Cognitive Biases That Can Wreck Your Analytical Efforts

I recently tweeted a link to an infographic on Life Hacker (link) highlighting 20 cognitive biases which can screw up your decision making.  It was striking how many of these are evident when organisations seek to enhance their analytics capabilities.  So here are my top six cognitive biases which can derail an organisation's analytics efforts, and how to tackle them.

1: Confirmation Bias
This is a widespread and pernicious bias.  It is when we seek or only listen to information which confirms our preconceptions or, at worst, our misconceptions.  As Raymond Nickerson, the American psychology professor, put it:

"If one were to attempt to identify a single problematic aspect of human reasoning that deserves attention above all others, the confirmation bias would have to be among the candidates for consideration"

You can hear it all the time in interactions with analysts and data scientists: the digital marketer who demands a report which proves digital media is better than "traditional" media to support a transfer of budget to digital; or a marketing director seeking 'proof' that it is declining customer service not marketing performance behind a drop in sales.

'Data science' can be perceived as a faddish term but I don't subscribe to this view.  Science progresses by seeking antithetical evidence to current theories. Data scientists should resist the temptation to exacerbate others' cognitive biases and leaders should ask bigger and more open questions.  The result should be smarter opinions, new insights and ultimately competitive advantage and business value.

2: Conservatism Bias
Conservatism bias is the tendency to favour prior over new evidence and therefore to hold onto old opinions even in the face of compelling new insights.  It's often another hurdle to overcome after confirmation bias.

Again, it is very common in my own speciality, marketing analytics.  A more enlightened leader requests or permits some new analysis which breaks with conventional beliefs in the organisation.  Once shared internally, it is beaten to death by a host of old tests and analysis which support the widely-held legacy beliefs.  I have seen organisations dredge up ancient data and reports in defence of inefficient marketing  activities.  This evidence was perfectly valid five years ago, but the market and consumers had moved on.

An effective way to counter an organisation's tendency to fall into conservative biases is to foster a culture of continuous test and learn, and establish a cross-functional approach to designing, running and analysing your tests.

3: Pro-innovation Bias
One of the challenges with actively addressing cognitive biases is that fixing one type can lead to others.  Those suffering from pro-innovation bias will champion a new idea at all costs and be blind to the limitations and weaknesses inherent in it.  "This $50,000 test we ran on marketing through Snapchat delivered huge returns - it's the future - we should redirect millions more into it!".  Don't let your desire to address conservatism bias slip into pro-innovation bias.  Adopting a holistic, cross-functional approach to analytics can certainly reduce the risks here.

4: Stereotyping
A stereotype is an exaggerated belief or distorted truth about a person or group that allows for little or no individual differences or social variation.  That definition may sound familiar to many marketers as its the basis for lazy customer segmentation.  A simple, static segmentation strategy may be a convenient first step, but it ignores a wealth of granular consumer level data and insight and can lead to dull, unoriginal customer experiences.

There are myriad tools and techniques available which enable organisations to advance from simple, stereotypical customer segments using “raw” behaviour, demographics or attitudinal data.  The holy grail should be to achieve relevance at scale for your customers through analytics which adapts as your customers needs change.

5: Information Bias
Information bias is the assumption that all information is useful when trying to make a decision and that more is always better.  In reality, the reverse is often the case, with additional information adding no value and only serving to complicate and confuse decision making.

This is an area where data scientists and analysts have an important role to play, distilling vast amounts of data and analytics into compelling, visual presentations and stories and in designing uncluttered dashboards and tools for decision makers which enable guided data discovery and insights.

6: Bandwagon Effect
Bandwagon effect bias is when we do something primarily because other people are doing it, regardless of our beliefs and knowledge.

One way to fall foul of the bandwagon effect is to allow your organisation to adopt simple strategies which are not underpinned with solid data and analytical evidence.  For example, a retailer whose pricing strategy is to quickly match competitors without regard for underlying consumer price sensitivity, or the brand who allows their media agency to convince them that maintaining share of voice against competitors in all channels is necessary.

Of all the biases highlighted here, the bandwagon effect is probably the easist to identify and counteract through the pervasive use of analytics to challenge and then underpin improved business strategy.

Addressing cognitive biases in your organisation is vital if you wish to fully benefit from the use of advanced analytics.  Failure to do so will limit your organisation's ability to embed analytics in the decision making process.  The extent to which analytics is embedded in an organisation's decision making is a key factor which separates low from high performers.  For more insights on this, I recommend research we ran with MIT titled "Winning with Analytics" (link).  If you don't identify and address cognitive biases you run the risk of your analytics efforts being expensive and academic exercises.  Truly addressing these biases in your organisation could have far-reaching implications for how you organise and your ability to transform through analytics.  You can start small though by being alert to cognitive biases in all their forms and having the courage to call them out when you see them.