The data dilemma: personality profiling in lending and insurance

July 11, 2018

Social media is on the march in the Middle East. The 2017 Arab Youth Survey found that Facebook was the most popular way for 18-24 year-olds to get their news.  Research by Hootsuite shows that more than a third of people in the Middle East are active social media users, with 64% of 18-24 year-olds sharing stories on Facebook.

Being active on social media has its rewards, but it also carries risks, as shown by the Cambridge Analytica scandal. What soon became clear was not just that data had been used without permission, but that businesses could use Facebook and other platforms to build detailed personality profiles and use them to influence people’s behaviour.

So how do organisations learn about individuals and what does it mean for the future use of personality profiling by insurers and lenders?

How do businesses source data?

Personal data is big business. The Internet Advertising Bureau estimates it generates €100 billion of marketing spend in the US and Europe.

The data industry consists of an often hidden network of organisations that collect, generate and supply information about individuals and their digital behaviour. Web trackers, data brokers, marketers and social media platforms are constantly capturing, buying or selling data about individuals.

This information can come from a range of openly available government sources such as electoral and land registries, telecoms listings and census data. It is then supplemented by web tracking technology that collects data about the sites you visit, your search history, what you buy and where you spend your time online. Mobile phones are powerful data capture tools with apps recording your activity, location and behaviour. Emails are also tracked using hidden lines of code.

Me and My Shadow, a pro-privacy campaign group, has developed a tool that claims to show how visits to a range of popular sites are tracked and the data shared. Try it for yourself and see how a global network of connections criss-crosses the globe as your online actions are recorded and passed between businesses.

Social media, surveys and personality profiles

Not only do social media platforms such as Facebook, Twitter and Instagram record your digital behaviour, they can also allow third parties to see who you’re friends with, what you’re interested in and the sort of updates and images you post.

And platforms such as Facebook don’t just show information about what someone likes and who they follow, they can also host a range of tools designed to harvest data. Personality quizzes, surveys and competitions are highly effective ways to build detailed pictures of an individual.

Such psychographic data allows profilers to classify individuals using five personality characteristics including introversion, conscientiousness and emotional stability. These profiles are used by businesses to ensure that the most relevant goods or services are advertised in the most effective ways.

Psychographic profiles become even more valuable when combined with an individual’s demographic information and their digital behavioural data.

Profiling and financial services

Insurers, lenders and other financial providers use a variety of data to identify potential customers and decide whether to offer insurance or loans. Credit agency information is traditionally at the heart of such decisions but businesses are looking beyond this to make better-informed judgements.

Admiral, the UK motor insurer, caused controversy in 2016 when it unveiled a new product,, that used analysis of Facebook posts to create a personality and risk profile for young drivers. Facebook blocked the product for breaking their rules so Admiral replaced analysis of posts with a personality quiz that led to the creation of a psychographic profile.

This information could then be used to assess risk: for example someone judged to be conscientious and an introvert might be offered a lower premium than an extrovert with a low agreeableness rating.

The quiz was reportedly created by Visual DNA who are very clear about how their profiling can be used:

“Our model allows us to reliably predict a customer’s ability to repay a loan. This is revolutionising banking. Now financial institutions can lend to customers based on their personality, even to people without credit histories.”

Such profiling is not without its risks, as Emma Eaton, senior associate at the law firm Clifford Chance explained in an insurance industry briefing. Focusing on GDPR legislation in EU countries, she stated that “just because an insurer believes it has a legitimate business interest in collecting personal data and the technology to automatically profile and make automated decisions is available, it does not mean [these] activities can be undertaken with impunity.”

Eaton added that insurers must assess whether profiling may create discrimination among data subjects and “explain the nature of the data collected, the logic of the profiling/automated decision-making that it is relying on, and the significance and envisaged consequences of the processing of the individual’s data.”

Total profiling

Such profiling is being taken to extreme measures in China where the government’s social credit score scheme will lead to the country’s 1.3 billion citizens being judged not just on their ability to repay loans, but on who they like on social media and whether they act in a manner that fits the state’s idea of good citizenship. A high score earns benefits such as travel visas, loans and car hire without deposits.

Huge tech businesses such as Alibaba and Tencent have government approval to use their vast banks of customer data to develop these rating systems. The scheme has already been rolled out on a voluntary basis and compulsory participation is scheduled for 2020.

The power of the algorithm

Such collection and use of personal data may seem like a massive invasion of privacy, especially to those used to western democracies. However the fact that millions of Chinese people have already signed up is proof that people will sacrifice their privacy for the right rewards. Motor insurance customers worldwide are also allowing themselves to be tracked by phone apps while they drive in order to secure lower premiums.

Or take the case of Lenddo, a fintech business that creates credit scores for millions of people in 15 countries including India, Mexico and Peru. These are customers it describes as ‘the emerging middle class’ who do not have traditional credit histories but do have mobile phones. Customers grant Lenddo one-off access to all their digital data, including geolocation, browsing history and social networks.

This information is passed through an algorithm to create a score that is ‘highly predictive of an individual’s credit worthiness’. (Learn more about providing finance in the developing world in this accompanying article.)

Careful expansion

The use of social media and personality profiling in finance and insurance is in a state of flux. The technical capabilities are racing ahead while governments struggle to catch up with legislation such as GDPR in the EU.

The fact that it was only in June 2018 that Apple banned app developers from harvesting details of a user’s contacts shows how even the biggest tech businesses are struggling to protect customers. Meanwhile Mark Zuckerberg has appeared before Congress and the European Parliament to explain how Facebook would prevent the further abuse of personal data.

The Facebook scandal has alerted the media to the scale and potential risks of profiling. Public attitudes are also changing as individuals become more aware of how data can be both used and abused. The evidence is clear, however, that individuals will continue to sign away their rights to data privacy as long as the incentive is attractive enough.

The challenge for insurers and lenders is to strike a balance between technological possibility, market opportunity, legislation and public perception.

Get it wrong and a brand can quickly become toxic, as the directors of Cambridge Analytica discovered to their cost.