One of the key characteristics of professional positions in organizations is that professionals have discretion over their choices. Physicians prescribing a course of medication, lawyers interpreting legal opinions, bank officers deciding on loan applications, and project managers deciding on work outputs all exercise substantial discretion.
As a result, their recommendations and observations can vary a great deal from firm to firm, and even within the same firm.
Usually, it is assumed that variability is limited; that substantially similar data, regardless of field, should yield very similar outcomes, whether those outcomes are medical advice, legal advice, loan approvals, or price and time estimates to a client.
A recent study in the Harvard Business Review, however, points out that many studies have shown that decisions and estimates can be different depending on such unrelated factors as weather, time of day, or the events occurring directly before the decision is made. The authors dubbed these variability-affecting factors “noise.”
The variability, moreover, can have negative impacts on business. Clients may jump ship, for example, believing that a recent cost estimate far over previous ones is unfair. Clients may leave doctors or lawyers whose advice seems noticeably different from other practitioners or freely available advice. Estimated project parameters may not be competitive with other firms, and bids fail as a result.
Even if variability does not cause direct business loss, a client’s sense that the information given is substantially different from one time to another can erode trust, relationships, and even brand. The authors term noise an “invisible tax” on businesses.
Reducing the Noise Factor
If unawareness of noise is the problem, then awareness can be a solution. The authors propose that firms conduct a “noise audit,” in which multiple professionals estimate what the outcome of a given scenario should have been. That gives a baseline sense of how much variable there actually is. In the study results, variance ranged as high as 70%.
In addition, variance in nearly all cases was much more than the organization’s leaders estimated it would be.
If there is a great deal of variation, firms might work to develop algorithms to guide judgement. In the medical community, algorithms are often used in combination with established, reviewed practices as a type of decision tree for arriving at recommendations for patients.
As a general rule, algorithms can guide many different types of decisions across various fields. In the approval of bank loans, for example, an algorithm could provide a standard score for various client inputs, with a standard deviation. At the least, the algorithm would decrease variability by reducing noise and in its place, substituting an easily used and transparent tool to facilitate judgement.
More Attention Paid
Noise audits and algorithm development also help counteract one of the pitfalls in all business decision-making, which is falling back on automatic responses and emotional proclivities. Some psychologists term this System One-dominated thinking. System Two thinking, by contrast, is more reasoned and analytic. In this sense, an algorithm serves as a tool to foster more reasoned thinking.
Regardless of the type of business or industry, minimizing variability in the decision-making process is important. Consistency, which is based on a reliable process, will have a positive impact on everything from customers’ trust to the business’ brand and reputation.