Benchmarking

Most benchmarking failures happen before any data is bought, in the choice of comparison group.

• Use where you actually compete for the specific skill. A payroll specialist in a manufacturing company competes with services firms and shared service centres, not only with other manufacturers.
• Match on location properly. Pay for the same role differs materially across Indian cities, and a national median describes nowhere.
• Match on size and stage. A forty-person company benchmarking against listed multinationals is comparing itself to employers with different structures, benefits and expectations.
• Look at where leavers go and where offers are lost. Those two lists are your real comparison group and they are free.
• Separate the roles where you must be competitive from the roles where you need not be. Paying at the same percentile across every function is expensive and unnecessary.

The disciplined approach is to match on job content rather than title, to look at the range and not just the midpoint, and to compare fixed pay separately from variable pay and from benefits, because those three are structured differently by every employer and blending them produces a number that describes nobody.

• Decide a position and state it: at market, above for critical roles, below with a stated compensating advantage. An undeclared position is decided by whoever negotiates hardest.
• Check internal equity before correcting anyone. Moving one person to market creates a comparison with everyone doing the same or similar work, and that comparison has to be defensible.
• Correct systematically rather than individually where a whole group is off market, since individual corrections in response to resignations teach employees how to get a raise.
• Consider what else the benchmark implies. If your pay is at market and you still lose people, the problem is elsewhere and more pay will not fix it.
• Benchmark the structure, not only the level. Fixed to variable ratios, allowance design and benefits differ, and a competitor paying the same total in a different shape may be more attractive for reasons that cost nothing to copy.
• Re-benchmark on a cycle rather than in response to a resignation, and align it with the increment cycle so the data is used while it is current.

Practice benchmarking, comparing process and outcomes rather than pay, is often more useful and is undersold.

• Attrition, time to hire, offer acceptance and joining ratio compare meaningfully across employers of similar size.
• Ratios such as HR staff per employee, or payroll processing days per cycle, indicate where you are carrying unnecessary work.
• Leave utilisation, absence and overtime hours compare within a sector and reveal scheduling problems that pay data will not.
• Internal benchmarking across your own locations is the cheapest source of all, since the variation between two of your own sites is data you already own.

The caution is narrow and worth stating. Comparing pay through a properly run survey with aggregated, historical data is ordinary practice. Agreeing with competitors on what to pay, or exchanging current or forward-looking pay intentions or no-poaching understandings directly between employers, is a different activity and can raise competition law issues. Keep benchmarking to aggregated data from a third party rather than to direct conversations about what each of you intends to pay next year, and take advice before entering any arrangement with another employer about hiring or wages.