How to Run a Job Evaluation Committee

How to Run a Job Evaluation Committee

A point-factor plan is only as good as the people applying it. You can build a beautiful factor structure with clean level definitions and defensible weights, and still end up with a job hierarchy nobody trusts — because one analyst scored 400 jobs alone in a spreadsheet over three weeks, and drifted somewhere around job 150.

The job evaluation committee is the fix. It is a small, trained, cross-functional panel that scores jobs against your compensable factors, resolves disagreements in the open, and leaves a written record of why each job landed where it did. Done well, it turns job evaluation from an HR opinion into a governed process. Done badly, it becomes a monthly meeting where the loudest VP re-argues his org chart. Here's how to run the first version.

TL;DR

• A job evaluation committee is a trained panel that scores jobs against your factor plan and reaches documented consensus — it is not a forum for negotiating individual salaries.
• Keep it to 5–7 voting members: comp/HR, plus operating leaders from different functions, plus (where relevant) employee or union representatives.
• Train everyone on the factor plan first, then calibrate on 8–12 benchmark jobs before scoring anything that counts.
• Use the EIGE pattern: members score independently first, then meet only on the sub-factors where they disagreed.
• Ontario's Pay Equity Tribunal explicitly tests whether a committee balances the parties' interests, is knowledgeable enough, and decides free of gender bias — so composition and training are compliance issues, not niceties.
• Write a one-page charter, log every score with a rationale, and publish an appeals route. Undocumented committees fail audits.

What a job evaluation committee is for

The committee has exactly one job: assign a defensible value to each job, using the plan you already agreed on. That value is about the job's content — the skill, effort, responsibility, and working conditions the role demands — not about the person in it, their performance, or what the market currently pays for the title.

That distinction is the single most important rule you will enforce as chair. Job evaluation is not performance evaluation. The moment someone says "well, Priya is exceptional in that role," you stop them and re-anchor on the job description. Market data comes later, when you price the structure — and it never overrides the internal hierarchy without a documented reason.

Three things the committee is not:

• Not a salary committee. It scores jobs. Pay ranges are a separate decision made after the point hierarchy exists.
• Not a title committee. Titles are marketing. If your titles are a mess, fix the job title hierarchy separately.
• Not a headcount forum. Whether a role should exist is an org design question, not an evaluation question.

Who should sit on it

Aim for 5–7 voting members. Below five, one strong personality dominates and you lose the reliability benefit of multiple raters. Above eight, sessions slow to a crawl and members disengage — you will feel this by hour three of your first workshop.

Two composition rules matter more than the org chart:

Cover the whole organization, not just the loud half. If every operating leader on your panel comes from engineering and sales, your plan will systematically undervalue administrative, care, and support roles — because nobody in the room can describe them accurately. This is the classic route to a gender-biased outcome, and regulators know it.

Balance the parties. In Ontario Nurses' Association v. Regional Municipality of Haldimand-Norfolk, Ontario's Pay Equity Hearings Tribunal set out five tests for whether a valuation tool was applied in a gender-neutral way. Three of them are about the committee itself: is it balancing the interests of the parties, is it sufficiently knowledgeable to allow the parties to meet their obligations, and is the decision-making free of gender bias? If you operate in Ontario, Quebec, or under the federal Pay Equity Act, your committee roster is a compliance artifact. Build it accordingly, and see our guide to gender-neutral job evaluation for the method-level controls.

Train first, score second

Never let a committee score a real job in its first session. Untrained raters anchor on titles, salaries, and the people they know, and those anchors are almost impossible to unwind later.

Run a half-day training workshop covering:

1. The four core factor families and your specific sub-factors — what each one measures and, critically, what it does not measure. (If your sub-factors are still fuzzy, start with compensable factors.)
1. The level definitions inside each sub-factor, read aloud, with examples of what a level 2 versus a level 4 looks like in your organization.
1. The weights and why they are what they are — see how to weight compensable factors.
1. Common bias traps: scoring the incumbent, scoring the title, scoring physical effort generously while under-scoring emotional demand and interpersonal complexity, and scoring "how hard it is to hire" instead of "what the job requires."

Then calibrate on 8–12 benchmark jobs that span your full range — a frontline operational role, an administrative role, a professional individual contributor, a first-line manager, a senior specialist, and a functional leader. Score them, compare, and argue. The point of calibration is not to get the "right" answer on those twelve jobs. It is to discover that Member C reads "problem solving level 3" completely differently from everyone else, and to fix that before it contaminates 300 scores.

Running your first committee and don't have a factor plan to train on? PointFactors ships with a validated point-factor structure and level definitions your panel can calibrate against on day one — see it in a demo.

The scoring session: independent first, consensus second

The highest-reliability pattern is the one EIGE documents in its EU-wide gender-neutral job evaluation toolkit: every member scores every job profile independently, assigning a level for each sub-factor and writing down the reason. The chair collects all proposed scores. Then the committee meets — but only on the sub-factors where members actually disagreed.

That last part is what makes committees affordable. In EIGE's worked example, a five-member panel scoring a kitchen porter role agreed unanimously on problem-solving (all level 2) and split only on physical effort (four members at level 3, one at level 4). There is nothing to discuss on problem-solving. The panel spent its two half-day sessions on the genuine splits, concluded that frequent lifting of up to 20 kg across 8–10 hour shifts justified level 4, and moved on.

A practical cadence that works:

• Batch size: 15–25 jobs per round. Independent scoring takes members roughly 10–15 minutes per job once trained.
• Session length: three hours maximum. Rater fatigue is real and it drifts scores downward.
• Disagreement threshold: discuss any sub-factor where members are more than one level apart, or where the split affects the job's grade placement. Ignore one-level splits that don't move the total across a grade boundary.
• Decision rule: consensus, meaning every member can live with the outcome. Write down what you'll do if consensus fails — usually the chair records both positions and escalates to a comp governance body rather than forcing a vote.

Then group the scored jobs. Most plans cluster jobs into grades using a band of total points; EIGE's standard approach uses a 10% range. Two jobs from unrelated departments landing in the same grade is a feature, not a bug — that is internal equity doing its work.

Document everything, or it didn't happen

The Ontario Tribunal's standard is that the parties made a reasonable effort to capture job content and applied the four factors — not that they achieved perfection. But you can only demonstrate reasonable effort if you wrote things down.

Minimum record per job:

• The job description version used, with date
• Each member's independent sub-factor scores and stated reasons
• The final agreed level per sub-factor, with the rationale where it differed from an individual score
• The total points, the resulting grade, and the date of decision
• Any dissent

Add a one-page charter before your first session covering scope, membership and terms, quorum, the decision rule, confidentiality, and the appeals route. And publish the appeals route: a manager who thinks a job was mis-scored should submit updated job documentation, not lobby a committee member in a hallway. Appeals are re-evaluations of evidence, and they follow the same process as any other job.

Common failure modes

Drift. Scores from month nine don't match scores from month one. Fix it by re-scoring three benchmark jobs at the start of every fourth session and comparing to the original results.

The absent expert. A member misses two sessions, returns, and re-litigates settled jobs. Fix it with a quorum rule and a "no reopening without new job information" clause in the charter.

Salary leakage. Someone shares what a job currently pays and the room scores backwards from it. Fix it by keeping pay data out of the scoring session entirely — the chair holds it.

The one-person committee. A single analyst scores everything and the "committee" rubber-stamps it. This is the most common failure in mid-market companies, and it fails every regulatory test in the Ontario framework. If you don't have the bandwidth for full independent scoring, run software-assisted first-pass scoring and have the panel review and adjust — but the panel must genuinely be able to change the answer.

FAQ

How many people should be on a job evaluation committee? Five to seven voting members is the sweet spot. You need enough raters that no single perspective dominates, and few enough that sessions stay productive. Larger organizations often run several parallel panels by job family, with one cross-panel calibration group to keep scoring consistent.

Should employees or union representatives sit on the committee? In unionized environments and in pay equity jurisdictions like Ontario and Quebec, yes — regulators explicitly test whether the committee balances the parties' interests. Elsewhere it is optional but usually worth it: employee representation improves both accuracy on frontline roles and acceptance of the results.

How long does it take to evaluate a job? Budget 10–15 minutes per job per rater for independent scoring once the panel is trained, plus consensus time on contested sub-factors. A trained committee can process 15–25 jobs in a three-hour session. Your first two sessions will be half that speed.

What if the committee can't agree? Define the fallback in your charter before you need it. The usual approach: the chair documents both positions, the job is provisionally graded at the lower score, and the disagreement escalates to a compensation governance body or the plan's owner. Avoid majority votes — they convert a methodology question into a political one.

Can software replace the committee? It can replace the manual scoring grind, not the governance. Tools like PointFactors produce consistent first-pass scores from job documentation in minutes, which removes drift and clerical error. The panel still owns the review, the exceptions, and the record — and regulators still expect a human decision trail. See our comparison of job evaluation software.

How often should the committee meet after the initial project? Quarterly is typical for maintenance: new jobs, materially changed jobs, and appeals. Plan a full re-validation of the factor plan every three to five years, or sooner after a merger or a major restructure.

Does the committee decide pay? No. It decides job value in points. Pay ranges are built afterward from the point hierarchy plus market data, which is a separate exercise with a separate approval path.

Getting started

If you're standing up a committee this quarter, the order is: write the charter, pick 5–7 members who actually cover your organization, run a half-day training workshop, calibrate on a dozen benchmark jobs, then start scoring in batches of 20. Document every score with a reason. That sequence is boring and it is the whole game — the committees that fail are the ones that skipped straight to scoring.

Want the factor plan, the level definitions, and the scoring records handled for you? PointFactors gives your committee a validated point-factor structure, AI-assisted first-pass scores, and a complete audit trail of every decision — so your panel spends its time on the jobs that are genuinely contested instead of the 80% that aren't. Book a demo or see pricing.

Justin Hampton is founder and CEO of PointFactors, an AI-powered point-factor job evaluation platform for HR and compensation teams.