A Marketer’s Mac & Cheese: Data Science and Behavioural Science
Vanja Ljevar, Chief Data Scientist at Kubik Intelligence, explores the convergence of behavioural science and data science and discusses how brands should use data more intelligently.
Topics
What to Read Next
- Teads Partners with XPLN.AI to Advance Attention Measurement
- Factory Berlin, All Voice AI Launch Real-Time Voice Monetisation Platform
- Continua Adds Amazon, Google, Instagram, & TikTok Shopping to Social AI
- Airtory, CultureSync Media Partner to Deliver Rich Media Advertising
- VuePlanner Integrates Sundogs’ Creative Insights into CreativeVue
The tin man famously lacked a heart – but what if big data and AI can supercharge intuition? By combining behavioural science, data science, and generative AI, it is now possible to create targeted communications and user experiences that speak to people on the deepest level, understanding them perhaps even better than they understand themselves, says Vanja Ljevar, Chief Data Scientist at Kubik Intelligence.
Examining one’s history of transactions, browsing activity, and social media posts can unveil a wealth of insights into our inner thoughts and behaviours. However, according to Ljevar, many companies either do not gather this data or struggle to harness its full potential.
Martechvibe spoke to the data scientist about how behavioural science and data science together concoct the best analytics to understand customers better. “Consumers are so-called ‘cognitive misers’, with limited brainpower for paying attention to things, and have to rely on subconscious shortcuts instead. This is where targeted messaging comes into play.” She discusses how brands should use data more intelligently, and how to turn data points into psychology insights.
Excerpts from the interview:
Can we talk about the difference between zero and first party data – which is more trustworthy?
Both zero and first party data can be incredibly powerful, each with its respective strengths and weaknesses.
The great thing about zero party data is that it enables us to ask specific questions directly addressing business challenges. In essence, zero party data can provide invaluable insights into customer wants and needs, which can be leveraged to provide a more personalised experience. Furthermore, it also enables us to dig deeper, especially when it comes to asking about customers’ psychological characteristics (which can then be used in creating optimised targeting) – and this is the kind of information that is otherwise not available on an individual level. There is a valid argument that zero party data is, therefore, superior to any other data source.
However, this kind of data is not without its challenges – and they are usually connected to various biases and absence of representative samples. Some people are not very open in expressing their true opinions and beliefs, often due to social desirability of responses and demand characteristics. For example, people would never openly admit they feel stigmatised. They are aware that their responses are monitored and therefore, surveys represent a laboratory setting where people are ‘put under a microscope’.
First-party data (often big data) is typically secondary. However, it still enables analysts, given careful ethical forethought, to gather information from people who are less open to be a part of a survey or are more convenient participants for other biasing reasons. Of course, a significant challenge inherent in big data lies in the subsequent data cleaning process. All customer-related systems collect some customer data. The challenge stems from the fact that they all gather, store, and manage data differently, leading to inaccurate and inconsistent data between systems. Clean data is vital for a good analysis. However, in the case of big data, it is possible that data stored in various formats can be invalid in some way. Simply said, many companies amass vast amounts of first party data. But, just because data is there, does not necessarily mean it will be useful.
Which one of these data sources is more trustworthy – depends on what kind of a question we have in mind. Finally, it is worth noting that it still requires the expertise of a researcher (analyst) to responsibly distinguish between which data could and should be converted into information and which data cannot.
How does the mix of data science and behavioural science help brands understand their customers better?
Data science and behavioural science are like mac and cheese — one goes really well with the other. Data science can provide valuable insights about customers, using both descriptive and predictive analytics to intelligently extract data-derived behavioural insights. This serves as a basis for psychology-based nudging techniques that resonate most with customers and give companies’ ultimate competitive edge.
At Kubik Intelligence, we start with data: We help our clients understand why certain patterns in data exist and what they say about their customers. The tools and methods we use are rooted in peer-reviewed, cutting-edge, specifically-tailored research that brings killer insights, usually unavailable to off-the-shelf products and services. In their own right, these types of insights can illuminate gaps and opportunities within a company’s data and can emphasise ‘quick wins’ for the business.
The next step is psychology: It is often possible to infer psychological features from customer data, however the most reliable approach is to create a psychometric survey, which is then connected with customer behavioural data (from transactions, etc.) – so that we can segment the audience and extract underlying groups of customers. We can create recommendations based on behavioural science about who to target and how. This means our clients can curate experiences tailored to each customer’s unique preferences and needs – fostering deeper emotional connections with their audiences and of course, driving loyalty, conversion and sales.
How does behavioural data help campaigns on a population and individual level?
The pioneering field called data ethnography points out it is almost possible to blur the lines between the population and individual level-data. Thanks to a diverse set of digital footprints, social scientists have never had such a level of granularity and variability in observing human behaviour, which, depending on the research objective, may create even more precise, individual-focused insights than traditional qualitative approaches. This means that we can target groups in our population with such a precision as if we were speaking to individuals.
For example, one research paper states that social media can be used to infer information such as psychology, political views, happiness and others – all of which may be uncovered based on features like the number or likes or other properties of users’ Facebook profile. These behavioural features based on digital footprints enable us to gather a large number of individuals’ thoughts and emotional states, but also to do so in a scalable manner – while, of course, respecting privacy and ethical concerns.
This is invaluable for marketing campaigns: By utilising these data insights, businesses can focus on creating interactions and communication that resonate on a deeper, personal level. This leads to a higher sense of loyalty and creates a feeling among customers that they are genuinely understood and cared for.
What advice would you give marketers and data scientists to speak a common language?
There are sometimes challenges in finding a lingua franca and these can be solved by following the next steps:
It is crucial to start with a mutual goal and a set of agreed questions. This is extremely important for data scientists as it enables them to determine the most appropriate data source and analytical technique they should use. This enables them to decide on the optimal reporting and visualisation style.
Knowledge sharing is essential. This starts with establishing a data culture and regular meetings that break the invisible boundaries between departments. More frequently than not, marketing departments rely on data that is often in silos; data scientists can help with removing these obstacles, but only if they are aware of them. This may involve using a common data platform or integrating marketing automation tools with data analysis tools.
Finally, it is important to establish an honest feedback loop, where both marketers and analysts are encouraged to share their views.
