Lies, damn lies, and Bigfoot

Lurking in the shadows and the worst nightmares of market researchers across the world is the monster which we call Big Data. We call it Big Data because (a bit like Bigfoot) we don’t really know what it looks like.

Wikipedia provides the following definition:

Big data is an all-encompassing term for any collection of data sets so large and complex that it becomes difficult to process using on-hand data management tools or traditional data processing applications.

The ‘internet of things’, where everything from the household fridge to the heart pacemaker is packed full of sensors and generating vast quantities of data, is changing the ways in which humans and machines interact. The fear is that in the future, organisations will be able somehow to integrate all the data they have from every possible source and use this to measure and predict customer behaviour far more accurately than traditional market research has ever been able to.

Amazon, itself a major user of big data in its retailing and logistics operations, now offers big data consulting and services to other organisations.

Recently, Elon Musk, the billionaire founder of Tesla, suggested that the biggest threat to humanity might come from Artificial Intelligence, perhaps when the machines work out for themselves that the most awkward, unpredictable and irrational operators in any given system are the humans, and consequently decide to wipe us all out.

From my experiences both as a human and as a qualitative researcher, I have a few questions and observations about all of this.

Firstly, we already possess machines which are capable of simultaneously generating, combining, reacting to and interpreting vast quantities of information from multiple data sources in real
time. Critically, these machines are able to select which data sets are important at any point in time and to combine and compare them in order to derive meaning. So let’s compare big data within an organisation, say Amazon, with big data within you and me.

I’m not an expert on Amazon, but I’m assuming it gets data from customers, suppliers, financial services organisations, perhaps even the vehicles it uses via GPS. It uses all of this information in very clever ways. It sends me emails suggesting books on music and science because it’s looked at what I’ve bought in the past. It also suggests that as a good father I buy the latest Jacqueline Wilson book for girls, and gardening books (it should see my back garden!).

In short, Amazon uses big data to make decisions about what it offers customers. It has a limitation in that it can’t distinguish between what the customer buys for him or herself and what is bought for others.

As humans, we’re constantly trying to make sense of the world around us. Whilst all of our sensory equipment is ‘switched on’ all of the time, we can only pay attention to a few inputs at any one time and we can be highly selective if we want to.

We learn and make meaning of the world via the processes of embodied cognition. From an early age we learn about physical balance in stages as we learn to sit up, toddle and ride a bike. But beyond this, balance has become an abstract concept for humans. We use it to describe social negotiation, justice, diet and more. Essentially, we discover short hand methods for conceptualising externally generated big data via metaphors related to our own sensory experiences.

Currently, this is one of the big advantages we have over machines and computers. We have automated systems that prioritise data (e.g. instinct and pain), and we have unconscious systems which learn and recognise patterns in data in ways that we’re only just starting to understand, and take decisions based on these. At the top level we have what we like to call consciousness which provides the illusion of control and favours the linear, the rational and the logical (much like the machines we’ve built in our own image).

Machines are able to beat humans at chess through sheer computational firepower, but they don’t play chess in the way that humans do. The great chess players learn patterns of play to the extent that they are almost ‘instinctively’ able to make the right move in a given situation.

Whenever I receive a qualitative brief, there is typically some mention of the logical and the emotional, as if these were two different channels for decision making. Whilst Daniel Kahneman’s system 1 and system 2 thinking (described in his book ‘Thinking, Fast and Slow’) is useful and insightful, there is a tendency for people to mix metaphors (pattern recognition behaviour again) and say that system 2 (slow) decisions are rational whereas system 1 (fast) decisions are emotional. That word, emotional, has both positive and negative associations. We’re often cautioned against taking emotional decisions or reacting emotionally to a given situation; as if the emotional side of our nature is some wild beast which needs to be tamed. In truth, every decision made within conscious awareness is a blend of conscious ‘rational’ and unconscious ‘emotional’ thinking, and probably more influenced by the latter than the former.

In a recent experiment, respondents were asked to select a property to purchase from a collection of 12 flats with different characteristics. One half of the sample was given time and space to consider which were best. The other half were deliberately distracted whilst completing the task, but ended up making better decisions. It’s tempting to cite this as an example of humans making better use of big data at an unconscious level.

Market research analysis, whether qualitative or quantitative, is centred around pattern recognition and deductive thinking. This is where the ‘insight’ comes from. In quantitative research the patterns are in the numbers, and sometimes we use machines to help us to find these patterns using statistical methods. In qualitative research the patterns are buried within the words, tonality and other communications we receive from a given group of respondents. I often find myself challenged by a client to provide evidence for a particular insight or observation. When this happens, it’s rare that I’m immediately able to recall verbal evidence from a particular transcript or transcripts (although it’s easier if I did the fieldwork myself). I sometimes have to look back into the data to gather evidence, and sometimes I’m surprised by the quantity of evidence available. I’ve obviously picked up some kind of pattern or undercurrent in the data and made meaning from it at a level below my conscious awareness.

There’s no question that big data will continue to complement market research techniques, but there will need to be a step change (a quantum change, perhaps) in the way that machines ‘think’ before big data is able to consign human insight to the deleted items folder.