Best midpoint progression for salary grades: what the gap between grades should actually be

For most US companies under 200 employees, set midpoint progression at 12 to 15 percent between individual contributor grades and 18 to 20 percent at the point where professional or management work begins. The number is not a free choice: progression equals your range spread multiplied by one minus the overlap you want, so a 30 percent spread with a 50 percent overlap has already decided on 15 percent for you. Anything below about 12 percent produces promotions that pay nothing.

Midpoint progression is the gap between the midpoint of one grade and the midpoint of the grade above it. It gets treated as a detail, somewhere behind the more visible question of how wide each grade should be, and then it quietly decides three things that matter far more: how many grades you need, how big a promotion feels in the bank account, and how much of each grade sits inside the one below it.

The number you think you are choosing is already determined

Comp guidance usually lists range spread, midpoint progression and grade overlap as three separate decisions with three separate recommended ranges. They are one decision with three names. If every grade carries the same spread, then

progression = spread x (1 - overlap)

That identity holds exactly, and you can check it against a published structure. The federal General Schedule runs a 30.00 percent spread at every grade from GS-3 to GS-15. Between GS-14 and GS-15 the measured progression is 17.62 percent, and one minus 0.1762 over 0.30 is 0.413, against a measured overlap of 41.3 percent. The full grade by grade table, with midpoints computed from the published step 1 and step 10 rates, sits on the salary midpoint calculator page along with a builder that generates the whole ladder once you pick two of the three numbers.

The practical consequence is that the standard advice contradicts itself. Run a 40 percent spread, which plenty of guidance recommends for professional roles, hold overlap at the commonly cited 50 percent, and you have committed to a 20 percent progression. That is well outside the 10 to 15 percent the same guidance recommends. Pick the two you actually care about and read the third off the formula.

Four progressions, and what each one costs you

The comparison below assumes a 30 percent range spread throughout and a company whose lowest grade midpoint is $55,000 and whose highest is $165,000, which is a fairly ordinary span for a business under 200 people with a leadership team but no executives on public-company pay.

Grades needed is the number of steps required to compound $55,000 up to $165,000 at each progression, plus one. Overlap is derived from a constant 30 percent spread. Promotion raise is explained in the next section.

Low progression turns a promotion into a pay cut

This is the part that gets discovered in the middle of a promotion conversation rather than during structure design, and it is pure arithmetic. A strong, tenured employee is normally high in their grade, say a 110 percent compa ratio. When they get promoted, the sensible placement in the new grade is low, say 95 percent, because they are new to that level of work. Their new salary is 95 percent of the new midpoint, and the new midpoint is only one progression step above the old one.

At 10 percent progression, promoting your best person out of a grade they have outgrown cuts their pay by 5 percent on paper. Nobody actually does that, of course. What happens instead is that the promotion is granted with a manual exception on top, the new salary lands at 105 or 110 percent of the new midpoint, and within two years that person is at the ceiling of the new grade as well. Do this a dozen times and the structure has stopped describing anything.

The fix is not a bigger promotion budget. It is either more progression or more discipline about where people sit inside a grade, and the second is much harder to hold. If your promotions keep needing exceptions, run the compa ratios before and after through the compa ratio calculator and you will usually find the gap between grades is the thing that is too small.

Put the big gap where the job changes, not everywhere

Even progression is tidy and it is not what the biggest published US structure does. Through the General Schedule support grades the progression drifts down from 12.8 percent to 9.9 percent, then jumps to 19.9 percent into GS-12, then settles around 18 percent. The large step sits exactly where the nature of the work changes, at the boundary into full professional responsibility, and it is roughly double the ordinary gap either side of it.

Copy the pattern rather than the numbers. In a company under 200 people there are usually two such boundaries: the move from doing the work to owning an area, and the move from owning an area to managing people who own areas. Set 12 to 15 percent within each block and 18 to 20 percent across those two seams. The result is a ladder where lateral growth inside a block is cheap and a genuine level change is visibly paid for, which is the behaviour most small companies say they want and then accidentally price out.

How many grades you should end up with

Count backwards from your span. Take the midpoint of your lowest-paid role and the midpoint of your highest, then see how many progression steps fit between them. At 15 percent, a $55,000 to $165,000 span needs nine grades. At 10 percent it needs thirteen, and thirteen grades in a 60-person company means several grades containing one person each, which is not a structure so much as a list.

Fewer grades also survive scrutiny better. Every grade boundary is a place where an employee can reasonably ask why their job is on one side of it, and a boundary that exists only because the ladder needed filling is very hard to defend. This is the same argument that runs through salary structure design: the number of levels should come from how many genuinely different kinds of work you have, and the progression should then be chosen to span them, not the other way round.

One practical note for the moment you present this. A grade ladder is a table, and tables read badly in a leadership meeting where the real question is what it costs. Build the table first, price the current population against it, and then turn the whole thing into slides with the cost line on the same page as the structure. The conversation goes very differently when the two numbers are visible together.

What to do this week

Write down three numbers: your lowest grade midpoint, your highest, and the range spread you already use. The spread is usually the one you have, because it came from whoever built the first bands. Then pick the overlap you want, read the progression off the formula, and see how many grades that produces. If the answer is more than about eleven, your progression is too low for your size.

Then test it against two real people: your strongest individual contributor and the person you most expect to promote next year. Run their current compa ratio, place them in the grade above at the compa ratio you would honestly want them at, and look at the percentage change. If it is under about 5 percent, the structure will be overridden by hand the first time it is used, and you may as well fix it now. The width question sits alongside this one and is worked through in how wide a salary range should be.

Sources: Salary Table 2026-GS, US Office of Personnel Management, effective January 2026. Midpoints, range spreads, midpoint progressions and grade overlaps computed by Wagelist from the published step 1 and step 10 rates. Promotion figures are arithmetic from the compa ratio placements stated in each row.