We tested 16 things that supposedly make a job advert competitive. Fourteen did nothing.

We had a great story. Adverts that say "no experience needed" were pulling fewer applications than adverts that demand experience, which would mean the jobs opening the door widest are the ones nobody walks through. Counterintuitive, useful to both sides of the hiring desk, and completely wrong. This is what happened when we tested it properly.

Jobseekers are told constantly that certain phrases decide how hard a job is to get. No experience required. Full training provided. Salary stated. Remote working. Agency advert. It is repeated in career advice, in LinkedIn posts, and by us, and almost nobody checks it, because checking requires knowing how many people actually applied.

We can see that number. Reed publishes an application count on live adverts, so we hold 1,846 fresh UK adverts across 21 occupations with the size of the queue attached to each one. That is enough to ask which words go with a long queue and which go with a short one.

The finding that died

The first version of this analysis pooled every advert together and compared those saying "no experience needed" or "full training provided" against everything else. The open-door adverts had a median of 10 applications, the rest had 16. A 37% gap, in the surprising direction, ready to write up.

Then we did the thing that breaks most findings like this. Application counts vary far more between occupations than within them. A receptionist advert and a nursing advert differ by more than any phrasing ever will, so comparing all adverts with a phrase against all adverts without it mostly measures which occupations happen to use it. Care and warehouse work say "full training provided" constantly. Nursing does not. The gap we found was largely a map of which industries use the phrase.

Redone inside each occupation, on fresh adverts only, with duplicates collapsed:

Nothing. The interval sits across zero, and the individual occupations point in opposite directions: warehouse operative adverts offering training drew 15.5 more applications than warehouse adverts that did not, while administrator adverts drew 8.5 fewer.

One more check finished it off. Splitting by pay, the direction flips: below £25,000 the open-door adverts drew 20 applications against 11, and between £25,000 and £35,000 they drew 5 against 8. A finding that changes sign depending on how you slice it is not a finding.

What survived

Two things held up across enough occupations to be worth reporting.

Adverts requiring a driving licence were consistently less contested. A median of 5.75 fewer applications, holding in five of the six occupations where we could test it, with a confidence interval of -9 to -0.25. The gaps inside occupations are large: customer service adverts requiring a licence had a median of 1 applicant against 11 for those that did not, senior project manager 4 against 12, senior sales manager 2.5 against 10.

That result needs a caveat carried right next to it rather than buried at the bottom. Roles requiring a licence often involve travel, field work or awkward geography, so they may differ from desk roles in the same occupation group in ways that have nothing to do with the licence. We are not claiming that adding "driving licence required" to an advert would thin the queue. We are saying these adverts were consistently less contested, and that occupation is a coarse control.

Customer-facing language went with fewer applications too. Three fewer, holding in 10 of the 14 occupations tested, interval -6.5 to -0.5. Smaller effect, more occupations behind it.

Agency adverts looked like the strongest result of the lot, at +3.5 applications with an interval of +2 to +5.5 and 11 of 12 occupations pointing the same way. We are not publishing it as a finding, because agency adverts in our sample are a median of 14 days old at collection against 11 for direct employers, and three extra days on a board is worth applications on its own. When we split by advert age the sample fragmented and the direction went mixed in the freshest band. It might be real. The data we have cannot separate it from the age gap.

Everything we tested

Sixteen features, every one declared before the analysis ran and frozen in a dictionary file so the same test can be repeated. Negative means adverts with the feature drew fewer applications than adverts without it in the same occupation.

Fourteen of the sixteen either cross zero or rest on too few occupations to call. Salary visibility, remote working, weekend work, DBS checks, advert length, career progression, part-time hours: none of them separated cleanly once occupation was held constant. Immediate start looks convincing at -2.5 with a tight interval, but it only had two occupations behind it, which is not enough to say anything.

Why the null result is the point

Recruitment folklore sounds convincing because it is built on vivid examples. Everyone has seen an entry-level advert with 400 applicants. Everyone has seen a remote job go mad. Large datasets are useful precisely because they force those examples to survive a comparison against similar roles, and most of them do not.

The practical version for anyone applying: the wording of an advert tells you far less about your odds than the occupation does. We have measured that gap before, and it is enormous. A receptionist advert draws a median of 37 applicants and a nursing advert draws 1, a difference no phrase in this table comes close to. Choosing which queue to join beats reading the tea leaves in the advert copy.

For anyone writing adverts, the same result in reverse. Your phrasing is not what determines whether you get 4 applications or 40. If a role is not attracting applicants, the fix is more likely to be in the occupation, the pay, the location or the hours than in whether the advert promises career progression.

Method

Sample: 3,999 unique Reed adverts, of which 2,979 carry an application count, 2,216 were posted within 30 days of collection, 1,895 survive near-duplicate collapsing on employer plus title plus advert opening, and 1,846 sit in one of the 21 occupations holding at least 30 adverts. A feature is tested inside an occupation only where both arms hold at least 10 adverts.

Effects are computed as the difference in medians inside each occupation, then combined across occupations. Confidence intervals come from a bootstrap that resamples occupations rather than adverts, because adverts inside one occupation share a market and are not independent evidence. Feature definitions are frozen in a dictionary file with a version number, so every figure here can be regenerated exactly.

Limits worth stating plainly. This is cross-sectional: it describes adverts as they stood when we collected them, not how their queues grew. Application counts are what the board reported at that moment. Occupation is a coarse control, and within any occupation there are real differences between roles that our features do not capture. And this is one job board over one month, which is a snapshot rather than a law.

The wider dataset behind this work, 15,291 UK adverts in total, is described in our ten findings from 15,291 job adverts, and the occupation-level competition gap is in where the queue is shortest.

Frequently asked questions