Both art and science are involved in creating a salary range structure. I cannot fit it all into a blog. Perhaps I will write a book someday. I’ve created several hundred pay range structures to guide pay decisions. No two are the same.
For those of you who think it is a simple thing, here are some of the key questions and considerations. The answers to these questions will lead you to smart decisions on how to construct the pay ranges.
- Number of structures—should different business areas have separate structures, or use a single structure? Consider separate structures for sales, senior management, white collar versus blue collar, hourly versus salaried, union versus non-union, or strategic versus non-strategic (enabling or transactional) functions. Are there market pay differences by function, at the same grade level?
- Number of grades—how many reporting levels are there? How flat does the organisation want to be? How frequent should promotion opportunities be? How much resourcing does the HR dept have to conduct job evaluations (more grade levels require more job evaluations)?
- Make or buy?—does the organisation prefer to hire fresh graduates and develop them over time, saving on salary, but allowing greater control over the development of work attitudes, alignment to company culture and values, and industry knowledge? Or does the company prefer to hire people with more experience, people who can “hit the ground running” and already know the industry, but who come with expectations, current pay packages or work habits that create challenges? The answer to these questions can influence decisions about “anchoring” the range minima and midpoints of certain grades to the corresponding market pay levels that will make the chosen approach effective. A buy approach may require wider or less rigid ranges.
- Internal versus external focus—should grading be strictly determined by job evaluation, looking at job content, or will market value influence grading to ensure ranges are competitive?
- Job evaluation methodology—Does your job evaluation method result in a number of job levels that aligns naturally to the levels of your structure? If your structure gives you level scores ranging from 40 to 60, will you build a structure with 21 levels to match, or do you need only 12 or 13 levels? The market may recognize 3 levels of IT help desk, and your job evaluation system can distinguish them well, yet if your structure has 8 levels, then you only need two of them for IT help desk. How will you therefore withhold a promotional opportunity offered by other companies to conform to a flat structure?
- Pay philosophy on market positioning—will there be a lot of allowances, which take pressure off of salary? Will incentive opportunities be at, below or above market norms? Does the company wish to align to the market, or lead the market with higher pay levels for certain business areas or job levels or locations? Do you want to lead on salary and lag on incentives? What is the ideal mix of compensation? How competitive are benefits or other rewards? To what extent can intangible rewards form part of the package, such as contributing to nation-building or opportunity for international projects?
- Philosophy on development—are frequent promotions desired, and if so, should promotions be based on the job only? Can people grow “in the job” or must there be growth “of the job” for a promotion?
- What is the role of job titles? Will titles be hard-wired to grade levels, or have a life of their own? If there are fewer, broader grades, will it be possible to use such title prefixes as Junior, Senior, Assistant, Deputy, 2nd Assistant, etc.? Must all “managers” be in the same grade? Should administrative support roles like PA be consistent, link to the level of manager supported, or be graded strictly based on job evaluation factors?
- Selection of pay statistics—survey data can be acquired easily. It can be bought with participation or a promise to participate. The challenge is knowing what numbers you want to use from the survey:
- Regressed data can be used directly if you wish to have a similar number of grades compared to the levels provided in the regressed data (e.g. Mercer position class, or Radford career level, etc.) If the company’s desired number of levels matches the levels of your data source and a one-size-fits-all structure is desired for overall internal pay equity, an organisation can align their grades directly to the survey levels and simply use regressed data as midpoints!
- Job specific data is the preferred approach for larger organisations with a diverse range of job families. Job market values within the same grade may vary significantly, so it is important you can find job-specific data. If you are competing with an industry dominated by 1 or 2 employers whose pay packages you do not intend to match, then use company-weighted data if available.
The answers to these questions and the combination of answers will point to a specific solution. From there you construct the ranges and model them for fit against the population Avoid creating a monster or a class system. Keep it simple and fair, yet effective at guiding individual pay decisions in a way that effectively attracts and retains talent, allows sufficient differentiation based on performance, give you access to the range of market talent you want.
Plus.. your ranges should look nice when shown as a graph. That was a joke.
There is no way to fully automate creation of a fit-for-purpose salary range structure, though I supposed it is possible to come close. I know of no company that does not use Excel, building ranges like a craft. There’s no app or bot that can do this to my knowledge.
3 thoughts on “Pay Structures—Simple?”
What are your views about using salary data regressed across a wide range of career streams and levels?
Based on experience, doing so often results in wide variances between actual data and regressed data and c&b professionals might run the risk of using a higher/lower midpoint than is necessary.. I personally avoid regressing data agressively at the first instance and will always compare actual vs regressed data to assess before making a decision…
Erik, your concerns about using regressed data are valid. Some like the aesthetic “appearance” of regressed midpoints, but at the cost of some market accuracy. Market accurate midpoints stay true to the data, even if your midpoint differentials are inconsistent. Smooth lines are overrated, while market accuracy should have the greater weight in most cases.
One approach that balances the two is to anchor certain grade midpoints to the market value, and only regress the intermediate grades. Message me on LinkedIn or email me for an explanation of this method. Email is shown in the right margin here in the website.
Sure. I will get in touch with you shortly!