Salary Band Construction: Market Data, Internal Equity, and Band Design

Salary band construction is the under-discussed structural foundation of a defensible compensation program. The bands determine which roles are in the same compensation envelope, where in the envelope each role’s market positioning sits, and how the within-band individual variation is bounded. Done well, salary bands produce internal equity that survives merit cycles, external competitiveness that aligns with the binding-constraint labor markets, and pay-transparency communications that don’t create unintended internal-equity flares. Done poorly, bands become bureaucratic furniture that constrains hiring without producing the equity or competitiveness benefits they were designed to deliver.

The empirical foundation for compensation banding is largely practitioner-developed rather than experimental. Lawler’s “Strategic Pay” series and the Milkovich and Newman compensation textbook articulate the dominant frameworks; the academic literature on internal-pay-equity effects (Card and Mas’s work on peer-salary effects, Trevor and colleagues on pay dispersion) provides the strongest empirical anchors for why band discipline matters. This article walks through the mechanics of band construction.

Data Notice: Salary-band magnitudes and methodologies cited reflect projections based on published compensation data sources (Radford, Mercer, Robert Half, Levels.fyi) and compensation-design literature at time of writing. Specific band widths, midpoints, and progression structures vary widely by industry and organization size; market-data selection significantly affects resulting band structures.

What a salary band actually is

A salary band defines a range of base-salary values (minimum, midpoint, maximum) for a level (or grade) within which roles of similar market value sit. A complete band structure includes:

Levels (or grades) that step from entry-level to senior leadership, defining the ladder.
Per-level band ranges specifying minimum, midpoint, and maximum base salary for the level. Band width (max minus min, divided by min) typically runs ~30-50%, with wider bands at senior levels.
Geographic differentials that shift band magnitudes by labor-market location.
Function differentials that shift band magnitudes by function (engineering vs sales vs operations vs customer-facing).
Compa-ratio targets for individual placement within the band (typically 80-120% of midpoint, sometimes 90-110%).

The band structure interacts with the career-ladder structure: levels in the band are aligned with levels in the career ladder, and the lateral-vs-promotion movement between roles maps onto the band structure. For broader treatment of how bands integrate with career progression, see career ladder design.

Sourcing market data

Salary-band midpoints are anchored to market data. The practical decision is which market and which data source. The principal market-data options:

Radford (technology and life sciences) is the dominant source for technology compensation data, particularly engineering, product, design, and technical leadership roles.
Mercer provides broad cross-industry compensation surveys with geographic breakdowns suitable for general-corporate roles and global programs.
Robert Half publishes annual salary guides with job-family granularity and geographic detail, particularly useful for finance, accounting, administrative, and operations roles.
Levels.fyi aggregates self-reported technology compensation data, useful for specific-company benchmarking but with self-selection bias that generally skews magnitudes upward relative to managed-survey data.
Pave and other compensation-data platforms aggregate participating-company submitted data, with varying coverage and methodology.
Industry-specific surveys (financial services, healthcare, professional services) are dominant in their respective industries.

The market-data choice is consequential. Different data sources can produce midpoints for the same role that vary by ~10-25%, depending on data sample, geographic weighting, and self-reporting characteristics. A defensible program specifies the primary data source, the secondary sources used for triangulation, and the methodology for resolving discrepancies.

The percentile decision

A second consequential decision is which market percentile the band midpoint targets. Common conventions:

50th percentile (median): the typical-cost-control default, intended to keep total compensation broadly in line with market.
65th-75th percentile: typical for binding-constraint functions where competitive intensity for talent is high.
90th percentile or above: less common at the midpoint level; more often used as the maximum of the band for senior leadership roles in competitive markets.

The percentile decision should be function-and-level specific. Anchoring all bands to the 50th percentile produces predictable hiring difficulty in binding-constraint functions; anchoring all bands to the 75th percentile materially inflates total compensation cost across functions where the binding constraint does not apply. The right structure pays-at-or-above-market on the binding constraint and pays-at-market elsewhere.

For broader treatment of how the binding-constraint logic applies to compensation design, see compensation design evidence, and for the economics of when more aggressive compensation positioning pays back, see hiring cost economics.

Band width and progression

Band width (the spread between minimum and maximum within a level) determines how much within-level variation the band accommodates. Wider bands accommodate more performance-and-tenure differentiation within a level; narrower bands force more frequent promotion-or-departure decisions.

Common conventions:

Narrower bands (~25-40% width) at entry and mid-levels, where the level is intended to function as a stepping-stone with relatively rapid progression to the next level.
Wider bands (~40-60% width) at senior IC and manager levels, where the level may be a multi-year destination with substantial within-level performance variation.
Wide bands (~60-80%+ width) at director-and-above levels, where individual-circumstance variation (responsibility scope, tenure, specialty) drives wider appropriate ranges.

Band progression (how midpoints step from one level to the next) typically runs ~15-25% per level at IC levels and larger steps at the manager-IC transition and the director-VP transitions. Progression that is too small collapses the level distinction; progression that is too large produces step-function leveling decisions that don’t match underlying scope variation.

Internal equity and the dispersion problem

Internal equity is a band-design objective alongside external competitiveness. The empirical literature (Card and Mas, Trevor and colleagues) documents that internal-pay-dispersion effects on satisfaction and performance are real and asymmetric: visible above-median peers reduce employee satisfaction more than visible below-median peers raise it, and the effect is amplified by pay-transparency.

The implication for band design is that within-band dispersion should be defensible on the merits (performance, tenure, scope, market) and should be explicable via the compa-ratio structure. Within-band dispersion that is unexplainable on the documented criteria produces internal-equity flares when it becomes visible.

Compa-ratio discipline (tracking individual placements as a fraction of band midpoint, monitoring distributions, and addressing outliers through explicit calibration) is the operational mechanism for internal-equity maintenance. A program that has bands but does not track compa-ratios across the population is operating bands as labels rather than as a discipline.

Pay-transparency interactions

Pay-transparency requirements (US state-level posting requirements, EU pay-transparency directive) interact with band design in several ways. The most important:

Posted ranges in job advertisements force the band ranges into public view. Bands that look defensible internally may produce candidate-side reactions when posted, particularly if the posted range is wide and the candidate’s expected placement is unclear.
Internal candidate-information access under pay-transparency rules forces internal communication about bands; opacity that may have been operationally acceptable becomes operationally untenable.
Cross-jurisdiction differential management. Geographic differentials become visible to employees in other geographies, raising fairness questions that compress the differentials.

The empirical work by Cullen and Pakzad-Hurson (NBER 2021) on pay-transparency effects documents that transparency tends to compress within-organization wage variance over time as employees and candidates anchor on observable peer compensation. Band design under pay-transparency should anticipate this compression effect.

For broader treatment of how compensation design integrates with the broader hiring system, see skills-based hiring evidence and internal mobility and promotion.

Designing the band structure

A defensible band structure has the following:

Documented level structure with role-to-level mappings that are reviewed and updated.
Specified market-data sources and percentile targets per function-level combination.
Geographic differentials that are sourced and reviewable, not invented.
Compa-ratio monitoring at the population and subpopulation level.
Pay-transparency-ready posted ranges that align with internal band structure rather than diverging.
Annual band-refresh cadence that updates midpoints to reflect market movement and tracks band-effective total-compensation cost.

The bands themselves are not the discipline; the operational use of bands is. Programs with carefully constructed bands that do not enforce compa-ratio discipline at hiring and merit cycles produce the same internal-equity problems as programs without bands.

Takeaway

Salary-band construction integrates market-data selection, percentile targeting, band width and progression, and internal-equity discipline into a structure that supports hiring, retention, and pay-transparency simultaneously. The defensible program specifies market-data sources, applies function-and-level appropriate percentile targets, sizes band width to the within-level variation needed, and operates the bands through compa-ratio discipline rather than as static furniture. Pay-transparency interactions will continue to constrain band-design choices over time, and band structures that anticipate the compression effect remain defensible longer than structures that don’t.

Sources

Schmidt, F. L., & Hunter, J. E. (1998). The validity and utility of selection methods in personnel psychology. Psychological Bulletin, 124(2), 262-274.
Sackett, P. R., & Lievens, F. (2008). Personnel selection. Annual Review of Psychology, 59, 419-450.
Lawler, E. E. (1990). Strategic Pay: Aligning Organizational Strategies and Pay Systems. Jossey-Bass.
Milkovich, G. T., & Newman, J. M. (2017). Compensation (12th ed.). McGraw-Hill Education.
Card, D., & Mas, A. (2012). Inequality at work: The effect of peer salaries on job satisfaction. American Economic Review, 102(6), 2981-3003.
Cullen, Z., & Pakzad-Hurson, B. (2021). Equilibrium effects of pay transparency. NBER Working Paper No. 28903.
Radford. (2024). Technology compensation survey. Aon plc.
Mercer. (2024). Global compensation planning report. Mercer LLC.