Job Title Levels: How Many Should You Have?

Job Title Levels: How Many Should You Have?

"How many levels should we have?" is the question that stalls more job architecture projects than any other. Someone proposes eight. Engineering says they need twelve because that is what their competitors publish. Finance wants five so the merit cycle is simpler. Nobody has a defensible way to settle it, so the project slips a quarter and the company keeps promoting people into titles that mean nothing.

The answer is not a number you pick. It is a number you calculate, from the actual spread of work in your organization and the pay differential you need between adjacent steps. This guide gives you the math, size-based benchmarks to sanity-check your result, and the sequence that keeps your level count from drifting back upward the moment the project ends.

TL;DR

• Level count is a function of two inputs: the point spread between your smallest and largest job, and the pay differential you want between adjacent levels. Everything else is negotiation.
• Rough benchmarks by headcount: under 100 employees, 5–8 levels; 100–2,000, 8–12; 2,000+, 12–18 with separate tracks by career stream.
• Aim for a 12–18% midpoint progression between adjacent levels. Below 10% and the promotion is symbolic; above 25% and people get stuck.
• Levels are not titles. One level can carry several titles across job families — it should carry exactly one grade.
• Derive levels from point-factor evaluation scores, not from headcount politics. A level boundary you cannot explain in factor terms will not survive its first appeal.

Levels, grades, and titles are three different things

Teams argue about level count because they are arguing about three things at once. Separate them first.

A job level is a band of evaluated work — a defined range of scope, complexity, and accountability. It is a measurement concept.

A pay grade is the salary range attached to a level. In most modern structures the mapping is one-to-one. In a broadbanded structure, several levels may share a wide band.

A job title is the label. Level 6 might contain Senior Financial Analyst, Senior Software Engineer, and Manager of Facilities — same evaluated weight, three titles, three job families.

That last distinction resolves most of the "engineering needs more levels" argument. Engineering rarely needs more levels; it needs more titles that map onto the levels you already have. Those are very different projects — one redesigns your pay structure, the other updates a naming standard.

The math: two inputs decide your level count

Here is the calculation, and it takes ten minutes once you have evaluation scores.

Input one: your point spread. Evaluate a representative sample — your smallest individual contributor role, your largest executive role, and 15 to 20 in between. Say your entry coordinator scores 180 points and your CFO scores 940. That is a ratio of roughly 5.2x.

Input two: your target midpoint progression. The percentage jump in salary midpoint between adjacent levels. Market practice sits between 12% and 18% for professional structures, wider (20%+) at executive levels where the market itself is more dispersed.

Now solve for the number of steps:

The arithmetic is `log(ratio) / log(1 + progression) + 1`. A 5.2x spread at 15% progression gives about 12 levels. That is your starting number — the one you defend in the steering committee, because it came from evaluated work rather than from preference.

Two practical adjustments. If your calculation lands above 15 levels, your progression is too narrow, not your organization too complex; widen it. If it lands below 6, you probably under-sampled the top or bottom of the organization.

Sanity-check against size

Benchmarks are not a substitute for the math, but they tell you when your answer is strange. Published survey practice clusters like this:

• Under 100 employees: 5 to 8 levels. Most jobs are generalist and scope differences are obvious without fine gradation.
• 100 to 2,000 employees: 8 to 12 levels. This is where the first real architecture usually gets built.
• 2,000+ employees: 12 to 18 levels, often with separate tracks for individual contributor, management, and executive streams that share a common grade spine.

For a public reference point, the U.S. federal General Schedule runs 15 grades from GS-1 to GS-15, with 10 steps inside each grade — and it covers roughly 1.5 million positions across every occupation the government employs. If 15 levels can carry that much variety, your 4,000-person company almost certainly does not need 22.

Building this from scratch? PointFactors scores your jobs against weighted compensable factors and shows you the point distribution before you commit to a level count — so the number comes out of the data instead of out of a workshop. See how the scoring works.

Job level examples: what a 10-level structure looks like

Concrete beats abstract. Here is a defensible 10-level spine for a 1,200-person company, with the point bands and typical titles that sit in each.

Notice three design choices worth copying.

The tracks converge on the same grades. A Principal Analyst at level 5 and a Manager at level 5 sit in the same salary range. That is the whole point of a dual career ladder — it only works if the levels are genuinely equivalent in evaluated scope, not just in rhetoric.

Point bands widen as you climb. Level 1 spans 60 points; level 8 spans 100. Higher-scoring jobs are more differentiated and less precisely measurable, so wider bands prevent meaningless reclassifications.

Management starts at level 3, not level 1. The first supervisory role does not automatically outrank a senior individual contributor. Tying the entry point to a score rather than to the org chart is what stops title inflation, as covered in our guide to job titling conventions.

Too many levels vs. too few: the failure modes

Both errors are expensive, in opposite ways.

Too many levels produces promotions with no pay attached. If adjacent midpoints differ by 6%, a promotion carrying a 4% increase leaves the person lower in the new range than they were in the old one — and they will notice. You also get near-identical jobs landing in different grades because a 15-point scoring difference crossed an arbitrary line, which shows up in a pay equity review as unexplained variance. More levels also mean more range overlap to manage and more boundary appeals to adjudicate.

Too few levels produces stalled careers and offer-driven pay. When a band spans 40% of salary, a strong performer can spend six years without a level change, so retention pressure moves entirely into off-cycle adjustments. Managers respond by inventing titles the structure does not recognize.

The diagnostic is simple. Pull last year's promotions. If the median promotional increase was under 8%, you have too many levels. If your median time-in-level exceeds four years outside the executive ranks, you have too few.

How to set your level count in five steps

1. Evaluate a benchmark sample. Twenty to thirty jobs spanning your full range of work, scored against weighted compensable factors. You cannot set levels without knowing the distribution you are cutting.
1. Plot the distribution. Look for natural clustering. Real organizations produce clumps around common scope profiles, and the gaps between clumps are the cheapest places to draw boundaries.
1. Run the math. Apply your target progression to your point spread. Compare the result to the size benchmarks above.
1. Cut the bands. Convert point ranges to grades using the approach in turning evaluation points into pay grades, then attach market-based ranges.
1. Test against edge cases. Take the ten jobs your managers argue about most. If the structure places each one somewhere you can explain in factor terms, it will hold. If two of them land wrong, fix the boundary now — not after go-live.

For the wider framework these levels plug into, see our guides to job architecture and job leveling. WorldatWork's guidance on crafting job architecture and SHRM's overview of salary structure design are both worth reading before you finalize the count.

FAQ

How many job levels should a company have? Most organizations land between 8 and 15. Calculate rather than guess: divide the log of your point spread by the log of one plus your target midpoint progression. A 5x spread at 15% progression gives roughly 12 levels. Then sanity-check against size — under 100 employees, 5 to 8 is normal; over 2,000, 12 to 18.

What is the difference between a job level and a pay grade? A level describes evaluated work — scope, complexity, accountability. A grade is the salary range attached to it. In most structures the relationship is one-to-one, so people use the words interchangeably. In a broadbanded structure, several levels can share one wide band.

Can different departments have different numbers of levels? They can have different numbers of titles, but they should share one grade spine. If engineering uses six IC titles and finance uses three, that is fine — as long as both map onto the same evaluated levels so a level 5 means the same amount of work everywhere.

How much pay difference should there be between job levels? Target 12% to 18% between adjacent midpoints for professional roles, widening to 20% or more at executive levels. Below 10%, promotions stop feeling like promotions. Above 25%, people stall because the next step is too far to reach.

Should job levels be visible to employees? Publishing the level framework and the criteria for each level is increasingly standard, and it is close to mandatory anywhere pay ranges must appear in job postings. Publishing every individual's level is a separate decision — most companies share the framework and the level attached to each posted job, not a directory.

How often should we revisit our level count? Review the framework every two to three years, or after any event that changes your scope distribution — an acquisition, a new business line, a significant layer removal. Re-evaluate individual jobs when duties materially change, not on a calendar.

What are job tiers? "Tier" is usually an informal grouping of levels — entry, professional, senior professional, leadership, executive. Tiers are useful for communication and for setting different progression rates. They are not a substitute for levels, because they are too coarse to attach a salary range to.

Decide it with data, not with a workshop

Level count causes so much argument because most teams try to settle it by discussion, and discussion has no stopping rule. Point scores do: once you know your work spans 180 to 940 points and you have committed to a 15% progression, the number of levels stops being a matter of opinion.

PointFactors evaluates your jobs against weighted compensable factors, shows you the resulting point distribution, and lets you test level boundaries before you commit — including what happens to every job when you move a cut point by 20 points. You get a level count you can defend to a skeptical executive, an appealing employee, or an auditor, because the reasoning is written down.

Book a demo and see your own jobs plotted on a point scale before you pick a number.

Justin Hampton is founder and CEO of PointFactors.