Huge amounts of data exist on customers and social media users, how useful is this in predicting choices around health, wealth and politics?
Available data on audience behaviour and preferences has grown tremendously in the digital era. From cellular geolocation data to fitness tracking to browsing history, firms know more about what their customers do with their products than ever before. However, the size and complexity of these data make it easy for firms to oversimplify observed patterns and mistake correlation for causation.
Understanding of economic and behavioural theories related to consumer behaviour and choice can help to build actionable understanding of customers for business leaders. Detailed customer profiles, detecting inflection points in customer profitability paths and determining life-cycle value are all part of strategies to retain and recruit customers.
The data trail left online, particularly expressed preferences, can reveal a great deal about an audience. Casual relationships can be identified in structured and unstructured datasets and can be immensely profitable in the right hands. Understanding customers in buyer-seller relationships is one part of the story and through customer loyalty schemes and online shopping, retailers have all the data they need.
Purchasing patterns, ‘likes’ and memberships of groups is used to track a range of attitudes and intentions but predominantly for advertising targeting. Audience characterisation in buyer-seller relationships has moved on from inferences made through identification of traditional demographics. But what about other relationships?
“Next generation audience analytics extend beyond demographics to reveal who we really are, our visceral reactions to information and other stimuli that are vital in shaping better choices and behavioural change pathways.”Andrew Roberts
Demographics answer an important question, “who are you?” Who you are remains important to campaigns of all kinds. The most basic use would be to segment audiences on three general factors—age, gender, and socio-economic position (education level, income, work hours, household size and residential area). Campaigns would target different audience segments with messages relevant to them. Data can also be used to predict patterns of decision-making and reaction to information or other influences. The target and prediction uses are distinctive, the latter is of far more value when shaping attitudes, intentions and ultimately behaviours. However these general demographics are limited, individual differences can override age and gender assumptions and more data is needed if we are to explore deviations from core economic theories to really understand how to intervene to shape audience decisions and behaviour.
Two next generation analytical areas relating to audience subcultures have advanced greatly through big data analytics:
- Political ideaology
- Cognitive reflection or deliberation
Through the wealth of online data, practitioners are able to make advances in characterisation of political ideology. Liberal or conservative has inherent value in political responsiveness and identity defensiveness, but it can also be applied to provide insight into ‘in group’ values also known as cultural subgroups. These have been shown to be extremely accurate in predicting audience responses to a wide range of societal issues, such as climate change and vaccination. They can also provide data on likely predispositions and how groups will react to information provided. This is valuable in guiding development of messages around health for example, but also in shaping campaigns around new technologies. For far too long campaign managers have believed such campaigns to be binary or zero sum. Either their message worked, or it had no effect. We now know of course that messages may not only be ineffective, they can have unintended or completely the opposite effect.
Data that will help segment audiences on cognitive reflection are helpful for assessing time preferences, risk preferences, probability weighting, ambiguity aversion, endowment effects, and anchoring. All have significant influences on judgement and decision making and span 4 nodes of the Reciprocom behavioural nexus (cognitive psychology, trust, risk, and culture/values). In practical terms, this data helps us look at audience ability (to process a message) and motivation to do so. These are important variables in how audience process information—is it relevant to them, does the content appeal, especially when interest is low.
Two data sets can provide direct and proxy data to give us deeper understanding of an audience, without having to perform in depth formative research, which takes time and money. Not only a quicker way to reach the intervention stage, a more effective way of making an intervention in complex areas of behaviour where often, there are significant barriers to change (predispositions, structure or environmental barriers).
Data collection and processing have advanced through our interactions being recorded, particularly online. They may or may not define, who you are. Demographic data remains a key tool in targeting and prediction, but we need to go far beyond general audience factors to make effective interventions to shape changes in habitual behaviour related to health, wealth and happiness. In key areas of such, opinions and predispositions are so engrained that efforts to provide general information or one-size fits all interventions are not only ineffective, they will contribute towards further divisions and polarisation in society on issues of great importance.
Insight will continue to explore audience characterisation as as a key component of designing and deploying behaviour change interventions in critical areas of lifestyle and well-being.