The Nice-to-Have List Stopped Filtering and Started Attracting

What is a nice-to-have list actually for?

It is there to make some readers decide not to apply, and that is the whole of it. Nothing downstream is obliged to act on a line marked preferred, so the block's work happened in the reader, who counted the lines they matched and decided whether to bother. The cap-the-list rules and the soften-the-verbs rules are all rules about that arithmetic.

The folklore version of the reader is wrong in a way worth knowing, because it changes what the block is doing. The line everyone repeats, that men apply at 60% of the qualifications and women at 100%, comes secondhand: Tara Mohr's piece, routinely miscited as its source, attributes it to an internal Hewlett Packard report and is skeptical of it. Her survey of over a thousand predominantly American professionals was the counter-evidence. Asked why they had not applied for a job whose qualifications they did not fully meet, both sexes named the same top reason, twice as common as any other, "I didn't think they would hire me since I didn't meet the qualifications, and I didn't want to waste my time and energy" (41% of women, 46% of men), while "I didn't think I could do the job well" was the least common answer of all, around 10% of women and 12% of men ¹.

People were not being deterred by doubt about their own ability. They were reading the list as a set of eligibility rules and concluding they were not eligible. A preferred qualification, to that reader, looked like a requirement in a softer font.

Measured, the gender gap is far smaller than the folklore. In an online experiment with 10,468 working-age participants, the Behavioural Insights Team found men said they would apply at 52.1% of an advert's requirements and women at 55.7%, with the gap appearing only among the less qualified half of the sample and no difference among the more qualified ². Nobody applied to anything: the measure is stated willingness on a seven-point scale, in one fictional role, inside an online panel. The gap is still roughly a tenth the size of the number it replaced.

Why does the standard advice no longer do what it promises?

Because it assumes somebody who reads the list, compares themselves against it line by line, and withdraws. That person still exists. What has changed around them is the volume that arrives anyway, and the fact that the posting has itself become an input to the applications that do arrive rather than only a gate in front of them.

The volume side is visible in ATS data. Across more than 109 million applications and 247,000 jobs from January 2021 through March 2026, Ashby reports the average recruiter processing 291 applications per hire against roughly 100 in early 2021, with the share of applications resulting in an interview falling from about 7-8% in 2021 to between 3.6% and 4.7% depending on role type ³. That is one vendor's customer base, weighted toward venture-backed technology companies, and the numerator counts application records, so duplicates and automated submissions inflate it. Ashby names automated applications as one driver without sizing it. Treat the direction as real and the level as local, which is roughly what the tripling of applications per opening amounts to.

The softening advice has a second problem that predates all of this: the assumption that more careful wording communicates more. In the same Behavioural Insights trial, willingness to apply was significantly greater for the vague and the stereotypically masculine versions of the advert than for the version spelling out required behaviours in specific detail, and the specific version failed its own manipulation check. Participants did not rate it as any clearer than the others, even though across all conditions the clearer somebody found the advert, the more willing they were to apply ⁴.

Those are exploratory results from a trial whose ambiguity manipulation the authors say plainly did not work, so nobody should read them as detailed requirements repel applicants. The narrower reading is the useful one, and it is uncomfortable: writing requirements as specific behaviours did not, by itself, make the advert read as clearer. You cannot assume that adding detail has communicated anything at all.

Replace the block with one line, or cut it

One line, in your own words, naming what would make somebody unusually good at this particular job and what evidence would show it. It has to be a sentence you could say out loud to the hiring manager without reading it off a template. If you cannot write that line, there is no preference underneath the block worth a reader's attention, and the block comes off without replacement.

What goes in its place is concrete. LinkedIn's analysis of global paid postings from employers posting 100 or more jobs, covering March 2021 to March 2023, found posts listing skills in the requirements section correlated with an 11% higher view-to-apply rate than posts that did not ⁵. It is correlational, it measures one platform's view-to-apply rate and says nothing about who was worth interviewing, and LinkedIn sells skills-based hiring products. Take it as a weak signal pointing the same way as the rest: naming the work beats naming a proxy for the work.

The test for whether a preference belongs anywhere is the same test the must-have list has to pass, which is that every requirement needs a stage that produces evidence for it. Run the preferred lines through it and most of them evaporate on contact.

• *Experience in a regulated industry* survives if the screen asks what regulator, and what the person actually had to do differently because of it.
• *Familiarity with our tech stack* survives if somebody in the loop will ask a question only a user of it could answer. Otherwise it is a keyword, and keywords now come back to you free.
• *Startup experience* almost never survives, because nobody in the loop can say what it would look like if a candidate had it.
• *Advanced degree preferred* survives only where the degree is doing work the job actually needs, and dropping the line does nothing until something replaces it.

The lines that survive are not nice-to-haves. They are requirements you were not confident enough to state, and stating them is the improvement.

Check which preferred lines carry weight before cutting

Some preferred lines are load-bearing for reasons that have nothing to do with attraction. Government contracting language, a licence a client contract names, a clearance, an accreditation body's wording, an internal levelling rule: these get copied into the preferred block because nobody knew where else to put them, and deleting one can cost you a contract rather than a candidate.

So the cut is a two-pass job. First pass, mark every line you cannot explain the origin of. Second pass, ask the person who would know, which is usually legal, procurement or whoever owns the client relationship. Anything touching a statutory or contractual minimum belongs in front of counsel, and that is doubly true of lines about AI use, where the wording is new and little about it is settled. The legal shape of requiring AI experience in a posting is its own question.

Start with the last requisition that filled. Read its preferred block, and for each line ask which stage of that search asked about it, and who asked. Most lists produce the same result: nobody asked about most of them, one line turned out to be a real requirement that got softened to avoid deterring people, and one line nobody can explain the origin of at all.

The softened requirement is the expensive one. It got demoted to a preference because somebody was worried about applicant volume, then it did the deterring anyway and got tested anyway, so the demotion bought nothing and cost clarity for everyone in the loop. Promote it back, give it a stage, and say so plainly. Whether the same reasoning applies to an AI line specifically is worked through in required versus preferred AI skills, where the wording is newer and the temptation to hedge is stronger.

See a sample report >