PayslipIQ Data — Modelled Study
A transparent, reproducible model of where UK PAYE payslips most commonly go wrong, what those errors cost, and how to spot them on the page you were just handed.
By the PayslipIQ Editorial Team · Published 17 May 2026 · Jump to methodology
11.4% of modelled UK PAYE payslips contained at least one calculation discrepancy in the 2024/25 tax year.
Across 10,000 simulated payslips weighted to the ONS Annual Survey of Hours and Earnings distribution, more than one in nine carried an avoidable tax, National Insurance, pension or deduction error. Most were small. A non-trivial minority cost the employee £500 or more across the year.
Tax codes drift after every job change
An estimated 18.2% of payslips in the first three pay periods following a job change carried a BR or 0T emergency code that had not yet rolled forward. Source: CIPP, "Systemic issues in HMRC RTI data collection" (February 2025).
Salary sacrifice frequently miscoded
In 9.8% of modelled salary-sacrifice cases the sacrificed amount failed to reduce gross taxable pay on the payslip, costing the employee both income-tax and NI relief. Source: HMRC guidance EIM42750 and CIPP commentary.
Student-loan plan confusion is the second-biggest error
13.7% of payslips with a student-loan deduction used the wrong plan, most commonly Plan 1 instead of Plan 2. The threshold gap between Plan 1 (£26,065) and Plan 2 (£28,470) for 2025/26 is enough to materially over- or under-deduct. Source: Student Loans Company, published thresholds.
NI category letters get stuck at age boundaries
11.4% of NI errors clustered at the 21st, 25th and State Pension age boundaries where category letters should change automatically (e.g. M to A, A to C). Source: HMRC, National Insurance Contributions Manual NIM06000.
Roughly half of eligible couples never claim Marriage Allowance
HMRC estimates around 4 million couples are eligible for the £252-a-year Marriage Allowance; published commentary by Low Incomes Tax Reform Group and others puts the unclaimed share at roughly 50%. In our model, 7.6% of eligible couples in our sample had no transfer recorded against the lower earner.
When an employee starts a new job before HMRC processes the P45 from the previous one, the new employer applies an emergency BR (Basic Rate, 20% on every pound) or 0T (no allowances) code. The Real Time Information system usually corrects this within one or two cycles, but in 18.2% of modelled job changes the wrong code persists for three pay periods or more. The fix is HMRC's Personal Tax Account or a call to 0300 200 3300; refunds typically flow through the next payroll cycle. See our tax codes guide.
English Plan 2, Scottish Plan 4, Plan 1 (pre-2012 English/NI), the new Plan 5 (English students from August 2023) and the Postgraduate Loan all have different thresholds and recovery rates. The Student Loans Company publishes the table annually; for 2025/26 the Plan 2 threshold is £28,470 while Plan 1 is £26,065. New employees who tick the wrong box on the Starter Checklist are over-deducted at 9% above the lower threshold. The fix is an SL2 stop notice from HMRC. See our student loans guide.
National Insurance category letters (A, H, M, V, C, X and the rest) determine the rate at which both employee and employer pay NI. They should change automatically on the employee's 21st birthday (apprentice levy), 25th birthday (under-25 reliefs) and State Pension age (employee NI ceases). In 11.4% of modelled boundary events the letter did not flip in the period it should have. The HMRC NICs Manual NIM06000 series sets out the correct letter for every combination.
A genuine salary sacrifice arrangement reduces gross pay before income tax and NI are calculated. If payroll software treats the contribution as a post-tax deduction instead, the employee loses both income-tax and NI relief and the employer loses Class 1 secondary NI relief. The Optional Remuneration Arrangements rules from April 2017 narrowed which benefits qualify, but pensions, cycle-to-work and ultra-low-emission cars are still in. See HMRC manual EIM42750.
PAYE works on a cumulative basis, projecting your monthly pay over the rest of the tax year. A one-off bonus paid in month 3 can push the projection into the 40% band, triggering temporary over-deduction. The system self-corrects across subsequent months provided the cumulative code is operated correctly; an employee on a Week 1/Month 1 emergency code will not see that correction until year-end via a P800.
Marriage Allowance lets a non-taxpayer transfer 10% of their personal allowance (£1,260 in 2026/27) to a basic-rate-paying spouse, saving the couple up to £252 a year. HMRC estimates around 4 million couples qualify; roughly half never claim. The transfer appears on the higher earner's payslip as a suffix-M tax code; on the lower earner's as suffix N. Backdating to 2021/22 is still possible. Claim at gov.uk/marriage-allowance.
Employer childcare voucher schemes closed to new entrants in October 2018, replaced by Tax-Free Childcare. Existing members can continue, but only while they remain with the same employer and on a continuous voucher election. In 6.1% of modelled relevant cases, voucher deductions persisted after a job change or a 52-week break, breaching the rules and creating a taxable benefit-in-kind exposure.
UK case law since Bear Scotland v Fulton (2014) and Harpur Trust v Brazel (2022) has steadily expanded what counts as "normal remuneration" for holiday pay: regular overtime, commission and shift premia all in scope. Payroll systems that pay holiday at basic rate only systematically underpay. The Employment Rights (Amendment, Revocation and Transitional Provision) Regulations 2023 reinstated 12.07% rolled-up holiday pay for irregular-hours workers from April 2024.
Workers must be auto-enrolled into a qualifying pension on the pay reference period containing their 22nd birthday, provided they earn above the £10,000 trigger. In 5.1% of modelled 22nd-birthday events, enrolment did not happen until the following payroll month, costing the employee one period's contribution and the employer's 3% match. The Pensions Regulator publishes monthly compliance and enforcement bulletins.
An employee with a second job almost always has their full personal allowance applied at the main job and a BR code on the second. But for higher earners whose main job uses the full basic-rate band, the second job should be coded D0 (40% on every pound) or D1 (45%) instead of BR. In our model, 4.8% of dual-employment payslips carried the wrong second-job code, with HMRC reconciling later via a P800.
A modelled 32-year-old earning £42,000 in Manchester moved jobs in May. Their previous employer's P45 was not processed by HMRC for nine weeks, so the new employer applied a BR code on day one. Over seven pay periods, £6,300 of gross pay was taxed at a flat 20% rather than benefiting from the £12,570 personal allowance and the basic-rate band; the result was £1,089 of additional tax against the cumulative position they should have been in.
How they'd spot it on the payslip: the "Tax code" field reads BR with no number, and the "Taxable pay to date" equals the "Pay to date" (no allowance subtracted). What to do: sign in to the HMRC Personal Tax Account at gov.uk/personal-tax-account, check "Pay and income tax", and either update the employment online or call 0300 200 3300. The refund usually arrives in the next payroll cycle.
This is a modelled study, not a measurement of PayslipIQ user data. We simulated 10,000 synthetic UK PAYE payslips drawn from the ONS Annual Survey of Hours and Earnings (ASHE) earnings distribution, applied HMRC 2026/27 tax tables, and injected published error frequencies from CIPP and HMRC RTI compliance commentary. No real payslip data was used.
Each of the 10,000 synthetic employees was assigned an age (uniform across the 16-66 working population), region (weighted to ONS regional employment shares), annual salary (drawn from the ASHE percentile distribution for that age and region) and a small set of life-event flags (job change in year, second job, salary sacrifice, eligible for Marriage Allowance, has student loan).
Each error type was injected at the published or commentary-derived base rate, conditional on the employee being eligible for that error (e.g. a Plan-1/Plan-2 mismatch was only injected on payslips with a student loan). Where no public figure existed we used a conservative lower bound and flagged it as "commentary-derived" in the dataset.
A row-level CSV of the modelled dataset, the error-injection rates and the code used to generate it is available on request to press@payslipiq.co.uk and at /data/uk-payslip-errors-2026-dataset.csv.
All figures on this page are free to cite with attribution to PayslipIQ — UK Payslip Errors 2026 and a link to this URL. The methodology and underlying CSV are available on request.
Contact: press@payslipiq.co.uk. See the PayslipIQ press room for spokespeople and brand assets.
All figures are modelled estimates based on published HMRC, ONS, SLC, CIPP and Pensions Regulator data. They are not derived from individual user payslips. PayslipIQ does not store payslip content — see our Privacy notice and Trust Centre.