Behavioural Insights, Data Science and Debt Recovery

Advances in behavioural insights and data science are driving innovation in debt collection.

Combining behavioural insights and data science to improve revenue collection is a relatively new practice. As the pressure to do more with less increases, organisations willing to explore the true potential of this emerging field could significantly strengthen their competitive advantage.

Organisations willing to explore the true potential of this emerging field could significantly strengthen their competitive advantage.

It’s behaviour Jim, but not as we know it!

The concept of ‘Behavioural Insights’ has grown in popularity over recent years as an evidence-based method of driving positive change. Commonly associated with the concept of ‘Nudge’, interventions based on behavioural insights often focus on making small adaptations to existing process and practice.

We increased payment rates by 12% by adding to just one extra sentence to a letter

These adaptations are based on insights into human decision-making and behaviour, often drawn from the growing field of behavioural economics. Despite the incremental, cost-effective nature of these adaptations, they have the potential to generate impressive impacts on collection rates.

For instance, some of the headlines from our recent work includes…

• Increased payment rates by 12% by adding to just one extra sentence to a letter
• Increased revenue by 46% by adapting an envelope
• Increased customer engagement by 24% by redesigning a letter
• Saved an organisation ￡1/4 million per year by removing ineffective activities

Are you sitting on a gold mine?

As both data volumes and processing power continue their exponential growth, the potential to more effectively understand, pre-empt and respond to customer behaviour through the application of data science remains relatively untapped.

There's always a drive for more and better data, but many organisations are simply failing to unlock the value of their existing data sets

Whilst there is always a drive for more and better data, many organisations are simply failing to unlock the value of their existing data sets: they are sitting on a gold mine of insight that could be driving more effective collection strategies, tools and practices. For instance, some of the ways existing data sets can drive improved performance include:

• Process Evaluation: understanding the frequency with which activities are executed and then overlaying the cost of those stages/activities and the resultant customer behaviour. If a process has clear high-frequency and/or high-cost activities that don’t have a positive impact on customer behaviour, then we have some clear hotspots for ‘remove or improve’ re-design. We recently saved a client ￡250,000 by identifying activities that were having no demonstrable impact on customer behaviour.
• Behavioural segmentation: identifying statistically significant clusters of customers that manifest similar sorts of behaviours. If there are certain groups of customers who are behaving similarly to each other, but different to other groups, you have a platform for a more targeted and tailored approach to managing those customers. For instance, a group of customers who tend to pay late, but pay eventually, would warrant a very different approach to customers who are repeatedly falling into severe debt. They both fall into debt, but they represent very different behavioural challenges and need to be invested in proportionate to their risk. ‘One size fits all’ is costly and ineffective, behavioural segmentation allows for more intelligent resource allocation and more effective interventions.
• Predictive Modelling: identifying variables that predict customer behaviours and – crucially – the extent to which they predict those behaviours. Statistical analysis allows us to start building predictive models that forecast shifts at the population, segment and individual level. This work can massively improve decision-making at every level of the organisation:
• Leadership teams can respond to pre-empt overarching population and business trends
• Strategy leads can better plan for shifts in demand amongst key customer segments
• Operational teams can deploy more efficient and effective collection and recovery tactics.

Knowing ‘what works’ is no longer enough

Behavioural insights teams are strongly associated with their use of Randomised Controlled Trials (RCTs) to ascertain the effectiveness of adaptations. Widely regarded as a ‘gold standard’ method of building an evidence-base, RCTs (run properly) identify the extent to which customer behaviour can be said to be causally-related to recovery activities.

RCT analyses need to progress beyond knowing what works, to knowing what works for whom .

Traditionally, an RCT will deliver the following sorts of results:

In this example, we tested different sorts of ‘social norming’ message against a control group that received the existing communication. Social Norming is a popular behavioural insights technique that involves informing a customer that most people like them are paying / paying on time.This social pressure increases the likelihood that that customer will subsequently make a payment.

As can be seen from the chart, we found that both types of new message performed better than the existing one to highly statistically significant degree. However, taking these findings at the level of group averages, hides the fact that different customers will respond in different ways. Just because a particular approach works better on average, doesn’t mean that that approach will better with all customers. As such, we need to use RCT analyses to progress beyond knowing what works to knowing what work for whom and then target and tailor our activities accordingly.

The potential for behavioural insights and data science to impact positively on revenue collection is immense and largely untapped. If you’d like to join us on this exciting journey, please get in touch: [email protected]