Become better at digital analysis by understanding your behavioural biases

When it comes to - well, just about everything - we all think we know best

The comedian George Carlin said it best:

“Have you ever noticed that anybody driving slower than you is an idiot, and anyone going faster than you is a maniac?”

Our status and self-esteem is intrinsically linked to the belief that we are right and our perspective on the world the most accurate. Otherwise, how could we justify what we do and what we say? How could we ever form opinions? It serves a purpose, and a useful one at that, but it sometimes works against us.

While it’s true that each one of us is unique, we’re also incredibly alike.So much so that the way we think is predictable and we’re all susceptible to bias, in just about every area of our lives. Advertisers capitalise on this to sell us products and services. They know how to win a customer’s favour and convince them using social proof and other tactics. It’s not witchcraft, just basic psychology.

Our brains are being flooded with information, constantly. Instead of processing everything, we’ve learnt to interpret small amounts of data and draw a conclusion from it using previous assumptions and experiences. These shortcuts are called ‘bias’ and they’re the weak spots that those in advertising play on when trying to appeal to us.

When analysing digital data, this intrinsic susceptibility to bias causes problems. We look at data and form initial judgements, even if the results are inconclusive or the data metrics we’re looking at aren’t particularly useful or meaningful. We then, subconsciously, look for information that supports our initial assumption.

In fact, it’s been shown that people experience a dopamine rush when they stick to their original beliefs. That’s why we’re so drawn to people who seem to think like us, and why echo chambers, where existing beliefs are amplified and reinforced by a community, exist. Our brains love it! But what this causes is a tendency to make a snap decision and then craft a narrative that supports it – after the fact. This way of thinking was useful to our survival when we were living in caves and needed to make speedy decisions in order to stay alive, but harmful when we’re sitting in front of a computer, analysing data in order to make informed decisions.

The tendency to over-simplify digital marketing data, such as website traffic or campaign statistics, to fit an assumption undermines the purpose of analytics, completely. Having more data makes us more confident, not more accurate, and that’s an important distinction to make. We need to accept that any conclusions we draw are interpretations and not facts and, although our brains adore the idea, there is very rarely a single, root cause to an issue that can be cleanly identified and fixed.

So how can we approach data analysis in a way that minimises the impact of our natural biases? Here are some ideas:

1. Select your measures carefully – focus your analysis by choosing your campaign measures in advance, instead of trying to find metrics that support your point of view afterwards.

2. Be equally sceptical about all data – don’t overvalue any data or sources. Remove all preconceptions before starting the work.

3.Check motives and emotions – wipe the slate clean and don’t form any emotions in advance. If you’re feeling frustrated or excited or bored, this might be guiding how you’re interpreting what you’re seeing, so take a break.

This is only the tip of the iceberg when it comes to cognitive bias and data analysis. If you’d like to delve deeper, there are a number of books on the subject, such as ‘The Choice Factory’ by Richard Shotton and Rolf Dobelli’s ‘The Art of Thinking Clearly’.