Must-have vs nice-to-have: writing requirements that work

A must-have is a requirement whose absence would make you reject an otherwise excellent candidate; everything that fails that test is a nice-to-have. Run the split honestly, cap the must-haves at five or fewer, and score the two lists differently: must-haves as pass-or-fail gates, nice-to-haves as ranking points. The shorter the must-have list, the wider your funnel, because every requirement you publish filters out real candidates, not just unqualified ones.

What is the test for a true must-have?

Put every requirement through the same thought experiment. The strongest candidate you have met all year applies for this role, and they are missing exactly this one thing. Do you reject them?

An immediate yes means it is a must-have. Work authorization. Fluent Dutch for a Dutch-speaking support desk. Production experience with the core technology when nobody on the team has time to train. These survive because the answer does not depend on how good the rest of the profile is.

A “well, it depends” means it failed. Five years of React fails the moment you picture a sharp engineer with four. A degree requirement fails for nearly every role outside regulated professions. Your exact tool stack fails whenever the underlying skill transfers: someone who has run sprints in Linear will be fine in Jira within a week.

Requirements that fail the test are usually one of three things in disguise: a proxy (years of experience standing in for skill), a preference (the hiring manager’s favourite stack), or an inheritance (whatever the last job ad said). All three belong in the nice-to-have column, if they belong anywhere.

Why cap must-haves at five?

Because requirements multiply rather than add. Each must-have removes a slice of the qualified pool, and the slices compound. Stack ten gates in a row and almost nobody passes, and the few who do are not necessarily your best candidates; they are the people whose career happens to mirror the job ad.

Lists also grow by committee. The manager wants the predecessor’s skills back, the team wants its gaps covered, the director adds a strategic wish, and the result is a posting that demands senior depth in four specialisms at a mid-level salary. If you cannot get the list under seven after honest sorting, you are usually looking at two roles, or at a conversation about scope that needs to happen before sourcing starts. We cover that conversation in managing unrealistic hiring manager expectations.

Three to five is the working cap. It feels uncomfortably short. That is the point: the discomfort is the sign that you are actually choosing, and choosing is the job.

How do you sort requirements during intake?

Do not edit the list live in the intake meeting. Collect first, sort second. Get every wish on the table, then run three passes:

1. The rejection test on each item, as above. Be strict about hesitation: hesitation means nice-to-have.
2. The forced cut. Ask the hiring manager which three they keep if they can only keep three. The order in which items get sacrificed reveals the real priorities, which rarely match the original list order.
3. The training test. Could a strong hire pick this up in their first quarter? If yes, it is a nice-to-have by definition, whatever the manager’s instinct says.

The output slots straight into the ad and the screening plan, the same flow we describe in turning intake notes into a job description. From there, the two lists get different treatment at every step:

	Must-haves	Nice-to-haves
Definition	Missing it means a no, regardless of the rest	Adds value, never disqualifies
How many	Three to five	Five or six, ranked by weight
In the job ad	A short “what you need” list	A “bonus points for” section
In screening	Pass or fail gate, checked first	Weighted points for ranking
Candidate missing it	Polite, fast rejection	Slightly lower rank, still in the funnel

How should you score candidates against the split?

The split only pays off if your scoring respects it. Two steps, in order.

Gate first. Check must-haves as pass or fail, with no partial credit: if a candidate can clear screening with four of your five must-haves, the fifth was never a must-have. Candidates who fail the gate get a fast, honest rejection, which is better candidate experience than lingering at the bottom of a ranked list for three weeks.

Rank second. Everyone who passes gets scored on nice-to-haves, weighted by how much each one actually matters. This is where four years instead of five, an adjacent stack, or relevant sector experience earn their points.

Resist the blended single score. Average everything into one number and a candidate with seven bonus skills outranks one who actually clears the bar, which defeats the whole exercise. The same logic applies if your ATS scores applicants automatically: the mechanics are covered in candidate scoring in an ATS explained, but the must/nice split is the input that decides whether those scores mean anything. Ranking applicants against twelve undifferentiated bullets produces a precise-looking number with nothing behind it.

How do shorter lists widen the funnel?

Candidates treat requirement lists as instructions, not aspirations. A meaningful share will not apply unless they meet close to everything, and that tendency is commonly stronger in some groups than others, which is how an inflated list quietly narrows your funnel along exactly the lines you do not want. It is one of the patterns we flag in is your job description too generic or biased. Every requirement you cut keeps a set of capable people in the funnel who would otherwise have filtered themselves out.

Long lists also send a signal. A ten-bullet wish list reads as an employer who has not decided what the job is, or who expects a unicorn at a mid-market salary. Strong candidates, the ones with options, read that signal accurately and move on.

The benefit compounds downstream too: five gates screen faster than twelve, debriefs argue about things that matter, and rejections are easier to explain because the bar was explicit. A short list is not a compromise on quality. It is what an actual decision about the role looks like.

Where does Recruitifly fit?

Recruitifly’s assistant, Fly, scores and ranks applicants against a job and builds shortlists from the result, and the must/nice split is exactly the structure that makes those rankings worth trusting: hard requirements gate, everything else ranks. Fly proposes and you confirm every change before it happens, so a scoring run never quietly rejects anyone. We are in private beta at the moment; if you want to see your own requirements list run through this on a live role, talk to us.