By Sanjay Nursing, Insurance Specialist at SAS
Insurers need to prove that they’re taking this responsibility seriously; but those relying on legacy pricing systems may inadvertently be putting compliance and customer trust at risk. We see a lot of insurers struggle with this. Sometimes the issue is a lack of established processes in place to make a good assessment; other times, it’s a lack of customer data. Even in scenarios when the foundational data exists, firms struggle to implement models end-to-end consistently and at scale. And, several years on from the regulations coming into force, many firms still struggle with proving that customers are receiving good outcomes under Consumer Duty.
What’s also important is that under new GDPR rules, consumers have the right to ask for explanations as to how their data has been used - so pressure doesn’t just come from the regulator.
The hidden risks of legacy pricing systems
Pricing is a key area that is subjected to regulatory scrutiny. Traditionally, the industry has relied on Generalised Linear Models (GLMs) for pricing. These models have historically been a strong choice because they are largely based on straightforward mathematical rules and defined risk factors, making GLMs easy to understand and explain to regulators and key stakeholders.
But GLMs don't have to be the enemy. The challenge is knowing when they are the right tool for the job – and when more sophisticated analytical techniques can provide a better understanding of risk and customer outcomes.
Insurance pricing has become increasingly complex. As insurers consider more variables – from vehicle types and driver characteristics to usage patterns and environmental factors – relying on a single modelling approach can make it harder to capture the full picture. More advanced techniques, including machine learning, can help insurers identify patterns and relationships that traditional models may not capture as effectively.
The opportunity, therefore, isn't simply to replace GLMs. It's to give pricing teams access to a broader analytical toolkit, while maintaining the transparency, governance and controls needed to understand and explain the resulting decisions.
Customers that stay should be rewarded, not punished
Legacy pricing systems have also tended to penalise customer loyalty through ‘price walking’ - where long-term customers face compounding yearly price hikes.
Price walking practices were banned by the FCA back in 2022, but staying compliant requires constant policy checks over time. Insurers must ensure that any promotional deals offered to new customers don’t conflict with prices for existing clients; they can quickly find themselves in hot water if it turns out loyal customers are being charged more.
Rather than relying on manual checks or rigid rules to determine whether these requirements are being met, machine learning and more ‘dynamic’ pricing solutions can help insurers stay compliant with FCA obligations while ensuring customers are rewarded for staying with them.
Moving away from rigid rules to real-time pricing
To help mitigate this additional risk, one option insurers have where pricing is concerned is to move away from rigid, legacy pricing systems like GLMs towards more ‘dynamic’ or real-time pricing options. These combine classic methods like GLMs with modern solutions - like machine learning - to provide more competitive quotes in real-time without losing out on explainability or transparency.
Well-governed, machine learning solutions help to support more personalised and accurate quotes for customers. Rather than assigning blanket risks or values based on a postcode, real-time pricing can safely analyse a number of granular individual variables, for example: Do you live on a main street or a cul-de-sac? What is the flood risk? Are there any schools nearby?
This also means looking more closely at the customer - assessing if they are elderly, vulnerable, likely to miss payments, or financially literate. By demonstrating greater customer understanding and awareness, policyholders across different demographics and vulnerability levels can be treated equitably rather than penalised.
Historically, machine learning use within insurance has raised concerns regarding bias and discrimination, emphasising the need for proper governance controls to be in place to ensure automated decisions are fair, compliant and fully auditable. That includes things like regularly checking training data, testing model outputs against different demographics, keeping humans in the loop, and monitoring for drift.
Not only does a well-governed pricing model mean improved accuracy and reduced risk for the insurer - it also rewards loyalty to help protect your customer base. Customers no longer have to threaten to leave just to get an accurate insurance quote. A specific - and hopefully better quote - becomes the norm, and renewing with the insurance company becomes an easy choice.
By operating at a more granular level, insurance processes become fairer, traceable, and defensible - protecting both the customer and the bottom line.
We’ve also noticed that this model brings strong benefits for the insurer in terms of building deeper customer relationships.
The same data foundation can also help insurers develop a deeper understanding of customer needs beyond the initial pricing decision. By bringing together relevant customer, product and behavioural information within appropriate governance frameworks, insurers can identify where products and coverage may be genuinely relevant – while ensuring that any use of customer data remains transparent and appropriate.
Achieving fairer outcomes
FCA regulations aren’t about mandating profit margins or setting fixed pricing. Instead, they require firms to evidence that their customers are paying a fair price in comparison to what they receive. Justifying pricing on complex modern insurance premiums with rigid, legacy systems that rely heavily on assumptions is unlikely to hold up to regulatory scrutiny.
Fulfilling customer fairness duties requires moving away from outdated, batch-calculated GLMs. By adopting modern, data-driven pricing architectures, insurers can transform compliance from a regulatory hurdle into a competitive advantage.
|
