn’t take decision making into account
Decision makers rely on data science and other sources of information to determine their moves, but data science doesn’t take decision making into account. Data scientists need to understand their organization’s decision making process and what decision makers require to ensure their insights realize their full potential within the organization.
The data itself presents roadblocks
On top of the human-side intricacies involved in the decision-making process, companies also face challenges relating to data. They have access to vast quantities of complex data, but it’s often in a raw, unstructured format. It lacks business context. It’s irrelevant to the decision at hand.
Furthermore, the most valuable data exists outside the organization in endless sources including blog posts, research papers, patent filings, and press releases. Deriving useful insights from this data requires advanced knowledge of math, finance, and analytics, but it’s hard to find talent with this skill set.
Finally, leveraging data for decisions requires more than data science. As we have seen, it requires integrating data science with the decision-making process so leaders can make the right decision, and the organization can implement it.
Understanding the decision-making process
To solve the dilemmas involved with decision making, it’s important to understand the process involved at each step of a decision.
The levels of strategic decisions
There are three levels of strategic decisions in the modern enterprise:
- Executive-level decisions: These are broad, organization-wide decisions, such as which products or services to offer, and where, when and how to offer them.
- Operational decisions made by mid-level managers: At lower levels in the organization, action is the priority. Decisions focus on how to improve operational efficiency.
- Automated decisions at the individual level: For example, an aptitude test given to prospective employees which uses pre-coded algorithms and hiring criteria.
All three levels utilize insights from data science, but they suffer from the individual and organizational flaws previously discussed.
Sound decisions follow a system
Rational, actionable decisions require a clear process that’s repeatable and dependable. They start with relevant, trustworthy data. They consider creative options with clear values and payoffs. They apply a process of systematic reasoning that circumvents individual or group biases. They consider the business context of the decision, then gain consensus from all parties involved so the decision can be implemented.
Organizations require relevant insights to make decisions
Sound decisions start with understanding that senior decision makers want to make a well-constructed, well-presented decision. In short, they want to make the best decision possible. To do this, they require timely, well-packaged insights they can feed into their decision-making process.
Decisions require collective commitment
Finally, in order to fully implement their decision within the organization, decision makers need to have collective commitment. This requires conflict resolution to address the diversity of opinion, since each decision involves multiple parties, each with their own values and frames.
decision science draws it all together
Forward-thinking firms are reframing data science so it can mesh with the way their executives and managers make decisions. This new approach to data science, known as decision science, allows organizations to systematically improve processes for decision-making, rather than simply solve problems with data.
By doing this, they can attack complex, analytic decision problems, decide on the best alternative, and implement the decision throughout the organization. Decision science goes beyond just getting the right answer. It facilities commitment from all parties by appropriately involving them in the entire process.
Data science vs. decision science
In short, data science enables better decision-making, but decision science completes the puzzle. Data engineers and scientists still have a seat at the table: decision-making is forward-looking, but it’s based on present and past data. It requires applicable and reliable data to base decisions on, as well as a predictive element to understand what will happen in the future. It requires techniques of analysis, along with ways of displaying the insights for decision makers.
There’s a happy feedback loop between data scientists and decision makers. Data science gives decision makers new alternatives, and can reframe the context of the decision using insights gained from market intelligence. Decision makers in turn give feedback to data scientists so they can retrieve more useful data down the road.
Decision science applies at every level of the organization
There are a handful of key moves that differentiate top-performing companies from the rest, per McKinsey. These include disciplined mergers and acquisitions (M&A), reallocating resources to more promising businesses and markets, and increasing gross margin by innovating business-model and developing pricing advantages.
At the operational level, improvements in productivity are another key predictor of standout performance. A recent survey of 1,300 global CEOs indicates 77% say their main focus for driving revenue growth is improving operational efficiencies. These operational managers are looking to automated technologies such as self-serve customer service, internet of things (IoT) monitoring devices and robots to increase efficiency and improve the customer experience.
Each level requires leaders to make serious decisions that determine whether an organization stands out or sinks. M&A’s, for example, entail developing a pipeline of potential targets, conducting due diligence on a smaller number of high value targets, and submitting bids on a handful. And deciding to replace a human customer service agent with a self-service kiosk can lead to higher revenue, but cause significant layoffs.
Managers and executives need a sophisticated, informed decision making process in place to ensure they arrive at the best decision and can carry it out within the organization.
AI is a game changer for decision science
New artificial intelligence (AI) technologies promise to reshape decision analysis in the coming years. It’s important to differentiate between what data and AI mean for decision making.
Data helps provide evidence-based decision making. Before the advent of data collection and analytics tools like Excel, humans had to rely on their own intuition to make decisions. As we have seen, cognitive biases mean these decisions are fraught with error. Data allows decision makers to go beyond intuition to make evidence-based decisions based on data analysis.
AI helps process that data. The internet and the proliferation of personal devices means a vast quantity is generated every second. Leveraging all this data is impossible for humans to do. AI can collect data from hundreds of sources, process that data from its raw form to a structured version, and apply powerful advanced analytics techniques to uncover relationships between data elements. Furthermore, it can help uncover relationships between variables that would otherwise go unnoticed.
But AI won’t replace humans. Instead, it frees humans from the task of processing structured data so they can focus on higher order activities such as developing corporate strategies, values, and vision.
Skai is an advanced analytics platform built to empower business decision makers with real-time, actionable data insights. Our founders, two Israeli military intelligence officers, realized they could apply the same processes they used on the battlefield to make better decisions in the boardroom. We’re unique in our ability to connect more than 13,000 external data sources using proprietary NLP and machine learning techniques. Our clients include many of the world’s leading consumer brands.
*This blog post originally appeared on Signals-Analytics.com. Kenshoo acquired Signals-Analytics in December 2020. Read the press release.