How to Implement a Job Evaluation System: A 90-Day Rollout Plan

How to Implement a Job Evaluation System: A 90-Day Rollout Plan

Most job evaluation projects do not fail because the method was wrong. They fail because nobody scoped the work, the committee met four times in five months, and by the time the factor plan was finalized the CHRO had moved on to a different priority. The methodology is the easy part — point-factor scoring is well-understood and has been for seventy years. The hard part is sequencing: what you do in week two versus week eight, who signs off on what, and how you keep 400 jobs from becoming a two-year archaeology project. This is the rollout plan I would hand a comp team of one or two people who have been told to "get job evaluation in place by the end of the quarter." It assumes you are building a point-factor system, but the sequence holds for any analytical method.

TL;DR

• Weeks 1–3: scope and mandate. Decide what the system must produce, get an executive sponsor, and freeze the job inventory. Do not start scoring.
• Weeks 4–6: build the factor plan. Choose 4–8 compensable factors, write degree definitions, set weights, and pressure-test them on three jobs you already have strong opinions about.
• Weeks 7–9: pilot on benchmark jobs. Score 25–40 benchmarks that span every family and level. This is where you find out whether your scales actually discriminate.
• Weeks 10–12: calibrate, map to grades, and communicate. Convert points to grades, reconcile against market data, and tell managers before they hear it secondhand.
• The whole job library does not have to be scored in 90 days. Benchmarks plus a slotting rule gets you a defensible structure; the long tail follows.

Before week one: decide what "done" means

Write a one-paragraph definition of success and get your sponsor to agree to it in writing. It should name the deliverable, not the activity. "A factor plan, scored benchmark jobs, and a grade structure that every job in the company can be mapped to by Q4" is a deliverable. "Implement job evaluation" is a wish.

This matters because job evaluation gets pulled in three directions at once, and the three demand different amounts of work:

If your sponsor wants the third and has budgeted for the first, you need that conversation in week one, not week eleven.

Two other decisions belong here. First, confirm you have a written compensation philosophy — WorldatWork's job evaluation survey found 65% of organizations have one, and the ones that do not tend to relitigate the same market-positioning argument at every committee meeting. Second, name the sponsor. Not a supporter — a sponsor who will tell a VP "no" when that VP wants their team upgraded.

Weeks 1–3: freeze the inventory

You cannot evaluate jobs you cannot list. Pull every active job title from your HRIS and expect the count to be wrong. A 600-person company routinely carries 400+ distinct titles, most of which are variants of about 120 real jobs.

Do three things:

• De-duplicate to real jobs. "Sr. Analyst, FP&A" and "Senior Financial Analyst" with the same duties are one job. Merge aggressively; you can always split later.
• Assess documentation quality. WorldatWork found 65% of organizations have up-to-date job descriptions for most or all jobs — meaning roughly a third do not. Rate each job green/amber/red. Red jobs need a job evaluation questionnaire before they can be scored, and that takes calendar time you must budget now.
• Freeze the list. Publish the inventory and declare that new or changed jobs after this date go into a second wave. Without a freeze, the scope grows every week and you never finish.

Also stand up your governance in this window. A job evaluation committee of five to seven people, meeting on a fixed cadence with a named decision rule, beats an open-ended working group every time. Book all twelve meetings now.

Weeks 4–6: build the factor plan

This is the intellectual core of the system, and it takes about three weeks of real work — not three months.

Choose your compensable factors first. Four to eight is the workable range. Fewer than four and the scores will not discriminate between jobs; more than eight and you are double-counting the same underlying construct under different names. Your factor set should trace back to the four statutory families the Equal Pay Act uses — skill, effort, responsibility, and working conditions — because that is the framework a regulator or a plaintiff's expert will apply. The EEOC's guidance is explicit that comparisons turn on actual job duties, not titles or classifications, which is exactly what a factor plan forces you to document.

Then write factor degree definitions. Each factor needs four to six degrees with behavioral anchors — observable differences, not adjective ladders. "Moderate complexity" and "significant complexity" are not degrees; they are synonyms with different intensities. "Solves problems using established methods" versus "adapts methods where precedent does not exist" is a degree difference two raters can agree on.

Finally, set your weights. Weighting compensable factors is where your compensation philosophy becomes arithmetic: if knowledge carries 30% of the points and working conditions carries 5%, you have made a statement about what your organization pays for. Test the weights on three jobs whose relative worth is not in dispute — say, a receptionist, a mid-level engineer, and a plant manager. If the scores do not order them the way every reasonable person in the building would, your weights are wrong. Fix them now, before the pilot, when changing them costs nothing.

A working factor plan is the point at which the project becomes real. If you want to see what a complete, tested plan looks like before you build your own, take a walkthrough of PointFactors — the factor structure and degree scales are visible in the product.

Weeks 7–9: pilot on benchmark jobs

Score 25 to 40 benchmark jobs chosen to span every job family and every level from entry to executive. Benchmarks do two jobs at once: they anchor the internal structure and they connect to survey data, so pick jobs that actually match published survey benchmarks.

Run the pilot with at least two independent raters per job, scoring blind, then compare. Disagreement is information. If two trained raters land three degrees apart on the same factor, the problem is the degree definition, not the raters. Log every disagreement, fix the scale language, and re-score. Expect to revise 20% to 30% of your degree definitions during the pilot. That is the pilot working, not failing.

Watch for two specific failure signs:

• Score compression. If 80% of your jobs land in a 150-point band on a 1,000-point scale, your factors are not discriminating. Usually the weights are too flat or the top degrees are unreachable.
• Title contamination. If raters score "Director of Marketing" high because of the word Director, you are evaluating titles, not jobs. Strip titles from the scoring packet and score the documented content.

Weeks 10–12: grades, market reconciliation, and communication

Convert the point totals into a structure. Turning job evaluation points into pay grades is a mechanical step once you have clean scores: sort by points, look for natural gaps, and set grade boundaries there rather than at arbitrary round numbers.

Then reconcile against the market. Job evaluation sets internal relativities; survey data sets the external pay level. When a job's internal rank and its market rate disagree sharply — a software engineer scoring like an accountant but paying 40% more — do not bend the score. Document the gap and handle it with a market premium or a separate structure. Bending scores to chase market rates destroys the internal logic that made the system defensible in the first place.

The remaining jobs do not all need to be scored before you go live. Use job slotting against the scored benchmarks for jobs that are clearly variants, and reserve full scoring for anything genuinely novel or contested.

Communication is the step teams skip, and it is the one that determines whether the system survives. Tell managers what changed, what did not, and what happens to an employee whose job now sits in a lower grade than their pay. Publish an appeals process at the same time you publish the results — an appeals route you announce afterward looks like damage control, while one you announce upfront looks like confidence.

The three mistakes that stall implementations

Scoring before scoping. Teams that start evaluating jobs in week one always re-do that work, because the factor plan changes underneath them. Every job scored before the plan is stable is wasted effort.

Letting the committee design by consensus. Committees are excellent at applying a factor plan and terrible at writing one. Draft the plan centrally, then bring it to the committee for challenge and ratification.

Treating go-live as the end. WorldatWork's data shows evaluation is overwhelmingly event-driven — 86% of organizations trigger an evaluation when a job is newly created and 85% when job content changes significantly, while only about 10% work to a fixed schedule. Event-driven is the right instinct, but it only works if somebody owns the trigger. Assign that ownership before the project team disbands.

FAQ

How long does a job evaluation implementation really take? Ninety days gets you a factor plan, scored benchmarks, and a grade structure for a company under roughly 1,500 employees with reasonable job documentation. Poor documentation adds four to eight weeks. Scoring the full job library, building level descriptors, and running a pay equity analysis on top pushes it to six to twelve months.

Do we need a consultant? For most mid-market organizations, no. WorldatWork's survey shows compensation staff conduct 86% to 88% of evaluations for non-executive jobs; consultants are used most at the executive level, where they handle 27% of evaluations. Buy expertise where the political stakes are highest, not for the whole build.

How many people do we need on the project? One owner at 50% time for the quarter, plus a committee of five to seven contributing four to six hours a month. Fewer than that and the meetings do not happen; more and the decisions do not.

Should we score every job before launching? No. Score benchmarks that cover every family and level, then slot the rest against them. Waiting for 100% coverage is the single most common reason these projects never launch.

What do we tell employees whose grade goes down? Nothing about their pay changes on day one. Grade movement usually means the range moves, not the paycheck. Explain red-circling rules before anyone asks, and hold current pay while the range catches up.

How do we keep the system from drifting? Re-evaluate on defined triggers — new job, material change in duties, reorganization — and audit the whole plan every two to three years. Log every scoring decision with its rationale so the audit has something to read.

Can AI shorten this? It compresses the scoring step substantially. Automated scoring against your factor plan removes the committee bottleneck and applies identical logic to every job, which also removes rater drift. It does not replace the scoping, the factor plan, or the communication — those are judgment calls that belong to you.

Where to start

Pick your sponsor and freeze your job inventory this week. Everything downstream depends on knowing what you are evaluating and who will defend the answer. The factor plan can be drafted in three weeks by one person who understands the business; the pilot will tell you within a fortnight whether the plan holds. What kills these projects is not difficulty — it is drift.

If you would rather not build the factor plan from scratch, start from one that already works. Book a walkthrough and we will score a handful of your real jobs against a tested point-factor plan so you can see the structure before you commit to a quarter of work — or look at pricing if you want to run it yourself.

Related reading: The Definitive Guide to the Point-Factor Method and The Job Evaluation Process: 7 Steps That Work.

Sources: WorldatWork, Job Evaluation and Market Pricing Practices Survey (February 2020); U.S. EEOC, Facts About Equal Pay and Compensation Discrimination; U.S. EEOC, Compliance Manual Section 10: Compensation Discrimination.

Justin Hampton is founder and CEO of PointFactors.