How the pay comparisons work

ONS figures come from the Annual Survey of Hours and Earnings (ASHE). They cover employee jobs, not self-employed earnings. We show the matched occupation and employee group across the UK, covering all levels, regions and industries. A related broader occupation is named explicitly. Community filters do not turn these figures into a regional or seniority-specific estimate.

We use the latest usable estimate for the exact occupation and employee group first. If that is unavailable, we use the previous revised year with its original date, then a published three-digit parent group. For nursery and guest-house proprietors, a reviewed related management role provides employee-pay context. Its label explains the relationship; employee pay does not measure business profits. No pay changes or inflation adjustments are invented.

Hourly figures exclude overtime but can include other earnings beyond basic pay. Annual gross figures include additional pay and use a different eligibility population, including the ONS same-job requirement. Part-time annual pay is actual annual pay, not a full-time equivalent. These measures are labelled separately and should not be treated as interchangeable.

Serving armed forces personnel are outside ASHE. Their card instead shows published regular-service pay scales by rank, using the accepted 2026 Armed Forces’ Pay Review Body main pay spines. The range covers increments and, for other ranks, the four main trade supplements. It includes X-Factor, the military service supplement. Specialist rates, higher selected RAF engineering points, reserves and extra allowances use separate scales. These are pay-scale limits, not survey medians or quartiles.

Community figures come from voluntary reports of gross base pay. They are not a representative survey, verified payslips or verified unique people. They describe the reports in the displayed group, not what every worker should be paid.

Search suggestions use the ONS SOC 2020 occupation catalogue and safe title aliases from its coding index, plus explicitly maintained product roles. A suggestion needs your confirmation. Ambiguous abbreviations are not automatically assigned. We keep level and specialism separate: an AI product coordinator is not silently turned into a product manager.

The approved title examples beside a group explain its coverage. They come from the catalogue, not from rare or unreviewed contributor titles. A group can be broad; confirm that its responsibilities fit. Product manager (digital/software) uses the broader ONS IT managers benchmark, which includes IT product managers. This does not combine its community reports with all IT managers. Other custom roles need a reviewed source relationship before official figures appear.

Industries are broad categories chosen for this tool, not a complete official industrial classification. Region means where the work is normally carried out, including the normal working location of a remote worker. We do not collect a postcode.

We use gross base pay before employee deductions, excluding bonus, overtime and commission. Extras can be entered separately and do not enter the published base-pay figures. Direct employees, agency PAYE employees and umbrella assignment rates are kept separate. An assignment rate includes employer costs and is not an employee wage.

Monthly pay is multiplied by 12 and four-weekly pay by 13. A salary is divided by paid weeks and normal paid hours to obtain an hourly rate. Hourly pay uses the rate entered; weekly pay is divided by normal paid hours. Annual estimates for hourly or weekly pay require known paid weeks. We do not assume 52 paid weeks where they are unknown.

Use actual part-time earnings with actual paid hours, excluding unpaid breaks. Unpaid weeks and term-time arrangements matter. Irregular work, allowances, benefits and responsibility differences can limit a comparison. A comparison cannot decide whether minimum wage, equal-pay or contractual rules have been met.

Each displayed group requires at least 20 eligible reports dated within the preceding 12 calendar months, including the snapshot month. A recognised role and usable hourly rate are required. Unmatched, incomplete and unusual rates wait for review; a report can also be rejected. An unusual-rate check is a quality filter, not a judgement about lawful pay.

We split by level, then broad region, then industry only when every resulting group has at least 20 reports. Otherwise we show the broader parent. Published groups do not overlap, and we do not publish a parent alongside its smaller groups. The actual scope is always labelled. A broader figure is not an exact match for every selected filter.

Monthly snapshots show the median and the lower and upper quartiles: the middle value and the boundaries of the middle half. We use linear interpolation between sorted values, round hourly results to 25p and show report counts in bands. Groups below 50 reports are marked as early samples. There is no individual-pay export or company ranking.

A snapshot is fixed once published. Later snapshots retain its group boundaries and require at least 20 changed reports in each changed group; small changes or changed boundaries are withheld. Editing, withdrawing or rejecting a report revokes affected role results until safe republication. Results older than the previous calendar month are hidden. These checks reduce disclosure risk; a numerical threshold alone is not a guarantee of anonymity.

Exploring a comparison does not submit your pay. Only the final Share my pay action sends a report. Your entered title and optional employer name stay private and are never included in public benchmark responses. Employer names are not required to take part and are not used to sell individual reports.

We do not ask for your name, email, account, date of birth, full address or payslip. Ordinary hosting requests still include technical information such as an IP address. Forms and receipts are excluded from session replay and automatic interaction capture. Limited abuse counters use short-lived keyed hashes, not the report itself.

Save the private receipt. Anyone holding it can read, edit or withdraw that report. We store a hash of the receipt, not the original secret. We cannot recover a lost receipt using an email address. Private reports expire after at most 24 calendar months of pay history; rejected reports are removed after 30 days and review records after 90 days. Withdrawal removes the report and affected live comparisons, but cannot recall copies already saved by others.