How to run an annual pay review: process and governance

Before you start

Legal basis. Under the Equality Act 2010 [2]:

• an employee doing equal work to a colleague of the opposite sex is entitled to terms no less favourable, unless the employer shows a material factor that does not involve less favourable treatment because of sex;
• once a woman's maternity leave ends, her pay must include any increase she would have had if she had not been on leave;
• a term stopping staff discussing pay is unenforceable where they seek to find out whether pay is linked to a protected characteristic.

GOV.UK states that minimum wage rates change on 1 April every year [3], though no statute fixes that date: it is set by the amending regulations, and a new rate applies from the first pay reference period starting on or after it [6]. Employers with 250 or more employees must publish gender pay gap figures [5].

Who to involve. Reward or HR, finance, payroll, line managers, the signatory, and any trade union or employee representatives.

What you need. Contracts, pay policy, ranges, payroll and performance data, and last year's records. lumi figures below are national, from the 2026 H1 collection window: see the full findings.

Step 1: Set the calendar

Give each stage an owner and a date: data cut-off, manager briefing, proposals, calibration, fairness review, sign-off, letters, payroll cut-off, effective date and audit. In the lumi reward benchmark, "Annually" is the common answer for the main cycle, at 84.1% (n = 251). "Twice a year" (6.8%) and "Quarterly" (4.0%) are rare, and 5.2% have no regular cycle.

Fix the external dates:

• National Living Wage. From 1 April 2026 the rate for workers aged 21 and over is £12.71 an hour [3]. A new rate applies from the first pay reference period starting on or after the change [6]. The Low Pay Commission has been asked to advise by the end of October 2026 on rates from 1 April 2027 [7].
• Gender pay gap snapshot. 5 April for most employers and 31 March for most public authorities, with publication within a year [5].

Worked example (illustrative rates). Inputs: a team member on £12.40 an hour, paid monthly from the 1st, moves to £12.71 on 1 April. A supervisor on £13.30 waits for a 1 July review.

Plan how you will handle that gap.

Step 2: Prepare the data

Build one record per employee: grade, range, pay, hours, rating, start date, last increase, leave status and working pattern, plus any equality data held for monitoring. Freeze it at the cut-off and reconcile it with payroll.

In the lumi reward benchmark, "Spreadsheets" is the common answer for running pay review cycles, at 63.3% (n = 264). On a spreadsheet, lock formulas, limit editors and keep a version log.

53.9% define formal pay ranges for all permanent roles and 26.2% do so partially (n = 267). List roles without a range and agree what managers will use instead.

Step 3: Choose how increases are differentiated

In the lumi reward benchmark (n = 256), not using a formal merit matrix ("No") is the common answer, at 52.0%. "Performance only" is an alternative (30.9%) and "Full merit matrix" is rare (17.2%). Asked whether increases differ by rating (n = 258), "No – flat or near-flat" is common (45.0%), "Yes – moderately differentiated" an alternative (36.4%) and "Yes – strongly differentiated" rare (18.6%).

Write down the rule and who is eligible, including people on leave, before managers start.

Step 4: Set position-in-range rules

Compa-ratio is pay ÷ range midpoint. Range penetration is (pay − minimum) ÷ (maximum − minimum).

Worked example (illustrative). Inputs: a range of £28,000 to £36,000, so the midpoint is £32,000.

In the lumi reward benchmark (n = 160, "Not applicable" excluded), "At midpoint (100%)" is the common target for a fully competent employee (60.6%). "Above midpoint" is an alternative (20.6%) and "Below midpoint" is rare (18.8%).

"Frozen until range moves" is common (50.2%), "Lump-sum in lieu" an alternative (30.5%) and "Continue increases" rare (19.2%) (n = 213, "Not applicable" excluded).

Step 5: Brief managers

Give managers the timetable, the rules, how leave and absence are treated, how to record reasons and flag exceptions, and what to say before sign-off. In the lumi reward benchmark (n = 265), "Guidance pack only" is the common answer (44.9%), "Neither" an alternative (43.4%) and "Mandatory training" rare (11.7%). A pack is cheaper. Training builds shared understanding but takes time.

Step 6: Calibrate and review for fairness

Before anything is communicated:

• compare proposals across managers for similar roles, recording each change and why;
• compare ratings, increases and compa-ratio movement by sex and working pattern;
• check that people on maternity leave are treated as if not on leave [2], and that no rate falls below the minimum wage [3].

A rise for someone on maternity leave has two consequences, not one. Pay on return must include any increase she would have had [2]. Separately, where a rise is awarded, or would have been but for the leave, and applies to any part of the period from the start of the relevant period to the end of statutory maternity leave, her normal weekly earnings for SMP must be recalculated as if it had applied in each week of the relevant period (Alabaster; SMP (General) Regulations 1986, reg 21(7)) [11]. Tell payroll: SMP already paid is topped up.

In the lumi reward benchmark (n = 267), "No" is the common answer to reviewing pay decisions for fairness before implementation (45.7%). "Yes - always" (32.6%) and "Sometimes" (21.7%) are alternatives.

An equality impact assessment (EIA) records the change, who it affects, any group disadvantage and any mitigation. Government guidance on the public sector equality duty prescribes no process, but says "the easiest way to demonstrate compliance is using written records from the time of the decision" [4].

"No" is common (65.7%). "Major changes only" (17.4%) and "Always" (17.0%) are rare (n = 265).

Step 7: Document exceptions

Treat pay outside the range, outside the rule or off-cycle as an exception. Record the reason, evidence, approver and date. In an equal pay claim, the employer must show the material factor behind a difference [2].

In the lumi reward benchmark (n = 267), "Always" is the common answer to documenting exceptions with justification (50.2%). "Sometimes" is an alternative (49.1%) and "Never" is rare (0.7%).

Step 8: Get sign-off

Give the signatory the total cost, the spread of outcomes, the fairness review, any EIA, the exceptions and open risks. Record who signed, when and what they saw.

In the lumi reward benchmark (n = 265), "HRD" is the common answer (25.3%). "CEO/MD" (23.0%) and "Remuneration Committee" (21.5%) are alternatives. "Joint HR + Finance" (15.8%) and "Finance/CFO" (14.3%) are rare.

Step 9: Communicate outcomes and set a challenge route

Tell each employee their new pay, the effective date, how it was decided and how to raise a question. Where the change alters the pay particulars in their written statement, confirm it in writing within one month of the change [8].

In the lumi reward benchmark (n = 264), "None" is the common answer (65.5%), "Grievance route only" an alternative (24.6%) and "Formal appeal route" rare (9.8%).

Grievances follow the Acas Code. Tribunals can adjust awards by up to 25% for an unreasonable failure to follow it [9]. From 1 October 2026 the time limit for bringing an employment tribunal claim rises from 3 to 6 months [10].

Step 10: Audit the outcomes

Using final payroll data:

• Gender pay. Compare ratings, increases and compa-ratio before and after, by sex within each grade. Acas describes an equal pay audit for employers with 50 or more staff and a shorter equal pay review for smaller ones [11]. Gender equality action plans have been voluntary since 6 April 2026 and are due to become mandatory in 2027, date to be confirmed [10].
• Compression. Check grade differentials (Step 1), newer against longer-serving staff in the same role, and people at the maximum.

In the lumi reward benchmark (n = 267), "Yes" is the common answer to reviewing outcomes for compression and internal equity (40.8%). "Sometimes" (33.7%) and "No" (25.5%) are alternatives.

Checklist

• Calendar set around 1 April and the gender pay gap snapshot
• Data frozen and reconciled with payroll
• Differentiation, eligibility and range-maximum rules written down
• Managers briefed and proposals calibrated
• Outcomes compared by sex; maternity leave, SMP recalculation and minimum wage checks done
• EIA decision recorded
• Exceptions documented with reason and approver
• Sign-off and signatory pack recorded
• Pay changes confirmed in writing within one month
• Gender pay and compression audit completed

Related

• Pay ranges and pay review governance: UK benchmark 2026 — the benchmark findings behind the figures in this guide.
• How to publish salary ranges in job adverts
• How to design a bonus scheme: eligibility, measures, gatekeepers, malus and clawback
• How lumi works: methodology — how these figures are collected and calculated.