This article was first published in the Financial Times on June 21, 2013
http://www.ft.com/cms/s/2/470f1d30-ca12-11e2-8f55-00144feab7de.html#axzz2WsELomvu
Companies must use a combination of data analytics and managerial experience
Is it advisable to let analytics replace human judgment or experience in business decision making? Not really. Given that there are several instances where complete reliance on analytics has resulted in faulty decisions, there is a clear case for business schools to highlight the relevance of human judgment in decision making.
Why has analytics acquired such prominence? Is it because of the vast amount of data that is now available? Or is it because of the ever-increasing computing power at the disposal of organisations? Both factors have contributed.
As complexity in the world of business grew, objective decision making became the need of the hour. Subsequently, several analytical models were developed by academia and industry experts.
For example, an important part of marketing analytics is churn analytics which helps organisations project customer attrition and retention rates. However, the effective application of this model depends on the judgment of the decision maker as well as proper communication within the organisation. This way other stakeholders within the organisation stand to gain from the experience of the decision maker, and the analytical model deployed can be understood holistically across the organisation.
Einstein once said: “Not everything that can be counted counts and not everything that counts can be counted.” The oft-quoted example of financial analytics going wrong before the 2007-08 recession substantiates this. The model was not faulty, but its deployment was. The models used by financial institutions clearly identified the subprime customers. Nevertheless, loans were given to them and the outcome was inevitable. By not paying much heed to what the numbers told them, top management at financial organisations faltered in their judgment and this led to a major global financial meltdown.
It is obvious that in putting all the ducks in a row, one cannot change some of the ducks that err and data can be chosen selectively or even fabricated to support a hypothesis. But if dishonest twisting of numbers is a concern while deploying analytics, rigidity in frameworks is another.
Take the plagiarism-check software used for school students, for example, where wrong implementation without sound judgment by the decision maker can lead to unfair punishment. The software looks for 1phrases with three or more words that are common across submissions. The similarity between submissions could be as innocuous as: “As per this reference . . . ” If two students start a sentence with this phrase, the software would brand them as cheats. Thus, if teachers do not read through all the submissions to elicit the finer nuances and blindly depend on analytics, they could jeopardise the future of their students.
To take quick decisions, managers often rely on real-time analytics. Whether the data comes in real time or not, it is the quality of judgment that is paramount.
From what we know, short-term data and information should not be the basis of critical decisions related to things such as budget reallocation. Since patterns and trends are better judged if studied over a longer period, models that use long-term data are typically better predictors. Thus, prudence demands that managers are cautious about the type of analytical models they use.
Business schools need to teach students that they must go beyond the hype of crunching numbers and understand the business problem first, because numbers may not tell the complete truth. Numbers are a drop in the bucket and will serve their purpose best when they are used in alliance with the depth of a business manager’s judgment and experience.
