Everybody reads the salary line for the number. There is a second thing in it that nobody reads: the shape. A band of £45,000 to £65,000 and a flat £12.21 an hour are two different messages about how much of the conversation is already over, and the shape sorts UK jobs almost perfectly by class.
We measured 3,925 unique UK job adverts across 33 occupations. Not what they pay, but whether they state pay at all, whether they state a band or a single figure, and how wide the band is.
Three quarters of adverts state a pay figure of some kind, which is more than most people expect. Of the adverts that state both ends of a band, 35.4% state the same number twice. That is not a range. That is a rate.
The lower the pay, the more exact the number
Cleaner adverts quote one fixed figure 86% of the time. Care worker 74%, support worker 72%, healthcare assistant 51%, receptionist 49%. Now the other end: AI engineer adverts quote a single figure in 0% of cases, teacher 6%, data scientist 3%, software engineer 13%.
Band width does the same thing in reverse. Measured as a share of its own midpoint, so a £40,000 to £60,000 band is 40% wide, the median UK band is 15.9%. Care worker bands run at 4.1%, cleaner 5.7%, healthcare assistant 6.5%. Sales executive bands run at 70.3%, teacher 41.6%, AI engineer 40.0%.
Across the 23 occupations where we can measure both, the correlation between median pay and band width is +0.53, and between median pay and the share of adverts quoting one exact figure it is -0.60. The better paid the occupation, the more room the advert leaves.
| Occupation | Median pay | Band width | One figure |
|---|---|---|---|
| Sales executive | - | 70.3% | 10% |
| Teacher | £49,400 | 41.6% | 6% |
| AI engineer | £37,500 | 40.0% | 0% |
| Data scientist | £75,000 | 26.1% | 3% |
| Software engineer | £50,000 | 20.0% | 13% |
| Senior sales manager | £42,500 | 18.2% | 21% |
| Product manager | £58,750 | 17.1% | 12% |
| Warehouse operative | £28,256 | 17.1% | 39% |
| Nurse | £41,500 | 17.0% | 22% |
| Marketing manager | £38,013 | 16.7% | 31% |
| Teaching assistant | £25,415 | 16.2% | 2% |
| Senior project manager | £50,000 | 15.4% | 35% |
| Graduate | £26,000 | 14.6% | 3% |
| Customer service | £27,887 | 14.5% | 55% |
| Retail assistant | £22,000 | 13.5% | 63% |
| Operations manager | £39,850 | 11.3% | 26% |
| Administrator | £27,625 | 10.2% | 52% |
| Receptionist | £26,426 | 8.0% | 49% |
| Support worker | £25,760 | 7.6% | 72% |
| Healthcare assistant | £27,040 | 6.5% | 51% |
| Cleaner | £25,584 | 5.7% | 86% |
| Care worker | £26,208 | 4.1% | 74% |
Band width is the gap between the two ends as a share of the midpoint. "One figure" is the share of adverts stating pay that state the same number at both ends. Sales executive pay is blank because that pull publishes no pay period, so the figure cannot be annualised.
What the shape is actually telling you
A single figure almost always means the rate was decided somewhere above the person hiring you. A national minimum wage, an NHS band, a local authority scale, a franchise rate card, an agency margin. The advert is not being unfriendly. It is reporting a decision it did not make.
A wide band means the opposite: the employer has not decided yet, and intends to work out what you cost from what you say. That is more freedom and more risk at once. It is also why the widest bands cluster in sales and engineering, where the same job title covers wildly different levels of experience.
The practical consequence for anyone applying. If the advert states one exact figure, do not spend your energy planning a negotiation on the headline rate. Spend it on the things that are still loose: hours, shift premiums, contracted overtime, the start date, when your first review happens, and whether training is funded. If the advert states a band wider than about a fifth of its midpoint, the number in your head matters, because they are going to ask what you are looking for and the answer will anchor the offer.
One honest exception in the table. Teaching assistant adverts almost never quote a single figure, and yet nothing about that job is negotiable. The reason is that TA pay is quoted as a full-time-equivalent scale then paid pro rata for term time, so the advert carries a range that is really an arithmetic artefact. Where an occupation is paid on a published scale, read the band as a scale rather than as an invitation.
Three quarters state pay, and that is the good news
75.6% of the adverts in this sample state a pay figure. Pay transparency campaigns tend to talk as though the number is usually missing, and on this evidence, in these occupations, on this board, it usually is not.
The gap is not really about disclosure any more. It is about precision. An advert saying "£30,000 to £60,000 depending on experience" has technically disclosed pay and has told you almost nothing, and one in seven of the bands we measured is wider than 40% of its own midpoint.
Method
3,925 unique Reed adverts from three collection pulls in July 2026, deduplicated across pulls, with nursery nurses removed from nursing sets. Occupations are reported where at least 30 adverts exist. An advert enters the width figures only if it states both ends of a band; adverts with no pay figure are left out rather than guessed at.
Width is reported as a share of the midpoint rather than in pounds, because one of the three pulls publishes no pay period and a ratio does not need one. Where a period is published, hourly and daily rates are annualised at 37.5 hours a week and 52 weeks, which is a convention rather than a fact about any specific job. Median pay in the table is the midpoint of the stated band, annualised the same way.
Limits. This is one job board over one month, and Reed is stronger in commercial and care hiring than in, say, academia. Occupation labels come from the search term each pull used, so a couple of jobs appear under two names and are deliberately not merged. Advertised pay is what the employer published, not what anyone was paid. And we checked whether a wide band goes with more applications: the difference was about one application and the sign flipped between occupations, so there is nothing to report there.
The wider corpus behind this work is described in ten findings from 15,291 job adverts. We also watched adverts hourly to see how fast an application queue really builds.
Frequently asked questions
Why do some job adverts give one salary figure and others give a range?▾
A range usually means the employer expects to negotiate, or expects to pitch the offer against your current pay and experience. A single figure usually means the rate is set somewhere else, by a pay scale, a framework or a national minimum, and the person hiring cannot move it. In our sample of 3,925 UK adverts, 35.4% of the adverts that stated both ends stated the same number twice.
Can you negotiate a UK salary if the advert gives one number?▾
Sometimes, but the advert is telling you the odds are poor. Where a single figure dominates an occupation it is usually because pay is set by a scale or a rate card rather than by the hiring manager. In those roles the negotiable parts are hours, shift premiums, contracted overtime, start date and progression timing, not the headline rate.
How wide is a typical advertised UK salary band?▾
The median band spans 15.9% of its own midpoint, so roughly £4,000 on a £27,000 job. A quarter of bands are under 9.6% and a quarter are over 28.6%. About one in seven bands is wider than 40% of its midpoint, which is where the number stops being informative.
How many UK job adverts state pay at all?▾
75.6% of the 3,925 adverts we measured state some pay figure. That is higher than most people assume, but it varies enormously by occupation: teaching assistant, support worker and teacher adverts state pay almost always, while marketing manager adverts state it half the time.
Does a wide salary band mean more competition?▾
We checked and could not find a stable answer. Comparing adverts above and below their own occupation median band width, the difference in applications was about one application and the direction flipped between occupations. We are not publishing it as a finding.