Compensation design

Compensation is the most visible expression of a company's priorities. Done well, it aligns incentives, attracts the right people, and makes retention cheap. Done casually, it generates resentment, turnover, and the wrong behaviors. Almost every early company I've seen starts with ad-hoc compensation decisions and spends years cleaning up the inequities those decisions created.

The components

• Base salary, predictable, non-negotiable-after-sign cash
• Variable / bonus, performance-tied cash
• Equity, stock options, RSUs, or profit sharing
• Benefits, health, retirement match, PTO, other non-cash value
• Non-monetary, remote flexibility, title, ownership, growth path

Total compensation = all of the above. Candidates evaluate the full package. Compete on total comp, not on base.

Building a compensation structure

1. Define levels

Before any individual compensation decision, define the leveling system. For each function:

• Individual contributor track: L3 (early career) → L4 (mid) → L5 (senior) → L6 (staff) → L7 (principal)
• Manager track: M3 (manager) → M4 (senior manager) → M5 (director) → M6 (VP)
• Each level has a written description of expectations, scope, autonomy, impact

2. Set bands by level

For each level, a salary band: min, mid, max. Typical spread: mid ±15%. Every role in that level must fall within the band. No exceptions.

3. Benchmark to market

Use at least two data sources:

• Third-party benchmarks (Pave, Radford, Mercer, Option Impact for equity)
• Your actual offer acceptance / decline data

Target a market position: typically 50th–75th percentile of your comp market. Above 75th is expensive; below 50th means you lose candidates.

4. Document the philosophy

Write a one-page compensation philosophy. Answers: what percentile do we target? How do we weight cash vs equity? How much variable comp do we use? How do we handle raises? Publish it internally.

Variable comp, the dangerous lever

Variable compensation motivates whatever you measure. Pick wrong and people optimize for the wrong thing. Rules:

• Sales roles, variable comp should be meaningful (40–60% of OTE) and tied to closed revenue, not activities
• Customer success, lean variable (10–20%) tied to retention + expansion
• Engineers / PMs / designers, usually not variable, or minimal (5–10% company bonus). Variable comp in product teams tends to drive the wrong incentives
• Executives, typically 20–40% tied to company-level outcomes (revenue, profitability)

Whatever the comp plan rewards is what you'll get. If you reward bookings, you'll get bookings, even if they're bad deals.

Equity, the long tail

• Early stage, equity is usually the largest component of long-term value. Be generous with early hires; small equity differences compound massively
• Refresh grants, after 2–3 years, initial grants are vesting out. Refresh grants keep people engaged. Budget for this
• Vesting cliff. 1-year cliff is industry standard. Accelerate for terminations without cause, not for voluntary departures
• Secondary opportunities, at later stages, allowing modest secondary sales reduces pressure to leave for liquidity

Raise cadence

Raises happen on a schedule, not on who asks loudest:

• Annual, cost-of-living adjustments for all, tied to market data
• Merit, performance-tied increases, concentrated in top performers (top 20% get 50% of the merit pool)
• Promotion, when someone levels up, a compensation increase follows immediately, not at the next review
• Out-of-cycle, when someone's comp has fallen significantly out of market (e.g., market moved, they've been underleveled), fix it, don't wait

The transparency question

How transparent should compensation be?

• Fully transparent, publish individual comp to the whole company (rare; requires extraordinary culture)
• Structurally transparent, publish bands, levels, philosophy; individual comp is private (most common good answer)
• Opaque, nothing published (causes resentment; employees compare notes anyway)

Structurally transparent is the right answer for most companies. People need to know the system is fair; they don't need to know what their neighbor makes.

The pay equity audit

Annually, analyze comp by gender, race, tenure, and level. Questions:

• Within a level, is there unexplained variance by demographic?
• Are people who joined more recently paid more than tenured people at the same level? (The "tenure tax")
• Are people who negotiated hard paid more than people who didn't, at the same level?

Unexplained inequities get fixed. Not "noted for next cycle", fixed.

What good looks like

• Written compensation philosophy, levels, and bands
• Every offer maps to a level + band; no "special case" compensation
• Annual pay equity audit with specific actions taken on any findings
• Variable comp is tight: rewards the right outcomes, not activity proxies
• Retention offers never exceed what a proactive raise would've cost

Related: Role scorecard · Performance reviews · Hiring