The Startup Compensation Philosophy: A Founder''s Guide to Building a Pay System That Attracts Talent, Preserves Equity, and Scales Without Breaking
Most startups treat compensation as a series of one-off negotiations. The first ten engineers each got a different base, a different equity grant, and a different bonus structure — all priced from scratch, based on what the candidate asked for and what the founder guessed was reasonable. The eleventh engineer walks in and asks a colleague what they make and the founder discovers they''ve built a system that''s indefensible.
The founders who avoid this outcome commit to a compensation philosophy early — before the fifth hire, ideally, and certainly before the twentieth. Not a spreadsheet. A philosophy: a set of principles that determine how pay is set, how it changes, and how it''s communicated. Then a set of mechanics that operationalize the philosophy consistently.
This guide covers the specific components: the philosophy choices, the benchmarking discipline, the leveling framework, the equity refresh model, the raise and promotion cadence, and the transparency posture.
Ad-hoc pay creates inequities that eventually become public. No system stays private forever. Employees compare notes, current or former team members leak salary bands, someone posts on a Blind board. When the inequities become visible without a defensible philosophy behind them, trust collapses.
Compensation is a scarce resource. Startups have limited cash and limited equity. Every dollar and every share spent without discipline is one that can''t be spent later. Undisciplined comp burns both faster than expected.
Retention depends on it more than recruiting does. Pay above market gets people in the door. Pay perceived as fair keeps them there. Pay perceived as unfair drives them out — even if the absolute number is high.
It''s a signal about the company''s operating maturity. Serious executives evaluate the comp system as a proxy for how the company is run. A rigorous system signals rigor everywhere. An ad-hoc system signals ad-hoc everywhere.
Six foundational choices that define the pay system. Make them explicitly, not by default.
Choice 1: Where in the market do you pay? Options: 50th percentile (median). Standard for early-stage companies. Attractive to candidates who value equity upside. Requires strong story about why the equity is worth the discount. 75th percentile. Above-market cash. Attractive to candidates who need higher base for personal reasons. Burns cash faster. Common for later-stage or well-funded companies. 90th percentile. Top-of-market cash. Attractive to candidates who don''t care about equity. Very expensive. Common for FAANG-adjacent talent wars.
Pick one and defend it. Don''t pay 50th for some hires and 90th for others without a documented reason.
Choice 2: How do you weight cash vs. equity? A candidate valued at $300k total comp can be $200k cash / $100k equity, or $150k cash / $150k equity, or $250k cash / $50k equity. Early-stage typically over-indexes on equity; later stage typically over-indexes on cash. Have a documented ratio by level.
Same pay everywhere. Simple. Overpays in low-cost geos, underpays in high-cost geos.
Tiered geo bands (SF/NYC baseline, then discount tiers for other cities/countries). More complex, more defensible.
Local market rate for each geo. Most complex, most defensible, most expensive to administer.
No-negotiation offers. You make one strong offer. Take it or leave it. Removes bias toward negotiators.
Bounded negotiation. Small cash flexibility, no equity flexibility.
Full negotiation. Everything is a negotiation. Rewards negotiation skill, not job skill.
The trend among rigorous companies is toward no-negotiation or bounded — because full negotiation systematically overpays candidates from over-represented demographics and underpays others.
Point salaries per level. Every senior engineer makes exactly $X. Simple, transparent, but rigid.
Bands per level (e.g., $180k–$220k). More flexible, harder to game, requires discipline about where in the band people land.
Bands are standard once the company is past ~50 people. Point salaries work in the very early stages.
Fully transparent (every salary visible to every employee). Extreme, requires strong culture. Very few companies do this well.
Bands transparent. Every employee can see the salary bands for every level. Their own level and band is disclosed to them.
Confidential. Standard. Bands not shared broadly, individual comp confidential.
Bands-transparent is increasingly the norm for rigorous startups. It balances fairness against operational simplicity.
Compensation surveys (Pave, Option Impact, Aon Radford). Paid, high-quality, segmented by stage and geography.
Peer network. Sharing bands with 5–10 similar-stage companies produces informal but useful benchmarks.
Candidate data. Every offer negotiation produces a data point about market rates.
The refresh cadence: benchmarks change 5–15% per year in hot markets. Refresh bands annually. In extreme market moves (2021 tech comp inflation, 2023 tech comp deflation), refresh mid-year.
The rule: don''t make offers without recent benchmark data. "I think this is a reasonable number" is not a benchmark.
Leveling rubrics. For each level, a written rubric of expected behaviors. Scope, ambiguity, impact, cross-functional leadership. When you evaluate a candidate or promote an employee, you evaluate against the rubric — not against your gut.
The rule: every offer must be leveled before comp is set. "This person seems like an L3, we''ll pay L3 rates" is the workflow — not "let''s pay them $X and figure out the level later."
New-hire equity gets a lot of attention. Refresh equity — the grants employees get after their initial vest — gets much less, and drives more retention.
The problem without refresh. Employee joins in year 1 with a 4-year vest. By year 3, they''re 75% vested. Their equity upside is largely captured. A recruiter calling with a fresh 4-year grant elsewhere is now offering more forward-looking upside than they have staying. If you don''t refresh, they leave.
The refresh model. Every employee receives an annual equity grant on their anniversary. Sized as a percentage of their initial grant (typically 20–33%). Vests over 4 years like a new grant.
The math. By year 4, an employee has their original grant fully vested plus 3 refresh grants at various vest states. Their forward-looking vesting is meaningful. The retention incentive persists.
Performance-based refresh. Higher performers get larger refresh grants. Lower performers get smaller or zero. This lets equity work as a performance tool, not just a retention tool.
Ad-hoc raises destroy the comp system. Structured raise cycles preserve it.
Cadence: annual, usually tied to fiscal year end or a defined comp cycle (e.g., April 1).
Merit raises. Based on performance rating, applied to base. Typical range 2–5% for meeting expectations, 5–10% for exceeding, 0% for below.
Promotions. Level changes. Recommended by manager, calibrated across the company, approved by leadership. Comp adjusts to new band.
Market adjustments. For employees whose comp has fallen below the new band after benchmark refresh, one-time adjustments.
Calibration. Before raises are communicated, leadership team reviews all proposed raises together. Ensures consistency across managers, catches inequities, prevents rating inflation in one team vs. another.
Off-cycle raises. Should be rare. Exceptions for counter-offers, promotion to fill a critical role, or correcting a significant inequity. Every off-cycle raise erodes the discipline of the annual cycle — grant them sparingly.
At offer time: communicate the full package clearly. Base, bonus (if any), equity (with strike price and 409A), refresh policy, benefits value.
During the year: managers should be able to answer "am I paid fairly" honestly. If the answer is "you''re at the low end of your band because we hired you before the last refresh," say that.
At raise time: communicate the reasoning. "You got a 4% merit raise because you were rated ''meets expectations''; the median for that rating this cycle was 3.5%." Vague raise communications feel arbitrary.
1. No philosophy. Every offer is a fresh negotiation with no framework. Inequities pile up. Eventually visible. Eventually destructive. 2. Overpaying the loud ones. Candidates and employees who negotiate hard get more. Ones who don''t, get less. Systematic bias. Erodes trust when discovered. 3. No refresh. New hires get all the equity attention. Existing employees who joined years ago watch their vest complete and leave for new grants elsewhere. 4. Off-cycle raise chaos. Whenever an employee threatens to leave, they get a counter-offer. Everyone learns that threatening to leave is the way to get paid. Discipline collapses. 5. Level inflation. Every hire is a "senior" or "staff" because it makes the offer easier. Levels lose meaning. Real seniors resent the dilution. 6. Secrecy without integrity. Bands hidden from employees. When employees discover them through leaks, the discovery is worse than the original disclosure would have been.
Compensation is a system, not a series of negotiations. The philosophy choices — where in the market you pay, cash vs. equity mix, geo differentials, negotiation posture, bands vs. points, transparency posture — define the system. The mechanics — benchmarking, leveling, refresh, cadenced raises, calibration, honest communication — operationalize it.
The founders who build a rigorous comp system early scale it to 500+ people without a compensation crisis. The founders who don''t inherit years of accumulated inequities that eventually explode — usually at exactly the moment (a funding round, a departure, a public post) when the explosion does maximum damage.
