How to read hospitality salary surveys without misleading comparisons

Two hospitality salary surveys can give different answers without either being wrong. One may describe advertised packages for senior London hotels; another may measure actual base pay across a wider mix of employers.

Before using either figure to propose a pay change, establish what was measured. A useful benchmark compares sufficiently similar jobs, working patterns and reward components, and makes the remaining differences visible.

Match the work before matching the title

“Operations manager” can mean responsibility for one restaurant, several outlets or most of a hotel’s departments. Job titles alone do not establish comparability.

Create a short role-matching sheet covering direct reports, budget responsibility, operating hours, department scope, decision authority and required expertise. Note whether the role includes overnight responsibility, a substantial events operation or a multi-property remit.

Then record how well the survey definition fits: close match, partial match or unsuitable. Do not hide a partial match behind an impressive sample size. Where the external role has materially broader accountability, it may provide context without being a direct pay target.

WINC Wire’s discussion of job families and salary grades provides related organisational context. The task here is narrower: establish whether an external observation belongs in this particular comparison.

Identify the pay measure and its ingredients

The Office for National Statistics’ ASHE methodology guide distinguishes hourly, weekly and annual measures, including versions with or without overtime. A number labelled “earnings” should not automatically be read as basic annual salary.

For each survey, record whether the figure includes:

• Contractual base pay.
• Overtime, shift premiums or on-call payments.
• Guaranteed allowances.
• Bonus, commission or another variable payment.
• Tips, tronc or service-charge distributions, and how they are defined.
• Accommodation, meals or other benefits, and whether these are valued as cash equivalents.

Do not assume that different publishers treat any of these components identically. Ask whether variable amounts represent actual payments, a target opportunity or an optimistic maximum. A package with higher possible earnings may still provide less guaranteed income.

Keep base pay, guaranteed cash, variable cash and non-cash benefits on separate lines. That preserves information a single “total package” figure can conceal. This is comparison guidance, not a determination of how particular payments must be treated for legal or tax purposes.

Put hours on the same footing

Hypothetical example: Job A pays £31,200 for 40 paid hours each week over 52 paid weeks. Job B pays £30,000 for 37.5 paid hours each week over the same 52 weeks.

• Job A: £31,200 ÷ (40 × 52) = £15.00 per paid hour.
• Job B: £30,000 ÷ (37.5 × 52) = £15.38 per paid hour, rounded.

Job A has the higher annual salary; Job B has the higher hourly equivalent on these assumptions. Neither calculation includes overtime or unpaid additional hours. Do not present the conversion as an audit of actual hours worked or minimum-wage compliance.

A full-time-equivalent conversion answers another question. If a role pays £18,000 for 22.5 hours, and comparable full time is 37.5 hours, the equivalent annual base is £18,000 × 37.5 ÷ 22.5 = £30,000. Actual annual pay remains £18,000.

Use the relevant full-time hours, not an assumed universal standard. Seasonal annualisation also needs care: a short contract’s hourly rate does not establish guaranteed year-round earnings. Show paid weeks and guaranteed hours explicitly.

Distinguish the median from the mean

The median is the middle observation when pay is ordered; the mean adds pay and divides by the number of observations. The ONS guide to interpreting ASHE explains why the median is less affected by a small number of high salaries.

Hypothetical five-job sample: annual base salaries of £28,000, £29,000, £30,000, £31,000 and £60,000 have a median of £30,000 and a mean of £35,600. A claim that “average pay is £35,600” is mathematically correct but may suggest a typical salary that few people in this sample receive.

Ask which average is reported. Read quartiles or other percentiles where available, and do not treat the highest observed salary as a market rate. A percentile describes a position in the surveyed distribution; it does not prescribe where your organisation must pay.

Inspect who entered the sample

Find the number of observations for the specific role and location, not only the survey’s overall respondent count. Ten thousand responses across an industry do not guarantee a reliable result for one specialist hotel role.

Ask whether observations represent employees, jobs, employers, vacancies or recruiter placements. A large employer contributing many records may influence an employee-weighted result differently from a survey giving each employer equal weight. Duplicate adverts and repeated placements can also distort apparently large datasets.

The ASHE quality and methodology report explains sampling and non-sampling error, including coverage and non-response issues. Published uncertainty indicators matter: small or suppressed cells should not be turned into precise salary recommendations.

For commercial surveys, request the methodology if it is missing. An unexplained number may be useful as a lead for further investigation, but it is weak evidence for a specific adjustment.

Check place, date and composition

Separate workplace location from employee residence, and local labour markets from national averages. Compare like operating settings where possible. If an exact local match is unavailable, describe the compromise rather than inventing a location premium.

Record the survey’s reference period, publication date and revision status. A recently published report can contain much older observations. Do not automatically increase an old benchmark by consumer inflation; changes in living costs and changes in market pay are different quantities.

Check whether a year-on-year movement reflects comparable roles and respondents. If a survey adds more senior posts, its median can rise without any continuing employee receiving a raise. Avoid averaging incompatible surveys together simply to produce a single reassuring number.

Use a one-page evidence note

For every benchmark recommendation, retain:

1. The role match and any material differences.
2. The source, reference date, location and sample unit.
3. The pay components, hours and conversion assumptions.
4. The selected statistic and available uncertainty information.
5. Other relevant evidence, such as recruitment outcomes, with its limitations.
6. The proposed decision and what evidence would change it.

Keep market evidence separate from internal fairness, affordability and reward design. WINC Wire’s guide to employee reward design offers a wider perspective on those choices. A salary survey is most useful when it informs a judgement, rather than disguising that judgement as an exact answer supplied by the market.

Sources

• ONS: ASHE methodology and guidance
• ONS: Guide to interpreting ASHE estimates
• ONS: ASHE quality and methodology information

AI-generated editorial illustration.