Auditors decide which underwriting files to review by combining risk-based selection with a defined sampling methodology. They look at the size and complexity of the book, the coverholder's history, the class of business and the data supplied in the bordereaux, then pick a mix of random and targeted files. The goal is a sample that tests the whole underwriting process, not only the easy cases.
No two delegated authority audits use exactly the same sample, because the right selection depends on the binder, the product and the control environment. The sections below explain the main questions insurers and coverholders ask about how that selection works.

What risk factors shape an auditor's file selection?

Auditors select files based on where underwriting risk is most likely to sit. The main drivers are the class of business, the size of premium and limits, the coverholder's audit history, the complexity of the binder terms and any recent changes in staff, systems or underwriting appetite.

A binder covering a niche or non-standard product will usually attract closer scrutiny than a high-volume personal lines scheme with automated rating. The reason is simple: where underwriters apply judgement, there is more room for authority breaches, inconsistent pricing and weak documentation.

Other factors that commonly shape selection include:

• The territories and licensing position, particularly where business is written across several jurisdictions
• Referrals, exclusions and prohibited risks set out in the binding authority agreement
• Findings from previous audits and whether earlier recommendations were closed out
• Concerns raised by the capacity provider, such as loss ratio deterioration or unusual premium patterns

Auditors usually review this information before fieldwork begins, so the sample reflects genuine risk rather than convenience.

How do auditors choose between random and targeted sampling?

Auditors use both. Random sampling gives an unbiased view of how the coverholder underwrites day to day, while targeted sampling focuses on files carrying higher risk. Most delegated underwriting audits combine the two, so the findings are statistically fair and also probe the areas most likely to contain errors.

A random element protects the credibility of the audit. If an auditor only chose files they expected to fail, the results would overstate the problem and the coverholder could reasonably challenge them. A purely random sample has the opposite weakness: it can miss the small number of high-value or unusual risks that matter most.

Targeted selections typically include:

• The largest premium or highest limit risks
• Risks written close to the edge of authority limits
• Referred or declined risks, to test whether referral processes were followed
• Cancellations, mid-term adjustments and late-bound business

Documenting the logic behind each selection also helps. A clear rationale makes the audit report easier to defend and easier for the coverholder to act on.

How many underwriting files does an audit usually cover?

There is no universal figure. Sample size depends on the volume of business, the risk profile of the binder and the capacity provider's requirements. Small binders may be reviewed almost in full, while larger books are tested through a sample that is large enough to draw reasonable conclusions about the wider portfolio.

When deciding on underwriting file sample size, auditors typically weigh these points:

1. The number of policies bound during the audit period
2. The inherent risk of the product and the strength of the coverholder's controls
3. Any minimum sample expectations set by the insurer or capacity provider
4. The time available on site or remotely, balanced against the depth of testing needed

A coverholder with strong controls and a clean history may justify a smaller sample, whereas one with prior findings or a new product line usually warrants more. Agreeing the approach with the commissioning insurer in advance avoids disputes later.

Which files do auditors request from the bordereaux?

Auditors request files by working from the bordereaux, which list the risks bound and premiums reported. They use them to select a spread of policies across the audit period, then ask for the matching underwriting files so they can trace each risk from submission through to bound policy and reported premium.

The bordereaux are a useful selection tool because they show the full population of business before any files are pulled. Auditors commonly filter them to identify high premium risks, unusual classes, late entries, cancellations and policies with unexpected rates or limits.

For each selected policy, the file should allow the auditor to follow the whole decision trail. That normally means the submission, the risk assessment, the pricing basis, any referral to the insurer, the quote, the policy documents and evidence that premium and terms agree with the bordereau entry. Gaps between the bordereaux and the file are often where the most significant findings arise.

What red flags make auditors extend the file sample?

Auditors extend the sample when early testing reveals errors that suggest a control weakness rather than an isolated mistake. Repeated authority breaches, missing referrals, pricing outside agreed rates and poor file documentation are the most common triggers for pulling additional files.

The aim of extending the sample is to establish whether a problem is systemic. One missing referral may be an oversight, but several across different underwriters point to a process failure. Other red flags include:

• Bordereaux data that does not reconcile with the underlying file
• Business written outside permitted classes, territories or limits
• Inconsistent rating between similar risks
• Files that cannot be produced, or are produced incomplete

Extending the sample is not a sign the audit has gone badly. It is how auditors confirm the scale of an issue so recommendations can be proportionate and clearly explained in the report.

How Otter Audit helps with underwriting file review

Otter Audit's ACII-qualified auditors design every file sample around the specific binder, product and risk profile, rather than applying a generic template. With visits to over 500 insurance firms across 50 countries, the team understands how selection should adapt to different classes and territories.

• Risk-based sampling that blends random and targeted selections
• Clear reports that explain each finding and the reasoning behind recommendations
• On-time delivery, backed by a six-year record of meeting agreed deadlines

To see how this works in practice, explore our delegated authority audits or contact the team to discuss your next audit.

Related Articles

• What information is needed before a coverholder audit begins?
• How should insurers prepare their intermediaries for an audit?
• What happens when audit evidence is incomplete or inconsistent?
• How does an auditor select files for a delegated authority review?
• How do auditors assess compliance with a binding authority agreement?