The engineering manager was reviewing two offer letters. Both candidates were being hired as senior engineers. Both had passed the same interview process, both would join the same team, both had similar backgrounds.
One had negotiated aggressively. She was coming in at $195K base salary.
The other, equally qualified but a less confident negotiator, was at $165K base.
Same role. Same level. Same team. Thirty thousand dollars different. Neither candidate knew about the other, but they would find out eventually—engineers always do.
This is what happens without compensation bands, or with bands that exist on paper but aren't enforced. Compensation becomes a function of negotiation skill rather than contribution. It creates legal risk, retention problems, and erodes trust the moment people compare notes.
After helping over 60 companies build engineering compensation structures at SmithSpektrum, I've seen what works and what doesn't. Well-designed bands solve real problems. Poorly designed bands create new ones[^1].
Why Bands Matter
Compensation bands solve several problems simultaneously.
Negotiation inequality is the most obvious problem. Without bands, confident negotiators earn more than equally qualified people who don't negotiate well. Research consistently shows that negotiation behavior varies by gender, personality, and cultural background—meaning unstructured compensation systematically disadvantages certain groups. Bands cap this variance: the same role pays within the same range regardless of negotiation style.
Pay equity gaps often emerge over time in organizations without structured compensation. Small decisions compound: this hire negotiated higher, that person didn't get a raise because of budget timing, this promotion came with less adjustment than that one. Without bands to anchor decisions, drift creates inequity. Bands provide guardrails that prevent accumulating unfairness.
Retention risk increases when compensation is chaotic. Engineers talk. They share salary information with friends, read surveys, compare notes with colleagues. When someone discovers they're paid significantly less than someone doing the same work, retention becomes precarious. Bands prevent the dramatic disparities that trigger departures.
Manager discretion becomes less problematic with bands. Individual managers making compensation decisions based on gut feeling will produce inconsistent outcomes—some generous, some tight, some biased by factors that shouldn't matter. Bands don't eliminate manager judgment, but they constrain it within a defensible range.
Offer consistency matters for hiring velocity. Without bands, every offer becomes a unique negotiation, eating management time and producing variable outcomes. With bands, offers are faster to construct and easier to defend.
The Mechanics of Bands
A compensation band defines the salary range for a given level: a minimum (the floor), a maximum (the ceiling), and typically a midpoint (the target for someone fully proficient at that level).
Band width—the distance from minimum to maximum—is typically 20-40% of the minimum. Narrower bands (15-20% width) work when you have many levels and frequent promotions. Standard bands (25-30% width) suit most engineering organizations. Wide bands (35-50% width) work when you have fewer levels and want more flexibility for growth within each level.
| Band Element | Purpose | Typical Range | Update Frequency |
|---|---|---|---|
| Minimum | Entry point for level | Market 25th percentile | Annually |
| Midpoint | Target for solid performers | Market 50th percentile | Annually |
| Maximum | Top performers, retention | Market 75th-90th percentile | Annually |
| Band width | Flexibility for negotiation | 40-60% of midpoint | At design time |
| Overlap | Allows promotion without raise | 10-20% between levels | At design time |
Bands usually overlap. The maximum of one level might exceed the minimum of the next level. This accounts for the reality that a strong performer at one level might outperform a new person at the next level. Complete separation between bands is unusual and often impractical.
For example, a senior engineer band might run from $160K to $210K. A staff engineer band might run from $195K to $260K. The overlap ($195K-$210K) acknowledges that a top senior engineer and a new staff engineer might have similar compensation, even though they're at different levels.
Designing Your Bands
Band design starts with levels, not numbers. Before setting salary ranges, define what each level means in your organization.
Level definitions should capture scope, impact, and independence. A junior engineer works on guided tasks with limited scope. A mid-level engineer operates independently on full features. A senior engineer leads technical work, mentors others, and handles significant projects. A staff engineer has cross-team impact, influences architecture, and operates with substantial independence. A principal engineer shapes organizational direction and solves the hardest problems.
These definitions anchor your compensation decisions. When someone asks "why am I at this level?" or "what do I need for promotion?", the answer should be substantive, not arbitrary.
Once levels are defined, gather market data. Multiple sources are better than one: Levels.fyi provides tech company data with high reliability. Radford/Aon offers comprehensive but expensive survey data. Carta publishes startup-focused compensation data. Glassdoor and LinkedIn Salary provide broader but less precise benchmarks. Competing offers tell you what the market actually pays for people you're trying to hire.
Set a percentile target that reflects your compensation philosophy. 25th percentile means you're paying below market—appropriate if you have strong non-cash value propositions (mission, equity, flexibility). 50th percentile is competitive baseline. 75th percentile reflects a premium talent strategy. 90th percentile is aggressive top-of-market positioning.
Most tech companies target 50th-75th percentile. Decide your strategy before setting numbers, and be prepared to explain it to candidates and employees.
Geography complicates everything. In a remote world, do you pay the same everywhere or adjust by location?
Options include: single global band (same pay regardless of location—simple but expensive if you're anchoring to San Francisco), location-adjusted bands (pay varies by candidate location—common but complex to administer), hub-based bands (pay based on nearest major tech hub—a middle ground), or zone-based bands (tiers of locations, e.g., Tier 1 = SF/NYC rates, Tier 2 = 90%, Tier 3 = 80%).
There's no correct answer. Location-based pay feels fair to people in high cost of living areas and unfair to people in low cost of living areas. Flat pay feels the opposite. Pick a philosophy and be prepared to defend it.
Placing People Within Bands
Once bands exist, you need principles for where within the band someone should be placed.
Band placement reflects how fully someone fulfills the level expectations. Someone new to a level—whether through promotion or hiring—typically starts in the lower half of the band. Someone fully proficient and performing strongly sits near the midpoint. Someone exceeding expectations at the level but not yet ready for promotion might sit in the upper portion.
Hiring placement often causes confusion. Should everyone start at the band minimum? Should strong candidates start higher?
I recommend placement based on demonstrated fit to the level. Someone who barely cleared the senior bar starts in the lower quarter of the senior band. Someone who clearly exceeded it starts higher. This creates room for growth within the band and maintains consistency.
Special factors can justify higher placement: rare skills that command market premiums, strong competing offers you're trying to beat, experience significantly beyond the level's baseline expectations. Document these factors when they apply.
What shouldn't affect placement: negotiation aggression, demographic characteristics, factors unrelated to the work. If two people with equivalent experience and skills would be placed differently based on how hard they negotiated, your bands aren't doing their job.
Band Progression and Reviews
Bands aren't static. They need annual attention.
Every year, refresh your market data and adjust bands accordingly. The market moves; bands that were competitive in 2024 may be behind by 2026. If you're not updating annually, you're implicitly choosing to fall behind market.
Annual adjustments for individuals combine merit increases (performance-based) with market adjustments (keeping pace with band movement). A typical approach: high performers get 5-10% increases, solid performers get 3-5%, and bands themselves shift 2-4% to track market movement.
Edge cases require clear policies. When bands increase and someone falls below the new minimum, bring them to minimum immediately—you can't pay below your own stated floor. When someone reaches the band maximum but isn't ready for promotion, you face a choice: hold their salary (which may trigger departure), expand the band (which affects your philosophy), or find creative alternatives (spot bonuses, additional equity).
"Red-circled" employees—those above band maximum, often from acquisitions or historical anomalies—present challenges. Options include freezing their salary until the band catches up, accepting the anomaly temporarily, or managing them out if the situation is untenable.
Transparency and Communication
How much should employees know about compensation bands?
The trend is toward increasing transparency, driven partly by legislation (many jurisdictions now require salary ranges in job postings) and partly by culture shifts (employees expect to understand their compensation).
Levels of transparency range from opaque (employees don't know bands exist) to partial (employees see their own band but not others) to full (all bands are published and individual positions are visible).
I recommend at minimum showing employees their own band and where they fall within it. This answers basic questions: Am I fairly paid? How much growth room do I have at this level? What happens if I get promoted?
Showing all engineering bands enables self-service understanding of career paths and reduces the asymmetric information that breeds suspicion. "I wonder if staff engineers make more than me" becomes "I know staff engineers are in X band, and I know where I am relative to that."
Full transparency—individual salaries, not just bands—remains controversial and relatively rare. Some companies embrace it as an accountability mechanism; most find it creates as many problems as it solves.
Whatever your transparency level, communicate about the philosophy. Why do bands exist? How were they set? How do people move through them? What decisions are within manager discretion versus constrained by policy? Proactive communication prevents reactive suspicion.
Common Mistakes
Design mistakes undermine bands before they launch. Bands too narrow leave no room for growth within a level, forcing constant promotions. Bands too wide create huge disparities at the same level. Copying another company's bands without context ignores that your market, philosophy, and levels may be different. Ignoring remote complexity leaves you improvising location decisions inconsistently.
Implementation mistakes undermine bands after they launch. Making exceptions constantly transforms bands into suggestions rather than constraints. Rigid enforcement that loses good candidates on edge cases takes philosophy too far. No manager training means inconsistent application. Poor communication means employees don't understand or trust the system.
Maintenance mistakes let bands decay over time. Set-and-forget means bands fall behind market. Reactive-only updates mean you're always playing catch-up. No documentation means institutional knowledge walks out when people leave. No audit means drift goes undetected until someone notices a serious inequity.
The most dangerous mistake is having bands that exist on paper but aren't followed in practice. This combines the worst of both worlds: administrative overhead without the fairness benefits. If you're not going to enforce bands, don't have them.
Measuring Effectiveness
How do you know if your bands are working?
Track the percentage of employees within their bands. If more than 5% are outside their bands (either above maximum or below minimum), something is wrong—either the bands don't match reality or the exceptions have gotten out of control.
Track compa-ratio: each person's salary divided by their band midpoint. Average compa-ratio across the organization should be close to 1.0, indicating that on average people are near the target for their level. Significant deviation suggests systematic over- or under-payment.
Track pay equity ratios: average compensation by demographic group at the same level. Differences that can't be explained by experience or performance indicate bias that bands should have prevented.
Track offer acceptance rate. If your bands are below market, you'll see it in declined offers. If acceptance rate drops below 70% for candidates who reach the offer stage, compensation may be the problem.
Track attrition where compensation is cited as a factor. Some attrition for compensation is inevitable—you can't and shouldn't match every competing offer. But if compensation is a primary factor in more than 15% of regretted departures, your bands may be below market.
The engineering manager with the $30K disparity between same-level hires? The company implemented bands the following year. They defined ranges for each level, set placement criteria, and trained managers on consistent application.
The two engineers were both brought within the senior band's appropriate range over the next review cycle. One's salary stayed flat while the band moved; the other received an adjustment to reach appropriate placement.
One engineer was mildly annoyed that their negotiated premium had normalized. The other was relieved to finally be paid fairly.
That's exactly what bands should do.
References
[^1]: SmithSpektrum compensation consulting, 60+ companies, 2019-2026. [^2]: Carta, "Total Compensation Report," 2025. [^3]: Radford/Aon, "Global Technology Survey," 2025. [^4]: PayScale, "Compensation Best Practices," 2025.
Building compensation bands? Contact SmithSpektrum for compensation structure design and benchmarking.
Author: Irvan Smith, Founder & Managing Director at SmithSpektrum