Data analytics in the real world

Our world seems to be obsessed with data analytics. However, despite the Gods of big tech telling us there is nothing sexier, nothing more worthy of our praise than data, I disagree. As Christian Lous Lange rightly pointed out, “Technology is a useful servant but a dangerous master”.

Technology is a dangerous master

Perhaps we are getting close to that territory of technology being our master. After all, the thought of not having a mobile phone scares the hell out of most people. Let’s not pretend it’s just for emergencies either because nearly every person I see is holding an Internet enabled, touch screen, smart phone.

Putting data up on a pedestal as a way to predict human behaviour for example, exposes a major flaw. Humans are not rational. For years, psychologists have tried to understand the complexity of human beliefs, emotions and behaviours but we’re still not there.

What’s the problem with data anlytics?

Humans are rarely rational

How can you explain someone who is trying to lose weight who justifies their decision to buy a chocolate bar at a supermarket? Sure there are impulse buys or tricks you can use to attempt to influence the decision making process but is this decision a logical one or emotional? Emotions themselves may prove difficult to identify and measure, using traditional methods.

Ask someone how thay are in Australia and you might hear, “I can’t complain” or “pretty good”. Even if we could measure emotions, they don’t often exist in isolation. Is it possible to feel lust and love at the same time? Is it possible to feel guilt and relief at the same time? Absolutely. Now introduce a third emotion to the mix and find out how that works when mixed with each person’s unique life experience.

People can be manipulated

Now, give someone a stimulant or a relaxant. Fill them with uppers, downers and aphrodesiacs, as we do in the real world. How do each of those substances work in isolation or together to alter a person’s state of mind? What about perfectly legal, non prescription, mood altering substances like sugar, caffeine or alcohol?

Now place impulse buys at the checkout, magazine convers with sexually provocative images on the cover, headlines designed to pull on heart strings or fill the reader with rage. Feed people the content that will stir a reaction based on algorithms designed by people playing God. Do you think all of this external stimulus might play a part in what we believe and how we behave?

Of course it does. After all, we are only human.

Context alters behaviour

People who exist online, exist in the real world too. How people behave when they’re in their parent’s basement, with the computer on and the lights off, is entirely different to how they behave out there in the real world.

We’ve probably all heard the term “keyboard warrior” to describe someone who might have the courage to speak up online but likely to lack that same courage, face-to-face, in the real world.

Another term to explain online behaviour is “armchair experts”. That’s when people have an unqualified opinion about something they have seen, read or heard about but little or no real world experience.

Data analytics has proven much more successful at tracking simple behaviours like mouse movements and clicking. Tracking and predicting the complexities of real world attitudes and behaviour is significantly more complex. The above examples might give you an idea of how altering a single aspect of the situation, like removing consequences, might give people a false sense of confidence.

Causation is hard to prove

We can easily measure a simple behaviour, like a mouse click. What caused that mouse click might be less clear. For example, did someone click because they responded to the picture or the text? Is it possible a cat walked over the mouse? Did the person clicking do it because they had desire for what was advertised or curiousity for why on Earth anyone else would?

Add to that, most data collected in the real world is not done in a controlled environment. How do we know that one factor caused an action, when that factor may co-exist with 1,000 more at the same time? If a person did not reply to an email, is that because they didn’t get it? Perhaps that’s easy to check. What’s less certain is whether they read it, cared about the content or respect the sender enough to reply. Maybe they simply got distracted by their significant other asking them if they put the rubbish out? Without detailed context, who knows?

Assuming we had an infinate capacity to collect and store data. If someone makes an off handed remark in a drunken conversation, do we keep that on record forever? How do we find and retrieve that information to analyse whether that might have caused a certain behaviour 5 years after the fact? How do we differntiate that from the billions of data bits we would need to collect for every minute of every person’s life?

Data analytics need context

Data is just that. Only when we apply meaning to the data by someone interpreting it, we get information. Information is what people (and indeed machines) would need to make a decision. Data analytics is nothing without meaning applied.

But how do we know that the person who performing data analytics has done so correctly? How do we know the person presenting the information has correctly understood the meaning applied to it initially?

Language and culture alone is a filter that could skew the meaning someone applies to something. Often we hear of words in one language that have no direct translation to another. Then, there is a difference in how we apply words that seemingly have the same meaning. The word comprimise has more shame applied to it in some parts of the world than it does in others, like western cultures.

What if we keep only raw data?

You could propose to leave the raw data as is, which itself would open a debate about what context should be applied to the data and what it all means. Each person viewing the data could apply their own unique meaning, which itself would be problematic.

Again, decision making is a mystery psychologists have been trying to solve for centuries. What makes one person swear during an altercation, while another might walk away. Yet, a thrid person may decide to produce a gun and shoot someone. Let’s assume you have all of the data about the person’s childhood upbringing, trauma, failed relationships, successful relationships, current beliefs, peers and more. You still don’t know what a person will decide, given free will.

What does this mean for marketing?

Marketing has traditionally been difficult to measure and thankfully, modern technology means we can measure some aspects of it. For example, on websites and social media, we can see if a person clicks on something and perhaps even if they continue through to purchase.

What is less tangible?

While some asepcts of marketing are easy to meaure, like direct response advertising, others are not. Traditional branding on television, radio, networking, sponsorships or outdoor advertising might be difficult to track, despite out best efforts. But these things still matter because our customers and clients exist in the real world.

If it didn’t matter, Amazon would not spend money branding their delivery boxes and sticky tape, for countless people to see along the supply chain. Major corporations wouldn’t brand their vehciles with logos and corporate colours. McDonalds wouldn’t continue to use every form of advertising there is, online and offline.

Data doesn’t lie but it doesn’t tell the whole story

“Data doesn’t lie” is the mantra so many people, and organisations alike, are adopting. Perhaps it it true that data never lies. I believe it’s also true that data does not tell the whole story. Data should exist to assist decision makers, not take their place. Careful not to rely on data analytics as a false messiah.

Data analytics has come a long way and indeed it has its place in our understanding of the world. Incomplete data, lack of context, limited understanding of causation and indeed the mysteries of the human mind mean we need to respect what makes us human. Data is not the enemy but it’s a long way from being the answer.

Partner Promotions

Leave a Reply