The Ranking Method of Job Evaluation: How It Works

The Ranking Method of Job Evaluation: How It Works

The ranking method is the oldest and simplest way to evaluate jobs. You put every job in your organization in order, from most valuable to least valuable, based on an overall judgment about the whole job. No factors, no points, no scoring sheets. A panel looks at two job descriptions and decides which one is worth more.

That simplicity is why ranking still shows up in 40-person startups and small nonprofits that need a pay structure by Friday. It is also why ranking collapses the moment you grow past a couple of dozen jobs, or the moment someone asks you to prove — in writing, to a regulator or a tribunal — why the warehouse lead sits below the executive assistant. This article walks through how ranking actually works, shows you a worked example, and tells you exactly where the method runs out of road.

TL;DR

• Ranking is a non-analytical, whole-job method: you compare entire jobs against each other rather than scoring their component demands.
• Three common variants: simple ranking, alternation ranking, and paired comparison. Paired comparison is the most rigorous and the most expensive — 120 jobs means 7,140 separate comparisons.
• Ranking gives you an order, not a distance. You learn that Job A beats Job B, but never by how much — so you cannot build defensible grade boundaries from it.
• UK and EU guidance both push employers toward analytical, factor-based schemes. A non-analytical ranking will not give you a job evaluation defence under the Equality Act 2010.
• Ranking is a reasonable starting point under about 30 jobs. Past that, move to a point-factor system.

What the ranking method of job evaluation actually is

Job evaluation methods split into two families. Analytical methods break a job into separate demands — skill, effort, responsibility, working conditions — and score each one. Non-analytical methods compare whole jobs against each other and produce a single overall judgment. Ranking, job classification, and paired comparison all sit in the non-analytical family.

The ranking method asks one question about every pair of jobs: which of these two is worth more to the organization? It does not ask why. It does not ask by how much. It produces an ordered list, and you then chop that list into grades wherever the gaps look natural.

Two things to be clear about before you start. First, you are evaluating the job, not the person in it — a strong performer in a modest role does not move the role up the list. Second, "worth more" has to mean something specific and written down, or your panel will silently substitute seniority, headcount, or personal familiarity. Agree a one-sentence yardstick up front, such as "contribution to operating results, allowing for the knowledge and accountability the job requires."

The three ways people run a ranking exercise

Simple ranking is a card sort. Print one card per job, hand the deck to a panel of three to five people who know the work, and let them argue until they agree on an order. It feels intuitive, which is precisely the risk — intuition is where bias lives.

Alternation ranking exists because human judgment is better at the extremes than in the middle. Everyone agrees the CFO outranks the mailroom clerk. Almost nobody agrees on the senior accountant versus the production supervisor. By taking the top and bottom first and squeezing inward, you spend your panel's sharpest attention on the hard middle, where the money actually is.

Paired comparison is the most defensible version of an indefensible method. You build a matrix, compare every job against every other job, and let the win totals produce the order. The math is unforgiving: the number of comparisons is n(n−1)/2. Twelve jobs takes 66 comparisons — an afternoon. Fifty jobs takes 1,225. One hundred and twenty jobs takes 7,140, which at 30 seconds each is roughly 60 hours of panel time before anyone has written a single grade definition.

A worked example

Here is an eight-job ranking at a hypothetical 180-person manufacturer. The panel produced the order in the left column. I have added illustrative point-factor scores in the right column to show what ranking hides.

The ranking is not wrong. The problem is what it does not tell you. Ranks 3 and 4 are 15 points apart; ranks 2 and 3 are 225 points apart. On the ranked list, both look like a single step. If you draw your grade boundaries off rank alone, you will almost certainly put the Production Supervisor and the Controller in adjacent grades and pay them as though they are neighbors, when the work is nowhere close.

That gap is the whole argument. Ranking gives you ordinal information — first, second, third. Pay structures need interval information — how far apart, and therefore how much more. You cannot build defensible pay ranges or set salary range midpoints from an ordered list, because the list contains no distances.

Where ranking breaks down

It does not scale. Panels can hold roughly seven items in working comparison at once. Past 30 jobs, ranking stops being a judgment and starts being a series of half-remembered judgments.

It cannot absorb new jobs. Create a new role and you have to re-open the whole list, because the new job's position is defined only relative to its neighbors. A point-factor system scores the new job on its own and drops it into the existing structure in an hour.

It hides bias instead of surfacing it. This is the serious one. Whole-job comparison lets a panel undervalue demands that show up disproportionately in female-dominated roles — emotional demands, communication skill, responsibility for people rather than budget — without anyone ever having to say so out loud. The EHRC makes the point directly in its job evaluation guidance: schemes should "identify all the demands of a job, and not overvalue or ignore factors that are associated with jobs typically done by one gender or the other." A factor-based scheme forces that conversation into the open, because someone has to write down what "communication" is worth. Ranking never asks the question, so the assumption travels undetected into the pay structure.

It leaves you without a defence. More on that next.

Running a ranking exercise right now and worried the order won't hold up? The PointFactors scorecard lets you re-score the same jobs against weighted compensable factors and see where your ranked order and your scored order disagree. The disagreements are where your pay risk sits.

The legal problem with non-analytical schemes

Under the UK Equality Act 2010, a job evaluation study can effectively decide an equal value claim. Section 131(6) says a tribunal must find that two jobs are not of equal value where a job evaluation study has given them different values — unless there are reasonable grounds to suspect the study was based on a system that discriminates by sex, or is otherwise unreliable.

That is a powerful shield, and a ranking exercise does not qualify for it. The EHRC's guidance, drawn from case law, sets four criteria a scheme must meet to be relied on in defence of an equal value claim. The first is that it must be analytical — it must "assess and score each job in terms of the demands made on the employee under 'job factors'." The other three are thorough and impartial, gender neutral, and reliable. Whole-job ranking fails the first test by definition, which means it never reaches the other three.

The European picture points the same way. Directive (EU) 2023/970 on pay transparency requires employers to use objective, gender-neutral criteria — skills, effort, responsibility and working conditions, plus any other relevant factors — when determining whether work is of equal value. Those are the four classic compensable factor families. A method that never scores them cannot demonstrate it applied them. If you operate in the EU, read our guide to work of equal value alongside this one.

When ranking is still the right call

Ranking earns its place in three situations:

• You have fewer than about 30 jobs and no immediate regulatory exposure. A ranked order plus a sanity check against market data will get you a workable grade structure.
• You need a fast first pass. Ranking your top 15 benchmark jobs before a full evaluation is a good way to test whether your panel shares a mental model of value. If they violently disagree at this stage, fix that before you invest in scoring.
• You are validating a point-factor result. Rank the jobs independently, then compare against the scores. Where rank and score disagree sharply, one of them is wrong — usually the factor weighting. This is a cheap and genuinely useful quality check, and it is the best modern use of the method.

What ranking should not be is your permanent system of record. If you are also weighing classification or factor comparison, our breakdown of the four methods of job evaluation puts all of them side by side.

Moving from ranking to a point-factor system

1. Keep the ranked order. It is your control group. You will use it to sanity-check the scores.
1. Choose four to seven compensable factors. Start with the statutory families — skill, effort, responsibility, working conditions — and add sub-factors that reflect how your organization actually creates value. See compensable factors for definitions and examples.
1. Weight the factors before you score anything. Weighting after you see results is how you rationalize the answer you already wanted. Our guide to weighting compensable factors covers the mechanics.
1. Score 10–15 benchmark jobs first. Get panel agreement on those, then work outward. Benchmarks anchor everything that follows.
1. Compare the two orders. Every job that moves more than one position between your ranked list and your scored list deserves a written explanation. Those explanations become your audit trail.

FAQ

What is the ranking method of job evaluation? It is a non-analytical job evaluation method in which whole jobs are compared against each other and placed in order from highest to lowest value, without breaking the jobs into separately scored factors.

What is the difference between ranking and the point-factor method? Ranking produces an order; the point-factor method produces a score. Ranking tells you Job A outranks Job B. A point-factor system tells you Job A scores 520 and Job B scores 395, which lets you set grade boundaries, ranges, and midpoints with actual arithmetic behind them.

How many comparisons does paired comparison require? n(n−1)/2, where n is the number of jobs. Ten jobs takes 45 comparisons, 50 jobs takes 1,225, and 120 jobs takes 7,140. Effort grows roughly with the square of the job count, which is why the method rarely survives contact with a mid-sized organization.

Is the ranking method legally defensible? Not as an equal pay defence. UK guidance derived from case law requires a job evaluation study to be analytical — scoring jobs under defined job factors — before it can be relied on under section 131 of the Equality Act 2010. Whole-job ranking is non-analytical and does not meet that standard.

Is ranking the same as job classification? No, though both are non-analytical. Ranking compares jobs against each other. Classification compares each job against pre-written grade descriptions and slots it into the best match. Classification scales better because you do not have to re-open the whole structure to add a job.

Can ranking be gender neutral? It is much harder to demonstrate. Because ranking never records why one job outranks another, there is no evidence that demands common in female-dominated roles were identified and valued. Analytical schemes create that evidence as a by-product of scoring.

How many jobs can ranking realistically handle? Around 25 to 30 before consistency degrades badly. Above that, use an analytical method or a hybrid where ranking sets the benchmark anchors and scoring does the rest.

Rank once. Score from then on.

If you inherited a ranked list, you already have something useful — a shared view of order. What you do not have is a structure you can explain to an auditor, a works council, or an employee who wants to know why their grade is what it is. PointFactors turns whole-job judgments into weighted, documented, point-factor scores in days rather than months, and every score comes with the rationale attached.

Book a demo and bring your ranked list. We will score your benchmark jobs against weighted compensable factors and show you where the ranking and the evidence part company — usually in three or four places you will not have expected. You can also see how the job evaluation platform handles ongoing re-evaluation, or check pricing if you are scoping a project.

Justin Hampton is founder and CEO of PointFactors.