As insurers seek to refine pricing strategies that boost profitability without losing customer trust, price optimisation remains a hot topic in the InsurTech space. But while it can offer strong margin improvements, it also raises ethical red flags, particularly when applied unfairly across new and existing customers. InsurTech pricing platform Quantee is actively exploring this challenge, developing approaches that align strategic business goals with ethical pricing conduct.
At its core, price optimisation involves adjusting profit margins across an insurance portfolio to maximise overall profitability.
This is done within a set of business constraints, these can be legal, ethical or based on strategic targets. In this context, ethical considerations are central to how insurers differentiate pricing among customers with similar risk profiles.
A common concern in price optimisation is the unequal treatment of customers. The technique often results in charging higher prices to those less sensitive to price increases, while offering lower prices to more price-conscious customers.
While this might seem fair when linked to ability or willingness to pay, it crosses a line when driven by factors like customer trust or inertia.
One such problematic practice is price walking, where renewing customers are offered higher prices than new ones simply because they are less likely to shop around.
This behaviour has been banned in several countries, including the UK, and can erode trust and damage long-term profitability even if it increases short-term gains.
Quantee is now examining how insurers can perform price optimisation under strict ethical constraints that eliminate price walking, while still allowing them to meet distinct sales goals for new and existing business.
Setting boundaries without bias
To avoid price walking, insurers must ensure that their optimisation models do not distinguish between whether a customer is new or existing when setting prices.
However, insurers still need to model retention and conversion separately, as behaviour between the two groups differs significantly.
Quantee highlights two major approaches to solving this:
-
Separate optimisation – Optimise prices for new business and renewals independently, then blend the final prices using regression-based weighting.
-
Joint optimisation – Merge new and existing customer data, suppressing the customer tenure information during price setting, while keeping it available for sales constraints.
Both approaches require a careful balancing act: price recommendations must not be biased by customer status, yet the business must still hit retention and acquisition targets across different customer groups.
Inside separate optimisation approach
This method begins by importing datasets for new and existing business separately and building distinct demand models. Each is optimised with its own constraints—such as desired conversion rates for new business and retention rates for renewals.
The optimised pricing outputs are then combined. This is done by joining the datasets and applying a regression model to determine a midpoint price, weighted by the expected likelihood of sale. At this point, the algorithm no longer knows whether a customer is new or existing.
For example, if a customer profile would be quoted £100 for renewal (with 60% retention) and £90 for new business (with 5% conversion), and the number of prospective clients matches these probabilities, a weighted average is applied to set the final price.
This approach is relatively simple, transparent, and ensures that pricing decisions do not explicitly rely on customer tenure. However, the process of averaging can slightly distort demand constraint targets, requiring careful monitoring.
Exploring joint optimisation
Joint optimisation takes a more integrated approach. New and renewal data are merged into a single dataset, with separate flags for conversion and retention. Importantly, the pricing model is prevented from accessing whether a customer is new or existing—ensuring price walking does not occur.
Even though customer characteristics may differ between new and renewal portfolios, joining the datasets allows the model to operate with a broad view. Separate models are still used for conversion and retention rates, but pricing is set using a single, unified optimisation engine.
Conversion and retention constraints are applied on totals rather than rates. This is necessary because, after merging, rates no longer represent a clean probability of conversion, but a mix of customer likelihood and dataset composition.
This method allows for more flexibility and better accuracy when trying to reach specific business goals. It also enables efficient frontier modelling—helping insurers visualise trade-offs between new sales and retention performance for the same pricing decision.
Comparing performance outcomes
In practice, both methods have merits. When tested on real-world-like data, separate optimisation models performed best when evaluated against their own business segment. However, when comparing total outcomes, joint optimisation proved more effective at hitting overall conversion and retention targets.
Quantee’s testing showed that while joint optimisation sacrifices a small amount of margin on new business, it delivers greater gains in retention and better aligns with business constraints.
This approach also reduces the risk of misalignment during the final price averaging process, making it a more robust long-term solution.
However, it comes with complexity. Joint optimisation pipelines are harder to build and sensitive to inconsistencies between demand models. In contrast, the separate optimisation and averaging approach is more straightforward, easier to explain internally, and quicker to deploy.
Final considerations
Even when not directly engaging in price walking, optimisation algorithms can still result in systematic pricing differences that favour or disadvantage certain groups. For instance, areas with a high proportion of renewals might end up with lower margins compared to regions dominated by new business—depending on how constraints are configured.
The shape and calibration of demand models significantly influence these outcomes. To improve fairness, some suggest separating customer status modelling (new or renewal) from pricing elasticity modelling. This allows for cleaner demand predictions and more ethical pricing outcomes, especially when building from merged datasets.
Quantee’s ongoing research in this space aims to deliver tools that not only optimise profit, but also preserve trust and meet evolving regulatory expectations.
As insurers look to modernise pricing strategies, solutions like Quantee’s are helping bridge the gap between data science and ethics—ensuring long-term growth is built on fair and compliant foundations.
Read the full blog from Quantee here.
Keep up with all the latest FinTech news here
Copyright © 2025 FinTech Global
