Unpopular view: most frameworks carry two levels too many in the middle

Agree with the diagnosis, less sure about the cure. The demand for movement every two years is real, particularly early in a career. We kept the levels but stopped pretending each was a different job: levels 2 to 4 are explicitly one job at three stages of proficiency, with lighter criteria and no panel. Panels only meet from level 5 upwards.

How did you handle the people in the two levels you removed? That is the part that stops most of us. Merging upwards costs money and merging downwards costs trust.

We merged upwards in both cases, about 610 people, and held their pay where it was within the new, wider range. Cost in the first year was close to nil. The real cost comes later, since they now have more headroom.

On the pay gap point, that is fair, and we added quarterly monitoring of range position by gender because of it.

Your two-managers test is a good one and I have used a version of it. We gave 20 managers anonymised role summaries and asked them to place each at a level. Agreement was above 80% everywhere except between levels 6 and 7, where it was 48%. That one result did more to justify a merger than any design argument.