Business insurance benchmark research: matching geography, product and formula

Public UK insurance data rarely isolates small-business customers in England, so an honest benchmark review often ends with contextual evidence rather than a direct performance target.

What to take away

Write the comparison question

Specify the internal metric and intended use. "Are claims slow?" is too vague. Define claim type, start event, completion event, reporting period, customer population and whether the result will trigger investigation, capacity or supplier review.

Create an acceptance table before searching:

For example, a broker wants to compare average claim settlement time for commercial property in England. The FCA general insurance value measures data 2025 gives UK retail claim figures. The geography row matches UK, but the customer row fails because it is retail, not commercial. The formula row fails because it measures payouts, not settlement time.

The row is rejected.

The Bank of England insurance aggregate annual report gives UK commercial property premium totals, but it has no England split and no settlement time. That row is also rejected.

Use regulatory data in its published scope

The FCA's general insurance value measures for 2025 report, 2025 edition, covers claims frequency, acceptance, average payout, claims complaints and claims cost relative to premium for UK retail general-insurance products meeting specified thresholds. The publication's data tables describe data-quality work and remaining inconsistencies.

This is not commercial-insurance performance for England. Business mix, product age, policy duration, volumes, inflation and timing can affect the values. The FCA says the historical measures are not designed for consumers to select a current policy.

The FCA's firm-specific complaints data, for example the 2024 H2 edition, covers reporting firms above a publication threshold and can include joint reporters and broad product groups. The underlying page records revisions made after a reporting anomaly. Check the notes and downloadable table before comparing firms.

Distinguish market scale from service quality

The Bank of England's insurance aggregate annual report, 2024 edition, publishes structured UK premium and claims-development information for Solvency II firms by lines including commercial property, employers' liability and professional indemnity.

Those aggregates help describe sector scale and development. They do not show an individual small firm's expected price, an England market share or a distributor's service standard. Avoid dividing UK premium totals by England business counts to invent an average.

For England-specific business population, the Office for National Statistics UK Business Counts show around 5.5 million businesses in England in 2024. That figure sets context but not insurance performance.

Build an internal peer group carefully

Where public matches do not exist, compare similar internal cases by trade, complexity, channel, policy type and period. Remove personal identifiers from analysis and set minimum group sizes. Document selection before seeing results to reduce favourable cherry-picking.

Use peers to identify an investigation, not automatic underperformance. A longer completion time may reflect a complex exposure or a failed process; case review must distinguish them.

Publish a benchmark evidence note

Store source URL, edition, retrieval date, table, formula, scope, exclusions, corrections and transformation.

Label the conclusion as direct, adjusted or contextual. Set an expiry based on the source cycle.

Where adjustment is proposed, publish the calculation and reason. Do not scale a UK figure to England by business-population share unless the measured variable and geography genuinely support that method. Premium, claims and complaints depend on business mix, exposure and reporting, not population alone.

Have a second researcher reopen the source and confirm the table, units and notes. Record corrections issued after the original release. At each review date, decide whether to refresh, retain as historical context or remove the benchmark from active reporting.

Before you act

Common questions

What should an acceptance table include before searching for a benchmark?

The acceptance table should cover geography, customer, product, period, formula, threshold and quality. Each field must match the decision. For example, geography must state England, Great Britain or UK accurately, and customer must specify consumer, microbusiness, SME, commercial or mixed.

Why is the FCA general insurance value measures report not suitable for commercial insurance performance in England?

The FCA report covers UK retail general-insurance products meeting specified thresholds. It is not commercial-insurance performance for England. Business mix, product age, policy duration, volumes, inflation and timing can affect the values, and the historical measures are not designed for consumers to select a current policy.

How should an organisation proceed if no external benchmark meets the acceptance table?

If no external value meets the acceptance table, say that clearly and use the organisation's baseline plus an improvement target. An unsupported benchmark is less informative than a transparent absence of comparable evidence. Label the conclusion as direct, adjusted or contextual and set an expiry based on the source cycle.