Methodology

How is the pump price split into its parts?

Each week's pump price is split into four parts. The first three are measured; the fourth is whatever is left.

Because the margin is a residual, anything the Brent benchmark misses — biofuel content, refinery timing, the crack spread — lands in it. It is not a measure of retailer profit.

How are year-on-year changes worked out?

Each figure is compared with the observation nearest to exactly one year before its own date, taken from the same series. Where the series is annual or semi-annual, the nearest observation can be some months from the anniversary, so the comparison date is shown next to the change rather than implied. A comparison more than 400 days from the anniversary is not shown at all.

Which weeks are flagged, and why?

Three checks run over the weekly decomposition on every build.

• Oil dearer, pump cheaper. Weeks where Brent in dollars was higher than the latest week but the pump price was lower. Runs of nearby qualifying weeks describe the same event, so only the most striking week of each episode is kept.
• The same in pence. The identical test with crude converted to pence per litre, which removes the exchange rate and isolates duty, VAT and margin.
• Wide margins. Weeks where the residual margin sat more than 1.5 standard deviations above its own trailing 52-week mean. Consecutive weeks are grouped into episodes.

How is the rockets-and-feathers estimate made?

Weekly changes in the pump price are regressed on the positive and negative parts of weekly changes in crude (in pence per litre, so the exchange rate is already inside the input), with lags from zero to eight weeks:

ΔPump_t = a + Σ_k β⁺_k ΔCrude⁺_t−k + Σ_k β⁻_k ΔCrude⁻_t−k

Three numbers come out. Cumulative pass-through is the sum of the coefficients in each direction: how much of a move in crude eventually reaches the pump. The median lag is the point at which half of that response has arrived, and is the headline figure. The mean lag is the coefficient-weighted average, which weights the tail of the response and can point the other way; when it does, the page says so rather than quoting whichever number reads better. Fitted by ordinary least squares with a small ridge term, since adjacent lag columns are nearly collinear.

Current fit: 1153 weekly observations, R² 0.607.

This is one specification estimated on national weekly averages. It describes how the average price across thousands of forecourts has moved; it is not evidence about any individual retailer.

What stops a bad value getting through?

Every fetched value passes two tests before it is accepted.

• Bounds. Absolute limits per series — UK petrol between 80p and 400p, Brent between $10 and $300, and so on. A value outside them is rejected whatever surrounds it, which is what catches a units error.
• Step change. A limit on the change between consecutive observations, typically 15% for weekly pump prices and 25% for daily crude. This one is a suspicion rather than a verdict: crude really did fall 30% in a day when the Gulf War started. So a value that breaks the limit is accepted when the next observation confirms the new level, and rejected when nothing confirms it — which includes the case that matters most on a nightly run, a lone bad value at the end of a series.

A rejected value never replaces a good one. A source that fails entirely never blanks a figure: the stored series is kept, the page shows the last good value with its as-of date, and a marker appears if it is older than that source normally publishes. The status page lists every source and its state.

Do the parts add up to the totals?

Every chart or table that splits a total into parts is checked on every build: the parts must add to the stated total within 0.5%, or the site is not published. Where a source's own parts fall short of its total, the gap is shown as a line of its own ("other NICs", "not itemised", "other or not yet recorded") rather than stretched over the other parts.

Tax receipts

The all-taxes page uses one measure end to end: HMRC tax receipts and National Insurance contributions for the UK: cash receipts by financial year, outturn, from HMRC’s annual receipts table. Its components are every column of that table that is not already counted inside another (PAYE and Self Assessment are inside income tax, offshore inside corporation tax), plus HMRC's penalties, which are not a tax but are in its total. Council tax, business rates, Vehicle Excise Duty and the TV licence are not collected by HMRC and are not in this total.

Before this check existed, the page compared HMRC's total with only 16 of its components, one of them the wrong column, and fell £34.5 billion (3.7%) short. With every component it ties to within HMRC's rounding.

Every other chart with parts and a total

How are the council report cards and the council map compiled?

Every council tax billing authority in England, Scotland and Wales has a report card on its council tax page, and every one appears on the council map. The rules are the same for both:

• Only official findings and published figures, each attributed to the body that made or published it, with the period it covers and a link to the original.
• No characterisation of any council, official or organisation in this site’s own words. Where a source uses a word such as “inadequate”, it is quoted as that source’s rating.
• No named junior staff, no individual complaint details, and nothing beyond what the source publishes. Ombudsman reports are linked, not summarised.
• Each figure is compared with the figure for its own nation, one measure at a time. There is no composite score and no ranking of councils.
• A council, or anyone else, can reply: every report card carries a contact address, and every correction is logged on the corrections page.

How are councils matched across publishers?

Most publishers name councils in words rather than codes, and in different ways (“Bristol, City of”, “Argyll & Bute”, “Durham County Council”). Every dataset is joined to the council’s ONS code through one set of name rules, and the nightly run reports any council tax area it cannot match. Scottish and Welsh councils take their codes from the ONS boundary file, because their council tax releases carry none. Where a council was recoded (Barnsley and Sheffield in 2025), older data is mapped to the current code.

What happens in two-tier areas?

In most of shire England a district council bills council tax and a county council runs social care, children’s services, highways and waste disposal. The report card and the map sit at the level of the billing authority. Where a measure belongs to the county (Ofsted, CQC and road condition), the district shows its county’s figure and says so, and each district’s page has a section with the county council’s own figures.

Where do the complaints figures come from?

Where do the finance figures come from?

Where do the audit figures come from?

Where do the pay figures come from?

Councils must publish, in their statement of accounts, how many employees were paid £50,000 or more (£60,000 in Wales) and what was paid to councillors in allowances. There is no central dataset of either, so these are read from each council’s accounts by hand, with the page they come from, and added council by council. 39 council(s) so far, from statements of accounts read by hand. From each council's statement of accounts. England and Scotland count employees paid £50,000 or more; Wales counts from £60,000. Councils differ in whether schools staff and senior officers are included in the count, which the note on each entry records. Coverage grows as statements of accounts are read. The thresholds are not uprated, so more staff cross them each year as pay rises, and councils differ in whether they count schools staff.

Where do the service ratings come from?

How are council tax per head and the rise worked out?

The rise is the change in the Band D bill on the year before, from the two years’ published figures (Wales publishes the change itself). Council tax per head is the area’s Band D bill multiplied by its Band D equivalent tax base, divided by its mid-year population: the total the area pays across every precepting body, spread over everyone who lives there. Scotland publishes no tax base in a comparable form, so it has no figure.

Which official events are listed?

Each council’s timeline lists decisions and publications by official bodies: its council tax, section 114 reports, intervention, Exceptional Financial Support, ombudsman public interest reports, audit opinions and reports, local election results (seats won at each election, from Democracy Club’s open data, CC BY 4.0), and reorganisation under a structural changes order. Every event is dated and linked to its source. News reports and third-party headlines are never included.

How is the map drawn?

Boundaries are the Office for National Statistics’ generalised local authority district boundaries for the edition matching the council tax year (May 2026), simplified to TopoJSON at build time and drawn without map tiles or any external map service. Councils created by reorganisation are drawn as they are now and their pages name the councils they replaced. Number measures are coloured in five bands, each holding about a fifth of councils with a figure; ratings get one colour each. Both use a single blue scale that can be read with the common forms of colour blindness, and “no data” and Northern Ireland are hatched rather than coloured.

Source: Office for National Statistics licensed under the Open Government Licence v.3.0. Contains OS data © Crown copyright and database right 2026.

How is "where your council tax goes" worked out?

Every council page has a drill-down from the bill for your band to the lines each council's spending return itemises. The figures are the councils' own returns to government:

In Scotland the whole bill goes to the council: police and fire are funded nationally.

One list of services for three nations

Each government collects the figures on its own forms. One taxonomy (education, social care, roads and so on, with sub-services below) is mapped from each nation's lines in the pipeline, and published as taxonomy.json with every mapping. A nation that does not publish a line has no figure for it; nothing is estimated. Where a service's itemised lines do not add up to its total, the gap is shown as "not itemised in the return" rather than spread across the other lines.

How your bill is shared out

Council tax is pooled with a council's grants, business rates and fees; it is not ring-fenced for any service. So the drill-down shares each part of your bill across the receiving body's services in proportion to its net spending on each: if a council's net spending is 40% adult social care, 40% of its part of your bill is shown against adult social care. It is an allocation, not a record of what your money bought. Two adjustments come first:

• Spending paid for by a ring-fenced grant is taken off, because council tax does not pay for it. In England that is the Dedicated Schools Grant (from education) and the public health grant (from public health). Scottish figures are already net of every grant. Wales does not publish its specific grants by service, so Welsh figures cannot be adjusted this way.
• A service that brings in more than it costs (parking, often) takes no share and is listed separately.

Each body's spending comes from its latest published return, which covers an earlier year than your bill. Other bands are Band D multiplied by the fraction set in law for each band.

Comparisons

"Similar councils" are the ten bodies of the same kind in the same nation closest in population (every one, where there are ten or fewer): district with district, county with county, police force with police force. Combined authorities and the Greater London Authority each do a different mix of things, so they are not compared. Per-resident figures use the ONS mid-year population estimates; for a county, police force, fire authority or combined authority, the population is the sum of the districts it covers. Year-on-year changes compare the same line in the body's two newest returns.

Where the rest of the money comes from

The share of a council's money that comes from council tax is its council tax (excluding parish precepts) divided by the total of these sources:

What it adds up to

What it cannot show

No council publishes spending at street or property level. The national returns stop at the sub-service lines listed; councils' budget books and statements of accounts have more. Council housing is paid for through the Housing Revenue Account from tenants' rents and is separate from council tax; it is shown alongside, not in, the split.

How is the spending over £500 compiled and classified?

English councils must publish every payment of £500 or more under the Local Government Transparency Code 2015; Scottish and Welsh councils are not bound by it. Each council publishes its own way, so this site keeps a list of where each council's files are and how to read them (registry.json), and every night reads the latest twelve months into one format: date, supplier, amount, the council's service area, expense type and description, and the file each payment came from.

The categories are this site's classification, not the council's, and every page says so. Rules are tried in the order above and the first that matches wins; most payments match none. The keyword lists, word for word, are in summary.json under rules.

Matching payments to services

To name suppliers in the bill drill-down, each payment is matched to a service in the shared list from the council's own service area, expense type and description, using keyword rules published in the same file. Council housing (the Housing Revenue Account) is kept apart. The matching is ours and is labelled as such; a payment the rules cannot place is left unmatched rather than guessed.

How are the “When it goes wrong” figures compiled?

• Major projects: the Government Major Projects Portfolio data, from March 2022, when delivery confidence settled on a red, amber and green scale. From the 2024-25 data a project carries either the independent rating or the department’s own, so the two are counted separately. Whole-life costs are summed only for projects with a published figure, as the file gives them: most years in cash, the latest in 2024-25 prices, so totals are not comparable across that change. Projects are matched to the previous year by their portfolio ID.
• Reports: the National Audit Office’s value-for-money reports and the Public Accounts Committee’s reports, read from their own APIs, with their own titles.
• Fraud and error: the Public Sector Fraud Authority’s, DWP’s and HMRC’s estimates, quoted exactly. They are estimates from samples, published as ranges.

• Losses and special payments: read by hand from each department’s latest annual report and accounts, with the page. Departments count and group cases differently, so the figures are not added up or ranked.
• MPs’ pay and costs: IPSA’s published salary table and its annual totals of each MP’s business costs. MPs first elected in July 2024 have part-year figures, so each MP is compared with the average for MPs in the same position. The costs are for running an MP’s office; they are not paid to the MP.
• Contracts: award notices published by central government departments on Contracts Finder and, from 24 February 2025, on Find a Tender, read a day at a time. The value is the value stated in the notice, which for a framework agreement can be the maximum that might ever be spent under it. It is awarded value, not money paid. Only notices whose buyer is a department by name are included; agencies publish under their own names.

Which sources are behind the report cards, the map and “When it goes wrong”?

Which source does each figure come from?

Figures kept by hand — Ofgem's cap components, the eVED rates, the policy timeline, the renewables scheme costs — are in config/manual.json with a source link and a verification date each, and that date is shown next to the figure on the page. A nightly check reads Ofgem's published unit rates and raises an issue if they no longer match what is in that file.