The rate is an average.
Your risk is a company.

The short answer

One in 199 companies on the effective register in England and Wales entered insolvency in the 12 months to 31 July 2026, a rate of 50.3 per 10,000, down from 52.5 the year before, according to the Insolvency Service. The fall is real. It is also an average, and the average is doing three jobs at once: dividing by a company register that keeps growing, blending sectors whose failure rates differ by more than four times, and mixing insolvency processes that mean different things for the creditors caught in them.

This guide takes the headline apart using the Insolvency Service's July 2026 statistics release and its Business Insolvency Demography publication, then sets out what a credit decision should actually take from a national number.

The denominator grows on its own

The published rate divides the number of companies that entered insolvency over twelve months by the mean number of companies on the effective Companies House register. Both halves move. When new incorporations run ahead of closures, the register grows, and a stable number of insolvencies produces a falling rate without a single company having become safer.

The Insolvency Service makes the point itself: today's rate remains far below the peak of 113.1 per 10,000 in the 2008-09 recession largely because the register has more than doubled since then. Half the gap between then and now is arithmetic, not health.

The spread the average conceals

Averages over 199 companies per failure tell you almost nothing about a named debtor. The demography publication, which measures businesses rather than registered companies, put the 2025 spread across the nine largest industries like this:

Accommodation and food has carried the highest rate every year since the series began in 2015; real estate the lowest every year since 2021. The two rate scales are not interchangeable, because one counts registered companies and the other counts businesses, but the ranking is consistent year after year. A hotel supplier and a letting agent do not share a risk profile, whatever the national number says.

Size and age move the number too

The same publication splits 2025 rates by band. Businesses with 20 to 49 employees failed at 269 per 10,000, the highest of any size band, while businesses with no employees or one failed at 82. Businesses turning over between £250,000 and £50 million failed at higher rates than the tails either side of them. On age, companies of seven to ten years failed at 170 per 10,000 and companies under two years at 37, the lowest band.

The pattern matters for credit terms. The mid-market band, where most trade credit actually sits, fails at closer to the top of these ranges than the bottom. Pricing a mid-sized hospitality operator off the national company rate understates the failure rate around it by several times over.

The mix shifted under the total

July 2026 saw 1,931 registered company insolvencies in England and Wales: 1,497 creditors' voluntary liquidations, 288 compulsory liquidations, 124 administrations and 22 company voluntary arrangements. More than three in four were CVLs.

The processes reach a creditor differently. A compulsory liquidation follows a winding-up petition a creditor has filed with the court, so by definition the debt has already gone legal; what happens after a petition is filed is its own timeline. A CVL is started by the directors themselves, often with little outward warning, and usually reaches the public through a first Gazette notice. A CVL-heavy total means quiet exits: failures that never generated a court record a supplier could have watched.

Administrations, the process most associated with rescuing a saleable business, were 33 per cent below June 2026 and 19 per cent below July 2025. A falling headline with a rising CVL share is a market failing quietly rather than loudly.

When the fall is genuine good news

None of this makes the statistic wrong. The twelve-month rolling construction exists precisely to smooth single-month noise, and a lower rate does mean fewer registered insolvencies per registered company. For anyone pricing a portfolio, a book of receivables or a regional economy, that is useful information and the fall should be welcomed.

What the rate cannot do is pick the name. No portfolio holds a representative sample of the register; it holds specific companies, in specific sectors, at specific sizes and ages. The rate prices populations. Credit decisions price names.

How to use the headline properly

Benchmark by sector, not by nation. Your expected failure rate looks like the sectors you sell into, not 50.3. A book that is half hospitality carries a different base rate from one that is half property management, and the two must not share a limit policy.

Weight by pounds, not by counts. Rates count failures. Losses count money. One mid-sized manufacturer failing at 269-equivalent rates moves a ledger more than a dozen two-person companies failing at 82.

Then leave the statistics behind. The company-level evidence is what decides the transaction: the trend in its own R-Score, its sector percentile, and what has landed on its record since its last accounts, which is the subject of why credit risk cannot be measured in snapshots. Around 3.4 million of the roughly 5.1 million searchable England and Wales companies carry a score, and the profile states outright whether it is improving, stable or declining.