research note
How old is the data Britain runs on?
Every indicator in this index holds the most recent figure its publisher has released. Taken together they are a median of 8 months old. That average hides the finding: the country measures what its institutions do far faster than what happens to the people they serve.
71 scored indicators, August 2026 release, measured at 19 August 2026.
The finding
2 months for hospital activity, 20 for population health
The gap between the fastest and slowest parts of the national picture is a factor of 10. Public Service Capacity is current to 2 months. Population Health runs 20 months behind, which means the most recent published view of how long people live in good health, or how many are dying of drugs, describes a country nearly two years ago.
Median age of the latest published figure, by domain. Bars in the warning tone are a year or more behind.
Why
Administrative data is a byproduct. Outcome data has to be commissioned.
The fast indicators are the ones that fall out of running a system. A hospital records an appointment because it has to schedule it. The Treasury knows what it borrowed because it borrowed it. A prison counts the people in it every week. That data arrives monthly or weekly because collecting it is inseparable from the operation itself.
The slow indicators mostly come from surveys and registrations: household income, food insecurity, mental health prevalence, healthy life expectancy. Somebody has to decide to ask, fund the asking, wait for fieldwork, and process the answers. The psychiatric morbidity survey behind the mental health indicator runs roughly every seven years, which is why the newest figure is 29 months old.
The consequence is a system that sees its own throughput in near real time and the condition of the population late. Waiting lists, borrowing and prison capacity are visible fast enough to manage week to week. Whether children are poorer, whether people are dying earlier, whether housing costs have outrun incomes, arrive a year or two after the fact, by which point the policy that caused the movement may itself be two governments old.
The range
Oldest and freshest indicators
Oldest
Method and limits
What this measures, and what it does not
Age is the number of months between the end of the period a figure covers and today. A figure published this week describing the year to March 2025 is seventeen months old, because that is how far back the evidence reaches, not how recently it was typed up. Financial years are treated as ending in March, quarters at their closing month.
This is a proxy, not a census. It covers the 71 indicators this index scores, chosen to describe national conditions. Government holds faster data internally that it does not publish, and publishes statistics this index does not use. The claim is about the evidence base a citizen, journalist or minister can actually consult, not about everything that exists.
Where an indicator sits behind its own publisher, the age shown here overstates the delay by that margin. The freshness process checks each indicator against its publisher and records when the next edition is due, so that gap is tracked rather than assumed.
Recomputable from the published data: the per indicator ages are in latency.json, and every period and source is in the release downloads.