Hiring guideChecked against EEOC, ADA, OSHA and FMCSA sources, 15 September 2026

Hiring Criteria: How to Decide Who Gets the Job

Hiring criteria are the written, weighted standards you use to compare candidates, fixed before you meet the first one and applied the same way to everybody. For frontline roles it comes down to one separation: must-haves are what the law or the shift requires on day one — a valid licence, a medical certificate, the ability to work the Saturday that is actually on the schedule — and everything else is trainable, which means it gets a weight rather than a veto. Most employers invert this. They screen hard on years of experience, which is trainable, and wave through someone who cannot work the posted shift, which is not. That inversion is the most common reason a hire that felt right in the room is gone inside ninety days. Below: how to build the two lists, how to weight them, a worked grid, which red flags are real, and how to write criteria that hold up if the decision is challenged.

The decision you are actually making

Ask a shop owner why he hired the last technician and you get “I just liked him.” Push once and you get reasons: straightforward, dealership experience, seemed like he would show up. Those are criteria, and they are not unreasonable. The problem is that they were invented after the interview, to explain a decision made in the first four minutes.

Criteria written afterwards fail three ways. They cannot be compared, because everyone gets a different set. They cannot be defended, because nothing shows they existed before the decision. And they drift toward whoever is in the room.

This page decides what goes on that list. It is not the rubric — for the form you score on, take the interview scorecard template. If you have not written down who you are looking for, start with the ideal candidate profile.

Two buckets, and only two

Must-haves: what the law or the shift requires on day one

A must-have is pass or fail. It is not weighted, not traded against charm, and no interview recovers it. Three legitimate sources.

1. A regulatory requirement you can name. A driver in interstate commerce must be at least 21, hold a currently valid commercial licence issued by one state, be medically qualified, and read and speak English sufficiently to understand highway signs — all of it in 49 CFR 391.11. If you cannot cite the rule, this is not a legal must-have.
2. An essential function of the job. The ADA regulations define these as “the fundamental job duties” and exclude “the marginal functions,” and the evidence that separates them includes — tellingly — “written job descriptions prepared before advertising or interviewing applicants for the job” (29 CFR 1630.2(n)). A qualified individual performs them with or without reasonable accommodation, so an essential function is a statement about the work, not about a body.
3. The schedule as it actually exists. The one that gets skipped. Someone who can work two of the three required days has failed a must-have. Employers reject people for four years of experience instead of five, then hire the one who said “I can probably make Saturdays work.”

Trainable: everything else, including most of what you screen on

This bucket is larger than employers expect, and full of things used as vetoes. Forklift certification is the clearest case. Under OSHA’s powered industrial truck rules the employer owns the training, and before an employee operates a truck in your workplace you must evaluate that operator and determine them competent (OSHA powered industrial trucks eTool). You are doing the evaluation either way, so “must already be certified” is a preference that saves a training day — and with warehouse applicants at a median $9.83, an expensive one. Same for route software and charting systems.

Years of experience is the biggest item in this bucket and the one most often treated as a must-have. “Five years minimum” is a proxy for three things you could measure directly: can work unsupervised, has seen this go wrong before, will not leave in a month. Write those down and you can score them.

Weighting: the part everyone skips

After the must-haves you have five or six things that matter in different amounts. An unweighted list is one where everything matters equally, which is a way of saying nothing does. Weighting stops a candidate winning on what is easy to show in an interview and losing quietly on what decides whether they last until March. The method is crude on purpose: 100 points, four to six criteria, nothing below 10 and nothing above 30. The floor stops you carrying passengers; the ceiling stops one item becoming a hidden must-have.

Weight by what breaks. Take the last four people who left this role inside three months and ask what went wrong. In frontline work it is almost always the schedule, the reality of the work, or nobody training them — rarely a skill. Check your own first-90-days turnover before assigning a number.

A worked example: criteria grid for a home care aide

The role: personal care aide, private homes across one county, Tuesday, Thursday and alternate Saturdays, 7am to 3pm, two clients with limited mobility, one with early-stage dementia.

Must-haves, checked before anyone is scheduled for an interview:

• Can work Tuesday, Thursday and alternate Saturdays, 7am to 3pm, from the posted start date, confirmed in writing at pre-screen. Not “mostly.” Not “once childcare is sorted.”
• Valid licence and reliable insured transport — clients are 9 to 40 minutes apart, with no transit route between two of them.
• Clears whatever pre-employment screening your state requires for this role, run in the order the law requires: the legal order of checks.
• Can perform the essential physical functions with or without reasonable accommodation, written as tasks (assisting a transfer from bed to chair) rather than attributes of a person.

Weighted criteria, scored 1–5 where 3 means “meets the bar”:

Three candidates, weighted out of 500.

Candidate A — the one you liked, the one with the experience — is never scored, because the job is on Saturdays. That is the argument of this page in one row. Employers hire A anyway, and are surprised in October.

The grid also does not decide between B and C. Twenty points out of 500 is not a real gap. What it does is tell you what to check: for B, how long your transfer training really takes; for C, what happened in those three jobs, which is a question for the reference check. It also lets you explain the outcome in a sentence that is true, specific and about the job — the difference between a normal rejection email and one you would rather not have in writing.

Writing criteria that survive a challenge

The standard is the EEOC’s: a selection procedure that screens out a protected group must be job related and consistent with business necessity. The Uniform Guidelines cover “tests and other selection procedures which are used as a basis for any employment decision” — interviews and scoring grids included — and treat a selection rate below four-fifths of the highest group’s as evidence of adverse impact (29 CFR 1607.3 and 1607.4(D)). For a twelve-person shop that means four habits, not a validation study.

1. Write them before you advertise. The ADA regulations treat a job description prepared before advertising or interviewing as evidence of what is essential; one written after a dispute is evidence of nothing. That is the real reason to take the job description seriously.
2. Tie every criterion to a task, and use the same ones for everyone. “Lifts and repositions an adult with assistance” is a task; “physically fit” is a judgment about a person. Inconsistency is how a defensible set of criteria becomes an indefensible decision.
3. Keep the completed scores. You are keeping hiring records anyway, and the scored grids are the only contemporaneous evidence of what the decision rested on.
4. Do not let a criterion smuggle in a question you cannot ask. If scoring it requires knowing someone’s age, health, religion, family status or national origin, the criterion is the problem — see illegal interview questions. The EEOC’s summary of prohibited practices is plain: pre-employment information should be limited to what is essential to determining whether a person is qualified.

Red flags: which are real and which are folklore

Most lists of twenty candidate red flags are recycled. Here is the split.

Real, and worth acting on

• An unexplained gap in a licence or certification. Not a gap in employment — a gap in the credential. A nursing licence lapsed for eleven months, a CDL surrendered and reissued. There is usually an explanation and it is usually fine, and it is verifiable at the issuing board.
• A pattern of very short tenures with no external cause. Four jobs of 45 to 70 days, all ended by the candidate, none with a reason they can describe. One or two short stints mean nothing; the repeating shape with no story is what you ask about.
• Safety incidents blamed entirely on other people. Everyone has been in an incident. The signal is an account in which they had no part and learned nothing.
• Concrete claims they cannot describe. Someone who ran a machine daily for two years can describe its startup sequence and what always goes wrong with it. Vagueness about a task claimed as routine is the cheapest lie detector in interviewing.

Fake, and expensive to believe

• Employment gaps. Caregiving, a health event, a layoff, a stretch of cash work. Ask in one sentence and move on. Screening on gaps removes disproportionate numbers of women and older workers from a pool you paid for.
• Job hopping, in sectors where hopping is the norm. In warehousing, food service, home care and much of the trades, moving for a dollar an hour is rational. If the local market changes jobs every fourteen months, a candidate who does is average, not risky.
• A blanket bar on anyone with a criminal record. The expensive one. The EEOC’s guidance states that “the fact of an arrest does not establish that criminal conduct has occurred” and that an exclusion based on an arrest in itself is not job related. For convictions it points to three factors from Green v. Missouri Pacific Railroad — the nature and gravity of the offence, the time elapsed, and the nature of the job — and describes a targeted screen followed by an individualised assessment as the route to business necessity. Where a law requires exclusion for a specific role, that requirement stands and gives you a defence; the EEOC Q&A says so. Many states and cities separately regulate when you may ask — check yours before the question goes on an application form.
• “Overqualified” and nervousness. The first is frequently a proxy for age; if the worry is that they leave for more money, have the pay conversation out loud. The second is the correct response to sitting across from the person who decides whether you make rent.

Attitude versus experience: the honest verdict

“Hire for attitude, train for skill” is half right. Which half depends on one number you can measure: how long your operation takes to get someone from zero to independently productive.

Two warnings. “Attitude” has to be defined as behaviour or it is not a criterion: “positive attitude” scores nothing; “called ahead the one time they were late” scores something. And vague attitude criteria are where bias hides — the EEOC warns against decisions based on stereotypes and assumptions, and an unfalsifiable criterion is the perfect vehicle for one. To assess how someone works with your team, use defined behaviours and consistent questions: that is what culture-fit questions done properly are for, and it is not the same as deciding whether you like someone.

Fix the criteria, then leave them alone

Criteria are set once per opening, before the job is advertised, by whoever owns the role — one person and twenty minutes. Write them, weight them, put them where the panel can see them, and do not change them mid-process. Changing a weight after meeting a candidate is the after-the-fact justification this page started with, in a spreadsheet.

One legitimate reason exists to change them mid-search: the market has told you the person you described does not exist at the pay you are offering. Then change them in writing, for everyone, and re-score anyone already in the process. The signal is usually financial first. In our benchmark, CDL driver campaigns run a median $26.86 per applicant at an 8% apply rate, against $3.76 and 31% for caregiver roles and $13.88 across all 891 campaigns. Some of that gap is the licence. Some of it is requirements that were never really requirements. If four weeks of advertising has produced nobody who clears your must-have list, the list is a hypothesis that just failed.

The sequence itself is on hiring process steps and the interview on how to interview someone. Without criteria, a structured interview is a tidier way of deciding you liked someone. The grade your criteria get is what happens at day 90, the only honest definition of a quality hire.