How to Design a Job Evaluation Plan (Step by Step)

How to Design a Job Evaluation Plan (Step by Step)

Most comp teams inherit a job evaluation plan. Someone built it in 2011, the consultant who designed it left, and nobody remembers why "Problem Solving" is worth 180 points while "Working Conditions" is worth 40. The plan still runs, but nobody can defend it — and a plan you can't defend is a liability the first time a manager appeals a grade or a regulator asks how you decided two jobs were equal.

This guide walks through designing a point-factor job evaluation plan from a blank page: picking factors, writing degree scales, setting weights, allocating points, and pressure-testing the whole thing before you score a single real job. Budget six to ten weeks for a first plan. The design work is genuinely hard, but it is finite, and you only do it once every several years.

TL;DR

• A job evaluation plan has four parts: compensable factors, degree definitions, factor weights, and a point scale. Design them in that order.
• Eight to twelve sub-factors is the practical range. Fewer than six can't separate jobs; more than fourteen creates rater fatigue and false precision.
• Weights encode your strategy. Set them with the leadership team, in percentages that total 100%, before you convert anything to points.
• Test the draft plan on 15–25 benchmark jobs you already agree about. If the plan ranks them in an order that surprises you, fix the plan — not the jobs.
• Document every decision. The rationale is what makes the plan defensible later.

What a job evaluation plan actually is

A job evaluation plan is the scoring instrument your organization uses to measure the relative internal worth of jobs. In a point-factor system, the plan consists of four components:

1. Compensable factors — the dimensions every job is measured on.
1. Degree definitions — the levels within each factor, written so two trained raters land on the same one.
1. Weights — how much each factor contributes to the total.
1. A point scale — the numeric values attached to each degree.

Together these are sometimes called the point manual. Everything downstream — grades, ranges, career paths — depends on this document being right.

Worth stating plainly: job evaluation measures the job, not the person in it. It is not performance evaluation, and it does not consider how well the incumbent performs, how long they have been there, or what the market pays. Market data enters later, when you convert points into pay grades.

Step 1: Decide what the plan has to do

Before choosing factors, write down the plan's job. Three questions settle most design arguments later:

How many jobs, and how varied? A single-site manufacturer with 60 job titles needs a different instrument than a 4,000-person health system spanning clinical, technical, administrative, and skilled-trades work. Broad populations need factors that travel — "Problem Solving" works everywhere, "Exposure to Hazardous Materials" does not.

One plan or several? Organizations with genuinely different work sometimes run separate plans for, say, production and professional populations. This is legitimate, but it costs you the ability to compare across plans — which is exactly what pay equity analysis requires. Default to one plan unless you have a strong reason.

What's the legal exposure? If you operate in Canada, the EU, or a US state with proactive pay equity obligations, your plan needs to be gender-neutral by construction: factors that recognize work typically done by women (caring, coordination, emotional effort, dexterity) alongside the traditional male-coded ones (physical effort, budget accountability, supervision). The EEOC has been explicit that setting pay for female-dominated jobs below what your own job evaluation study suggests, while male-dominated jobs track the study, is evidence of compensation discrimination under Title VII (EEOC guidance). Your own plan can be used against you. Build it accordingly.

Step 2: Select compensable factors

The four statutory umbrella categories — skill, effort, responsibility, and working conditions — come from the Equal Pay Act of 1963 and still define the legal shape of the exercise. The EPA requires equal pay for jobs demanding substantially equal skill, effort, and responsibility under similar working conditions. Almost every credible plan organizes its factors under those four headings.

Inside them you choose sub-factors. A typical professional-services plan might look like this:

Three rules for selection:

• Every factor must be present in every job, in varying amounts. If a factor scores zero for 70% of your population, it is not a compensable factor — it is a job characteristic. Drop it or fold it into another.
• Factors must not overlap. "Years of experience" and "depth of technical knowledge" measure nearly the same thing, and including both double-counts it. Overlap is the single most common defect in inherited plans.
• Eight to twelve sub-factors is the sweet spot. Six or fewer and jobs bunch up; more than fourteen and raters start guessing. If you want the long list of options with definitions, see our breakdown of compensable factors.

Step 3: Write degree definitions

This is where plans succeed or fail. A degree is a level within a factor — usually four to six of them, arranged from lowest to highest. The test of a good degree scale is simple: two trained raters, reading the same job description, pick the same degree without discussing it.

Weak degree language uses relative words: "moderate complexity," "significant impact," "considerable experience." Those mean different things to different raters. Strong degree language uses observable anchors:

Financial Accountability — Degree 3: Recommends spending within an approved budget of $250K–$2M annually; approves individual transactions up to $25K; errors are typically caught in the normal review cycle.

You don't need the same number of degrees in every factor. Working Conditions might need three; Decision Authority might need six. Use only as many as you need to separate real jobs — extra degrees create the illusion of precision. Our guide to writing factor degree definitions has more examples and the common failure patterns.

Designing a plan from scratch and want the structure handled for you? See how PointFactors builds a factor plan in an afternoon instead of a quarter.

Step 4: Set weights before you set points

Weights are a strategy decision, not a technical one. They state, in numbers, what your organization pays for. A hospital system that weights Responsibility at 40% is saying something different from a research lab that weights Skill at 45%.

Run this as a facilitated session with the leadership team, not a spreadsheet exercise in the comp function. Ask each participant to allocate 100 points across the umbrella categories first, then across sub-factors within each. Discuss the spread. Converge. A plausible professional-population outcome:

Two cautions. First, do not set weights by reverse-engineering your current pay structure — that just launders existing inequities into a system that now looks objective. Second, expect to revisit weights only every three to five years, because re-weighting means re-scoring everything. Our guide on weighting compensable factors covers the arithmetic and the political dynamics.

Step 5: Build the point scale

Now convert weights to points. Pick a total maximum — 1,000 is conventional and easy to reason about. A factor weighted 15% has a 150-point maximum. Distribute those points across that factor's degrees.

You have two choices for the progression between degrees:

• Arithmetic: equal increments. A 150-point factor with five degrees becomes 30 / 60 / 90 / 120 / 150. Simple, transparent, easy to explain to a skeptical manager.
• Geometric: accelerating increments, e.g. 20 / 40 / 70 / 110 / 150. This stretches the top of the scale, which better reflects how pay actually behaves as scope grows.

Most plans use arithmetic progression for clarity and accept a slightly compressed top end, then handle the acceleration when they build pay ranges. Whichever you pick, apply it consistently across factors — mixing progressions inside one plan makes point totals uninterpretable.

Step 6: Test on benchmark jobs, then fix the plan

Do not roll out an untested plan. Pick 15–25 benchmark jobs that are well-documented, stable, and spread across every level and function. Critically, pick jobs whose relative standing your leadership already agrees on.

Score them. Then look at the rank order.

If the plan puts a senior engineer below a first-line supervisor and everyone in the room thinks that's wrong, you have a design defect — usually an under-weighted Skill factor or a degree scale that tops out too early. Adjust the plan and re-score. Two or three iterations is normal.

What you must not do is adjust the scores to reach a predetermined answer. The moment you start nudging individual jobs to land where you want them, the plan stops being an instrument and becomes a rationalization. Fix the rule, never the result.

Also run an inter-rater check: have two evaluators independently score the same five jobs. If they disagree by more than about 10% of total points, your degree definitions are too vague. Go back to Step 3.

Step 7: Document and govern

Write the point manual as a real document: each factor defined, each degree spelled out, the weighting rationale recorded, the benchmark test results attached. Then decide who owns it. A standing job evaluation committee with representation beyond the comp team is the usual answer, and it does two things — improves scoring consistency, and gives you a defensible process record.

Set a maintenance cadence now, while you are thinking about it: score new and materially changed jobs as they arise, audit a sample annually, and review factors and weights every three to five years.

How common is this, really?

WorldatWork's Job Evaluation and Market Pricing Practices survey (n=587) found 84% of organizations have an established method for evaluating jobs, but point-factor is the primary method for only about 16–21% depending on job family, with market pricing dominant at roughly two-thirds (WorldatWork, 2020).

That gap is the opportunity. Market pricing tells you what a job costs; it cannot tell you whether two differently-titled jobs are of equal value — which is the exact question pay equity law asks. The same survey found roughly 45% of respondents consider a non-biased evaluation methodology critical to pay fairness. A well-designed plan answers the question market data can't.

FAQ

How long does it take to design a job evaluation plan? Six to ten weeks for a first plan built internally: two weeks on scoping and factor selection, three to four on degree definitions, one on the weighting session, and two on benchmark testing and revision. Software shortens the drafting substantially, but the weighting conversation still takes the time it takes.

How many compensable factors should a plan have? Eight to twelve sub-factors, grouped under skill, effort, responsibility, and working conditions. Below six, jobs cluster and the plan can't discriminate between levels. Above fourteen, raters fatigue and marginal factors add noise rather than signal.

Should we buy an off-the-shelf plan or build our own? Off-the-shelf plans (Korn Ferry's Hay method, Mercer's IPE, and others) come pre-validated with market data tied to their scores, which is real value. Custom plans reflect your strategy more precisely and cost less to run. Many organizations start with a proprietary structure and customize the weights — see our comparisons of the Hay methodology and Mercer IPE.

Can one plan cover executives and hourly production jobs? Usually yes, if the factors are written broadly enough and the degree scales extend far enough at both ends. The risk is a plan that discriminates poorly at the extremes. Test explicitly with benchmark jobs at the top and bottom before committing.

How do we keep the plan gender-neutral? Include factors that capture work typically performed by women — caring and emotional effort, coordination, dexterity, multi-tasking under interruption — and weight them meaningfully, not token. Then audit: score your most female-dominated and most male-dominated jobs and check whether the point totals match your intuitions about their relative value. Canada, the EU, and several US jurisdictions expect this analysis in writing.

What's the difference between a job evaluation plan and a job architecture? The plan is the measuring instrument. The job architecture is the resulting structure — families, levels, grades, and titles. You build the plan first, apply it, and the architecture is what comes out the other side.

How often should we re-weight? Every three to five years, or after a genuine strategy shift such as a major acquisition or a pivot in the business model. Re-weighting requires re-scoring every job, so it is not a casual change. Degree definitions can be clarified more often; weights should be stable.

Get the plan built

A job evaluation plan is the most leveraged document in compensation. Every grade, every range, every promotion decision, and every pay equity defense traces back to it. It is worth designing carefully — and worth not designing from scratch in a spreadsheet if you don't have to.

PointFactors builds the factor plan, degree scales, and weighting for you, then scores your jobs against it — in days, not quarters. Book a demo and see your own jobs scored, or check pricing to see what it costs.

Justin Hampton is founder and CEO of PointFactors.