Can an Algorithm Identify Your Customer’s Psychological Traits?

Here’s why data ethnography works. It’s a diverse set of digital footprints that bring a higher level of granularity and variability in observing human behaviour.  This means we can make even more individual-focused insights than qualitative approaches.

They say the greatest test of a relationship happens when you start living together. This is because all the little things that we consider to be ‘normal’ and ‘the only and right way’ may not be ‘normal’ to our partner. But, we are all specific in many ways and we are also more mysterious to ourselves than we should be.

Have you ever wondered why you roll your sleeves the way you do; have your hair cut in that particular style (and not that other, horrible one); show your lower teeth when you smile to someone you fancy, or put all the dirty dishes onto one big tower, rather than scatter them all over the kitchen? 

Some of our little quirks have origins from our childhood, may represent a relic of protective strategies we developed along the way, or may simply represent little cues about our current mental state and even our inner worlds and psychology.

But, these little behaviours are not only limited to our physical worlds. In fact, every day we leave a myriad of digital footprints that can serve almost as our finger prints.

For example, have you ever considered what your ‘interesting’ Spotify playlists may say about you (I’m too afraid to ask!)? 

Some studies reveal that many of our digital footprints are indeed predictive of our personality. In a famous study, an algorithm was used to analyse the psychological traits’ relationships with Facebook likes. Extraverts ‘liked’ beer pong and disorganised people ‘liked’ comedy, as you’d expect; and another study found that, on average, the algorithm could predict personality better than could a family member, friend, workmate or cohabitant.

These small insights could be really useful to companies that would like to personalise their approach to customers based on big data they have at their disposal. But, the irony of big data is that the bigger it gets, the easier it is to forget that there are still people behind the numbers.

However, options of analysing the data don’t have to be so narrow. 

If your data could speak

Enter ‘data ethnography’.

This is a new field of research that uses big data to extract psychology and customer secrets via novel measures and a new way of looking at data.

Here’s why it works: a diverse set of digital footprints (including and not limited to: shopping data, Google searches, digital watch data, biometrics, social media comments and posts, etc) bring a huge level of granularity and variability in observing human behaviour! This means we can make even more individual-focused insights than qualitative approaches. 

For example, a significant work conducted by Kosinski et al. (2012) states that social media can be a source of highly sensitive information such as religious and political views, sexual orientation, intelligence, happiness and others – all of which may be uncovered based on features such as number of likes or other properties of users’ social media profiles. 

Language (as a type of big data of which there is an unprecedented amount of on social media) is in itself a powerful predictor of one’s’ emotional states, psychological traits and behaviours (e.g. if you’re open to new experiences, you probably prefer longer messages and abstract ideas, while conscientious people prefer concise, factual texts).

A great benefit of this analysis is that this kind of data is created in a non-laboratory setting (unlike surveys or focus groups when people know they are being ‘observed’). This also enables us to not only ‘access’ a large number of individuals’ thoughts and emotional states, but also to do so in a scalable manner.

How could it work?

It could be as simple as a ‘sandwich strategy’ – that is, a mixed approach where qualitative analysis of a subsample of big data comes first (in order to extract features) and then big data analysis comes second (where these features are scaled or incorporated in an AI model). For example, we may have a hypothesis that a ‘cool’ person does not use a grammatically correct way of writing (i.e. no capital letters or no commas) and we would, therefore, create a variable that represents the count of capital letters in someone’s text. Could this variable then be used to distinguish between ‘cool’ and ‘uncool’ people across several millions of tweets? Apparently it can! (According to our study, this is because ‘cool’ people choose not to obey any of the traditional rules, including the grammar). 

Data ethnography is a reflection on an idea that big data can provide a valuable insight about ‘shades of meaning’ similarly to qualitative work. Simply said, just as it is possible to obtain an insight about a person based on an uncomfortable laugh during a qualitative interview, it is also possible to better understand a person based on a particular type of greeting cards they buy, late night Amazon purchases, or number of people they follow on Instagram. 

This kind of work is an advocate of creating the space for fast evolving fields to aid the research of the social world with the depth and scope it requires and, ultimately – deserves.

So, when you start a Netflix show – do you watch it until the end (and nothing else, until it is finished). Well, you may have a high need for completion and have a consistent and conscientious personality that sees things through. 

Therefore, when you move in with your partner, make it dead clear that you like your wardrobe to be colour coordinated, dishes clean and grocery shopping done correctly.

And you may just make it as a couple.