To effectively engage in data-driven decision making we must confront the five biases of business un-intelligence . Until the day when future BI technologies can help mitigate cognitive bias, big decisions call for a collaborative effort that harnesses data and intuition.
The top reason that the highest performing organisations had for implementing BI, according to the 2014 annual Business Intelligence Gleansight Benchmark report, is executive level demand for data-driven decisions (92%). Yet unreasonable biases can impede both analysis and intuition, especially when we're making sense of big data, or tapping into the power of business intelligence.
Dear executives, if you think "data-driven" means that bias will no longer influence your decisions, think again. The best of us will double check our intuitions by accessing on-demand BI dashboards to consider the implications of every twist and turn in our business environment. We realise that bias can still skew our data-driven decisions, no matter how thoroughly our data has been cleansed and presented. It is time to stand up against these five cognitive biases and overcome data denial.
Five biases to overcomeThe mind is poor at recognising its own biases. There's a blind spot that makes us assume our view is the objective one, omitting the data we must see. In our community on Google+, Jan Wylie recommended that we check out an article from the SiliconAngle which summarises the biases that Dr. Michael Shermer, the quoted expert, suggests can affect us when we sort through data. Some additional insights from other sources have been included in these five common errors of bias, listed below.
1. Confirmation bias: Even when big data is available, the confirmation bias can cause us to overlook evidence that does not fit in with the story that we want to tell. We fail to ask the questions worth asking and search for data that confirms what we want to be true. Shermer warns, “You end up thinking you are being unbiased and that the facts speak for themselves, but, of course, they don’t.” Chip Heath, writes in his book Decisive: How to Make Better Choices in Life and Work:
"Confirmation bias is probably the single biggest problem in business, because even the most sophisticated people get it wrong. People go out and they're collecting the data, and they don't realize they're cooking the books.”2. Hindsight bias: When an unexpected event happens, we tend to look back as if we had predicted it all along. Shermer says, “That can lead you to reinterpret information to make it fit what actually took place. That can be misleading." David Snowden and Mary E. Boone, researchers of managing in complexity, explained in a 2007 HBR article that while best practices work just fine in simple contexts, difficulties arise when the the business environment shifts into complexity. They say that in such contexts, "hindsight does not lead to foresight because the external conditions and systems constantly change." In other words, what went wrong yesterday may have little bearing on what will happen tomorrow.
3. Correlation vs. Causation: It's often critical to find the root cause of a business issue, but misleading correlations interfere with finding the cause. If you run an inappropriate estimation technique, correlations will inevitably appear: per capita consumption of cheese (US), for example, correlates with the number of people who have died by becoming tangled in their bedsheets. With big enough data, many spurious correlations spurious correlations can be found, but no matter how strong, these do not imply causation. Nevertheless, some correlations are potentially significant reasons to drill-down to unearth the likely causes of an effect.
4. Too Much Data: Being overloaded with data is a big part of the dilemma, particularly when your team lacks the skills of a data scientist who understands the value, or lack thereof, of the myriad sources of data at your disposal. As Shermer points out, "You end up with a deluge of things you could study, but which of them really matter?” Many organisations choose to outsource part of the challenge to an expert team of data scientists who understand the business well enough to provide customised dashboards that reduce information overload while offering pertinent views that are relevant to each manager or department.
5. The Wrong “Big Question”: Look out, we may ask the wrong question of our data. Worse, our efforts at data mining may miss the goal we had in mind because the question is quite simply, amiss. A question that quickly arises from intuition may do the trick, but to get to the best possible question, the analytical mind is required to make the most of business intelligence. In the book Thinking, Fast and Slow, Daniel Kahneman, the renowned psychologist and winner of the Nobel Prize in Economics, breaks down two systems in thinking that we access in our daily business affairs.
System 1 operates automatically and quickly, with little or no effort and no sense of voluntary control. System 2 allocates attention to the effortful mental activities that demand it, including complex computations.The questions we can ordinarily answer through intuition are System 1, whilst the bigger questions may take time to mull over by accessing System 2. Analytical contemplations can truly challenge biased decision making, but without tools at our disposal we may fail to see the big question that can help us meet our objectives.
Thinking in this way, with business intelligence by our side, can accelerate analytical inquiries, as well as sudden intuitive insights.
Collaborate in data-driven decisionsComing up with a promising question may start with the CEO, but with collaborative BI dashboards, insights may arise from managers and analysts who see a business challenge from their own unique perspectives. Collaboration can be useful in overcoming bias, but only if the team has diverse members who can share alternative perspectives that executives are willing to hear. Otherwise, the five biases listed above can spread among all members of the team like a fog, a familiar phenomenon known as group think.
Collaboration in decision making works best when the organisation's culture is both diverse and data-driven. In the WSJ's CIO Journal, an article by researchers Foster Provost and Tom Fawcett titled, Data Science and its Relationship to Big Data and Data-Driven Decision Making, provides a definition of data-driven decision making as “the practice of basing decisions on the analysis of data rather than purely on intuition.” Phil Simon, the author of Too Big to Ignore: The Business Case for Big Data, agrees and is quoted in a Forbes article:
Big Data has not, at least not yet, replaced intuition; the latter merely complements the former. The relationship between the two is a continuum, not a binary.” [...] “When used right,” says Simon, “Big Data can reduce uncertainty, not eliminate it. We can know more about previously unknowable things. This helps us make better predictions and better business decisions.” But it doesn’t mean we can rely solely on the numbers without injecting our knowledge, perspective, and (yes) intuition.
When intuition and analysis danceConsider the day when intuition can flow with the facts in relative harmony. Business intelligence tools of tomorrow should support this possibility. Embedded business analytics could shift with changing contexts, from simple to complex. New forms of artificial intelligence could suggest common questions to help us better mitigate cognitive biases. We may see beyond the immediate impulses of faulty intuitions, as well as those skewed forms of analysis that give big data a bad wrap.
As of yet, there is simply no coherent, comprehensive theory or practice of cognitive bias mitigation. The day may still come, however, when the truth that lies before us could more readily be discerned with the help of predictive analytics and automated assistants that provide us with alerts, keeping us informed of what we need to know, when we need to know it.
But, no matter how seemingly perfect future BI technologies become, we must be diligent in questioning our own minds and motives, alongside the data that we interpret, especially if machines already interpret the data for us. Appropriately engineering the cognitive models and algorithmic heuristics to process our data meaningfully may be hard, but we refuse to say it's impossible.
In the words of David Pidsley, Data Strategist at Cause Analytics, "let's un-invent business un-intelligence." Please join our Anticipate Change community and for more on this topic read our post about Decisions in the Age of Abundant Data.