Random holdouts (RHOs) have become a defining and often debated feature of accelerated underwriting (AUW) in individual life insurance. While they are frequently viewed as a necessary source of friction in an otherwise streamlined process, their true value is often misunderstood. 

This paper reframes RHOs across four dimensions – the good, the bad, the ugly, and the way forward–to clarify their intended role as a measurement tool, highlight where they fall short in practice, and consider how they can evolve to meet changing industry needs. Ultimately, the goal is not to eliminate RHOs, but to use them as a more effective, data-driven, and customer-friendly component of the underwriting toolkit.

• RHOs are a measurement tool, not a risk protection tool 
  Their primary purpose is to quantify mortality slippage and validate AUW performance, not to catch bad risks. When used as a blunt control rather than a diagnostic tool, they can increase cost and customer friction without delivering meaningful insight.

• How RHOs are implemented matters  
  Applying RHOs only to cases accepted by rules engines provides a clearer view of residual risk and a more accurate assessment of mortality slippage. Targeting higher ages and larger face amounts can improve insight into business segments with the potential for material mortality impact. 

• The opportunity is to modernize, not eliminate 
  While RHOs are not always viewed positively, the goal should be to modernize and reposition them as one component of a broader underwriting toolkit, working alongside predictive models, advisor monitoring, and targeted post-issue reviews.

• The future is less invasive and more data-driven 
  Emerging access to digital health data, such as clinical lab results and prescription histories, creates an opportunity for the Canadian insurance industry to rely less on traditional evidence that can be slow and invasive. These sources can provide a comprehensive view of an applicant’s health, reducing reliance on fluid evidence without adversely affecting mortality experience.

RHOs are fundamentally a measurement tool, not a protective value tool. Unlike targeted holdouts, which are triggered by specific risk signals, RHOs are applied randomly to cases that would otherwise be approved through AUW. Their role is to provide an unbiased benchmark for evaluating how well underwriting rules, models, and data sources perform in practice, particularly with respect to mortality slippage.  

For this paper, we define mortality slippage as the expected deterioration in mortality introduced by the removal of traditional requirements in the underwriting process. 

When designed correctly, RHOs provide a ground-truth view of residual risk. Differences in outcomes, such as higher substandard or decline rates within RHOs, can highlight non-disclosure, model limitations, and gaps in underwriting rules, enabling meaningful program refinement. Programs often begin with elevated RHO levels that are reduced over time as confidence in AUW grows. This can provide protective value through a “sentinel effect,” which discourages misrepresentation given the possibility of fluid testing. 

From an implementation perspective, insurers vary considerably in how they calculate mortality slippage and incorporate RHOs into that process. 

• Timing: Some insurers may assign RHOs based on a fixed percentage of applications, while others may kick out a defined number of cases per monitoring period. Some insurers select the RHOs prior to triage and kick-out models and some select them afterward from the narrower pool of cases that would have been decisioned without further requirements.   
• Selection logic: Some insurers may disproportionately select RHOs at higher ages and face amounts, whereas others apply consistent holdout rates across predefined age and amount bands. 
• Measurement capability: Some companies track mortality slippage at a policy level, while others assess it using portfolio-level risk class distributions.  

A general best practice is to apply RHOs after all underwriting rules and decisions have been completed and to limit them to cases the AUW engine would have accepted. This ensures RHOs isolate true residual risk and provide a cleaner assessment of mortality slippage. 

RHOs also deliver greater value at older issue ages and larger face amounts, where financial exposure is highest. From a cost-benefit perspective, lower RHO rates may be appropriate for younger applicants with smaller policies, helping preserve a streamlined customer experience. 

Tracking at the policy level yields deeper insights and greater clarity into overall portfolio slippage, but some companies are hindered by IT and workflow challenges.

There is no single “correct” RHO percentage. The goal is to generate sufficient credible data to quantify slippage with confidence while maintaining a balanced customer experience. As a point of reference, approximately 5% of eligible applications with a face amount of $500,000 or more were processed as RHOs in 2025.¹ However, levels vary by company, age, and face amount. 

What matters is generating enough fully underwritten observations to produce stable insights. Smaller samples are prone to volatility and can misrepresent performance. Exact numbers of holdouts will vary by program, but targeting several hundred to one thousand RHO cases per year allows for a credible assessment of mortality slippage and emerging trends. However, achieving the right volume can be challenging for smaller blocks of business, especially at higher ages and amounts.  

Ultimately, RHOs should be calibrated to produce credible insights, not simply to satisfy a predefined percentage target. This ensures they remain a disciplined mechanism for evaluating underwriting effectiveness and supporting continuous improvement. 

Let’s start with the obvious: RHOs are unpopular. They introduce friction into the customer experience. 

From the applicant’s perspective, being pulled into a traditional underwriting path after expecting a quick decision can leave them feeling disrupted and frustrated. The experience is intrusive, slow, and misaligned with modern expectations.  

Advisors experience this challenge just as acutely. RHOs introduce uncertainty into the sales process, extending cycle times and increasing the risk of attrition. Every additional requirement creates an opportunity for second thoughts, competing offers, or lost placements. As a result, RHOs are often viewed not as a control mechanism, but as a threat to client acquisition.  

From an insurer’s perspective, the operational impact can compound quickly. Longer cycle times, higher wastage, and advisor pushback all reinforce the perception that RHOs should be minimized rather than embraced. 

The real problem, however, is not that RHOs are unpopular. It’s that they are often used for the wrong purpose. 

At their core, RHOs are not  a risk-selection tool; they are measurement tool. Their purpose is to help insurers evaluate how well an AUW program is performing over time and answer these hard-but-critical questions: 

• Are underwriting rules and predictive models segmenting risk as intended? 
• Is mortality emerging in line with expectations? 
• Where is mortality slippage occurring and why? 

Too often, RHOs are treated as a blunt control – a way to “double-check” accelerated decisions or provide additional comfort. But in some cases, it is unclear whether the results are actually being used to improve the AUW program.  

Some carriers conduct RHOs without analytics models in place or a clear feedback loop to translate findings into rule changes, model refinement, or distribution oversight. Without a defined mechanism to extract and act on the learnings, RHOs generate friction without delivering insight. Insurers may collect the data but fail to extract the behavioral signals that matter, such as how applicants flow through rules, how advisors respond to the process, and how risk truly differentiates once full evidence is introduced. 

The result is an uncomfortable paradox: RHOs impose real costs on applicants, advisors, and operations, yet fail to deliver their full actuarial and underwriting value. That’s the ugly part. We tolerate the pain without fully capturing the insight RHOs are intended to provide. 

There is a path forward, and it does not involve eliminating RHOs altogether. 

The goal should be to modernize and reposition RHOs as one component of a broader underwriting toolkit, working alongside predictive models, advisor monitoring, and targeted post‑issue reviews. When used intentionally, RHOs can do what no model alone can: provide a clear lens into real‑world underwriting performance. 

Looking ahead, this requires a shift in both design and data. 

• First, RHOs should be purpose‑built to evaluate underwriting rules and models, not simply replicate traditional underwriting on a subset of AUW cases. That means defining what success looks like and understanding how the learnings will be used to improve the AUW program, whether that means fine-tuning predictive models, addressing advisor concerns, or refining application questions.  
• Second, RHOs should not stand alone. They are most effective when used alongside complementary tools such as: 
  • Predictive analytics, such as for smoker propensity, BMI misrepresentation, or non-disclosure of select medical conditions, to help identify applications that warrant additional underwriting evidence.  
  • Advisor behaviour monitoring, including tracking non-disclosure rates, to identify areas where distribution practices could use additional scrutiny.  
  • Post‑issue monitoring, i.e., monitoring after the policy is placed, typically with Attending Physician Statements (APS), which are not commonly used in Canada but have been proven to deliver excellent results in the United States. 

Finally, the experience itself is poised for meaningful evolution. Traditionally, RHOs have been synonymous with invasive evidence collection, such as in-person exams and fluid collection, which add friction. That paradigm is beginning to shift. Digital medical records, such as clinical lab data and prescription histories, are becoming more accessible in Canada. In many cases, this data overlaps with information obtained through traditional insurance labs, and can sometimes provide a more comprehensive view of an applicant’s medical history.

This creates an exciting opportunity to reimagine RHOs, replacing portions of traditional evidence requirements with digital data where appropriate. The result is a win on both sides: improved risk assessment coupled with a faster, more seamless, and customer-friendly underwriting journey.

We will explore this idea further in an upcoming paper on how digital health records have real potential to reduce reliance on invasive insurance labs by leveraging equivalent records in existing medical data when available. More on this topic to come. 

As underwriting continues to evolve, so too must the role of random holdouts. By shifting toward less invasive approaches, insurers can preserve the measurement value of RHOs while supporting a faster, more seamless customer experience. Used thoughtfully, RHOs remain one of the few mechanisms insurers have to truly understand how their AUW programs are performing, and that insight is worth preserving.