Next best action in insurance: how to move from intuition to data-driven decisions

What you should know:

• NBA brings structure to decision-making across the entire organization. It provides a consistent decisioning mechanism regardless of channel, line of business, or team, which translates into greater predictability of retention, premium growth, and operating costs.
• The fastest return on investment appears where decision volume is the highest—most notably in renewals and claims handling. These areas represent the most natural starting point for NBA initiatives.
• NBA does not require replacing the core system, but it does require organizational readiness. To work effectively, insurers need a basic level of data hygiene, clearly defined decision rules, and the ability to embed recommendations into CRM systems, agent portals, and operational workflows.

NBA as a new approach to decision-making in insurance

Over the past several years, the insurance sector has invested heavily in data, automation, and tools that support operations. Despite this, day-to-day work in renewals, agency sales, and claims handling is still largely driven by simple rules, fragmented decisions, and an incomplete view of the customer. As a result, organizations fail to fully leverage the data they already have, struggle with inconsistent processes, and face growing pressure on financial performance.

In this context, Next Best Action (NBA) emerges as an approach that fundamentally changes how decisions are made. This is true provided it is not treated as a one-off initiative, but as an organizational capability that enables the company to operate in real time and make more accurate decisions based on data, business logic, and a consistent approach to customer engagement. 

NBA does not replace people. Instead, it augments them by delivering up-to-date context and actionable recommendations at the right moment. 

Why next best action is becoming a strategic priority

For executive teams, this topic now has clear strategic relevance. Decisions made across hundreds of thousands of customer touchpoints directly affect retention, customer lifetime value, cost to serve, NPS, and overall operational efficiency.

In its latest Hype Cycle reports, Gartner classifies Decision Intelligence as a transformational technology, emphasizing that organizations are increasingly moving away from descriptive where processes are complex, channels are multi-layered, and customer expectations continue to rise—this shift is particularly significant. Decision Intelligence forms the foundation of Next Best Action solutions by enabling a consistent, re analytics toward automated, model-driven decision processes. In insurance— peatable, and scalable real-time decisioning mechanism built on a complete view of the customer.^1,2

As a result, NBA is becoming less of a competitive differentiator and more of an operational necessity. The critical question is how to implement it in a way that delivers measurable business outcomes without destabilizing the existing architecture or increasing operational burden.

Why NBA matters for insurers right now

The insurance sector is at a turning point. Marginal product differentiation, cost pressure, and rising customer expectations mean that competitive advantage no longer comes from the offer itself, but from the quality of decisions made about customers—across every channel and at every stage of the journey.

That is why Next Best Action should be viewed as a strategic tool addressing three critical industry challenges.

1. the fight for retention

Customers increasingly see little real difference between offers, and churn is often driven not by price, but by the feeling that their situation is not understood. Many churn signals are subtle and difficult to detect within traditional processes.

NBA makes it possible to identify genuine attrition risk and respond in a way that reflects the context of the individual customer, rather than relying on uniform, rule-based approaches. As a result, retention actions become more targeted and more cost-effective.

2. rising operating costs

Mass renewal campaigns, pressure to meet short SLAs in claims handling, and steadily increasing case volumes mean teams are working harder, but not necessarily more effectively. Many activities consume time without generating value for either the customer or the financial result.

NBA brings order to this complexity by directing employee attention to where value is actually created and eliminating low-return activities. Teams begin to work more intelligently, with clear priorities, better context, and greater predictability of outcomes.

3. high customer expectations

Customers expect service that is fast, predictable, and tailored to their situation, regardless of whether they engage through an agent, online, by phone, or via a call center. They are accustomed to the level of personalization offered by e-commerce and banking and are increasingly unwilling to “start from scratch” with each interaction.

NBA enables a consistent, individualized experience across every channel and every interaction, regardless of who happens to be serving the customer. This builds loyalty and reduces price sensitivity.

How NBA works in practice: four processes where the greatest value is created

NBA shows its true strength only when it becomes part of everyday decision-making—embedded in processes that directly affect business performance.

1. Retention: predicting attrition instead of firefighting

NBA analyzes churn risk signals such as changes in behavior, atypical payment patterns, reactions to premium levels, or abandoned quotes. Rather than acting only when a customer is already leaving, the organization can know in advance:

• which customers are worth supporting with an agent conversation,
• where a clear explanation of a premium change should be sent,
• when an alternative offer is likely to increase retention,
• and when no intervention is actually the best decision.

The result is a more stable portfolio and predictable written premium—outcomes that cannot be achieved through intuition alone.

2. Renewals: changing the economics of contact

NBA assesses both the likelihood of renewal and whether contact will materially influence the customer’s decision. This allows insurers to:

• avoid mass, low-effectiveness outbound campaigns,
• direct agents to cases where their conversation truly changes the outcome,
• reduce negative customer experiences,
• while increasing conversion rates.

This is a rare process where costs decline and results improve, driven by intelligent orchestration of effort rather than escalating volumes.

3. Agent enablement: consistent decisions and higher effectiveness

NBA acts as an advisor, providing agents with context and recommendations:

• which products fit the customer’s current needs,
• how to structure the conversation,
• where coverage gaps exist,
• which arguments work best in a given segment.

NBA does not constrain agents. It enables them to act in a more consultative and consistent way, regardless of experience level.

4. Claims handling: speed where possible, control where necessary

At first notice of loss, NBA evaluates the nature of the claim and selects the appropriate path:

• straight-through processing for simple, low-risk cases,
• escalation for more complex claims,
• assignment to the right team,
• additional verification where fraud signals are detected.

The result is faster resolution for most customers, stronger control where risk is higher, and more predictable claims costs.

At first notice of loss, NBA evaluates the nature of the claim and selects the appropriate path:

• straight-through processing for simple, low-risk cases,
• escalation for more complex claims,
• assignment to the right team,
• additional verification where fraud signals are detected.

NBA architecture: what it looks like in practice

Implementing Next Best Action does not have to mean adding yet another IT system. NBA is a decisioning layer that sits on top of existing solutions and enables the entire organization to make consistent, data-driven decisions. In practice, NBA consists of four cooperating layers, each responsible for a different stage of information processing and decision execution.

1. Data layer

This is the foundation from which NBA draws information. In practice, it includes:

• a data warehouse or lakehouse,
• batch and near-real-time integrations (for example, APIs, event streaming, Kafka),
• segmentation of transactional, behavioral, and contextual data,
• unified customer identification (customer 360).

Data from policy, claims, payment, CRM systems, digital channels, portals, and quoting tools must be available in a consistent structure.

2. Prediction layer

This is where the models and signals used by NBA are created. In practice, the layer includes:

• machine learning models (propensity to buy, churn, claims risk, fraud probability),
• NLU and NLP models analyzing inquiry content or claim descriptions,
• prescriptive models assessing the impact of interventions,
• real-time scoring and model serving,
• model quality and drift monitoring (MLOps).

This layer delivers “decision signals.” However, models alone are not enough. Real value emerges only when they are connected to business logic.

3. Policy and constraints layer

This module ensures alignment with regulations and company policies. In practice, it includes:

• business rules engines (BRM or BRMS),
• a catalog of constraints—product, distribution, and regulatory,
• operational and segmentation priorities,
• discount limits, approval thresholds, and exceptions,
• compliance-by-design logic.

This layer is critical because it is where NBA “learns” how the organization actually makes decisions. As a result, recommendations are not only relevant, but also executable and aligned with the company’s risk appetite.

4. Decision execution layer

This is where NBA decisions translate into measurable business impact. Even the best recommendation has no value if it is not delivered where it can change the behavior of an employee, a system, or a customer. The execution layer is therefore responsible for embedding decisions into day-to-day operational processes. In practice, it includes:

• integration with CRM systems and portals so decisions influence priorities, contact scenarios, and agent workflows,
• APIs and event-driven mechanisms that inject decisions into real-time processes,
• workflow engines (for example, Camunda, Pega, Appian) that route cases down the appropriate path,
• communication and campaign systems that personalize messaging for individual customers,
• conversation support modules (next best conversation) in agency and contact center channels,
• automation of micro-decisions in claims, payments, and back-office processes.

This layer is what moves NBA from concept to tangible impact on key performance indicators.

Organizational readiness for NBA: three pillars that determine success

For NBA to drive real business impact, an organization must be ready not only technologically, but above all operationally. The key lies in how the company uses data, how it makes decisions, and how those decisions are executed in practice.

Pillar 1: data you can manage, not just store

The goal is no longer simply to connect data, but to ensure that the organization:

• trusts the data and uses it in day-to-day work,
• operates on a single version of the truth about the customer,
• can respond to current, not just historical, data,
• fosters a culture of data sharing across teams.

This represents a shift from “we have data” to “we manage data operationally.”

Pillar 2: decision logic that the business understands

Technology can generate predictions, but the organization must be able to:

• translate models into concrete actions,
• define clear rules, priorities, and constraints,
• establish a shared process for determining what constitutes the “best action,”
• ensure full auditability and accountability for decisions.

This is a shift from “we have models” to “we know how to use them.”

Pillar 3: decision execution as part of everyday work

NBA works only when people and systems actually use it. The organization must be able to:

• embed decisions into daily workflows,
• change behaviors, not just processes,
• build the capabilities of agents and operational teams,
• continuously measure the impact of decisions on KPIs.

This is a shift from “we have integration” to “NBA changes the way we work.”

Common challenges in NBA implementations

Implementing NBA is a cross-functional initiative involving business, IT, sales, operations, and data science. It is therefore natural that points of tension arise. The most important thing to recognize is that these challenges are common and well understood—and that there are proven ways to address them.

• Starting with too broad a scope. Ambitious goals are valuable, but implementation is most effective when it begins with a single process—most often renewals or retention—and only later expands to additional areas. This approach enables faster, measurable results and helps build trust in the solution.
• Focusing on models instead of decisions. Many initiatives start by building predictive models rather than defining the decision that needs to be made. Leading organizations begin by clearly specifying the decision and the desired outcome, then identify the required data, and only then develop the model. This sequence shortens time to value and improves the relevance of recommendations.
• Lack of clear KPIs. NBA must be measurable. Retention, customer lifetime value (CLV), cost to serve, contact effectiveness, or claims cycle time are examples of metrics that allow organizations to assess impact. The most effective projects start by answering a simple question: “How will we know it is working?”
• Insufficient embedding in agent and customer-facing work. NBA recommendations must be understandable, concrete, and usable in customer conversations. That is why leading organizations involve frontline employees already at the design stage, rather than only during rollout. This significantly accelerates adoption and improves effectiveness.
• Lack of governance over decision logic. NBA is a dynamic mechanism. Decision logic evolves with market conditions, models require ongoing updates, and business priorities change over time. It is therefore critical to define ownership of decision rules, how often they are reviewed, and the criteria used for approval.
• Too large a technological leap. Mature organizations start with simpler rules and gradually enrich them with predictive models and AI capabilities. NBA does not have to be a complex platform from day one. It can evolve incrementally, in line with the organization’s growing capabilities.

Summary: 10 questions to ask before implementing NBA

1. Do we have a consistent, unified view of the customer based on policy, claims, payment, and interaction data?
2. Are we able to measure the value of the decisions we currently make in retention, renewals, claims handling, and sales?
3. Do we know which processes have the greatest potential to benefit from NBA right now?
4. Are there clear governance rules for managing decision logic so that changes to rules and models are controlled and auditable?
5. Do operational teams understand how they will use recommendations in their day-to-day work and what will be expected of them?
6. Will NBA recommendations be delivered where decisions are actually made—to CRM systems, agent portals, claims workflows, sales tools, or digital channels?
7. Is our data available in real time, or at least in cycles that enable meaningful operational decisions?
8. Do we have analytical capabilities that can build, test, and maintain predictive models?
9. Are our systems ready to integrate a decisioning layer via APIs, events, and workflows?
10. Are we prepared to start with a small MVP that delivers initial value within weeks, rather than as part of a multi-year program?