Pay Compression and a Pay Transparency Deadline

Do not spread 3.0% evenly. A flat merit increase is below inflation, so it is a real-terms pay cut for everyone and a competitive loss specifically for the engineers who are already below midpoint. Peanut-buttering spends the entire budget and buys nothing.

Segment by where the money changes behavior. The discriminating data point is 62% of engineers below midpoint while R&D attrition is 19% versus 11% company-wide. That is not a market-wide pay problem; it is a targeted compression problem in one function. Concentrate the pool on engineers whose salaries are furthest below range midpoint and whose scarcity is highest - likely 15–20% of the engineering population absorbing most of the pool, with a smaller general increase elsewhere.

Compare with the retention bonus. A one-time bonus to the top 40 engineers is cheaper in year one and worse in year two: it does not fix the base salary, so it leaves compression in place and the recipient is still underpaid the day the bonus is spent. Bonuses buy 12 months; they do not change the underlying competitive position. They can, however, be a legitimate bridge while the structure is fixed - with a clawback and a clear statement that base pay will be addressed.

Recommendation: use the merit pool for targeted base correction, use a small, time-limited retention bonus only for people with live external offers, and go back to the CFO with the attrition cost, not the increase cost. Nineteen percent R&D attrition carries replacement and project-delay costs that usually exceed the compression gap.

The law converts internal pay structures into public-facing documents. Everything that is currently informal becomes a question a candidate or employee can ask and a regulator can inspect.

Prerequisites:

• A defensible job architecture - every role mapped to a level and a range, with a rationale for the range. Sixty-two percent of engineers below midpoint is not automatically unlawful, but it must be explainable.
• A pay equity audit by level, function, tenure, and protected class, with remediation funded before disclosure, not after a complaint.
• A documented range-setting methodology using refreshed market data. The existing data is 18 months stale, which is precisely the condition that makes disclosure indefensible.
• Manager capability. Managers will receive pay-range questions on day one. Unprepared managers create the disputes that turn into claims.

Highest-risk gap: pay equity in engineering exposed by disclosure. The combination of stale market data, mass below-midpoint placement, and high attrition means the company cannot currently explain engineer pay by level or tenure. Disclosure to an internal candidate asking about their range will surface this immediately - internally, at scale, before any external scrutiny.

Sequencing: equity audit and job architecture first (months 1–4), range refresh and remediation (months 4–7), manager training and communication (months 7–9). Remediation must be spent, not planned, before the effective date.

The flaw is that the plan measures activity and booking value rather than realized revenue. When payouts exceed target while revenue misses by 8 points, the plan has become a cost center disconnected from the outcome the company is buying. Pay has lost line of sight to results - the defining failure mode of incentive design.

Likely mechanical causes to test:

• Measure mismatch. Commissions paid on bookings or contract value while revenue is recognized over time - the sales force is paid for signatures, not for revenue.
• No performance threshold or accelerator cliff. Payout begins at too low a threshold and accelerates before the revenue target is reached.
• Multi-year deals credited in year one, so payout timing runs ahead of revenue timing.
• Discounting penalty absent. Reps can hit volume by cutting price, which raises bookings and lowers revenue - and the plan rewards it.

The fix: shift the primary measure to an outcome the rep controls and the company books - revenue recognized, margin, or gross profit - with a threshold below which no incentive pays and a decelerator for over-discounting. Cap the total payout and separate any one-time deal credit across the recognition period.

Transition carefully. A mid-year plan change breaks trust and can trigger claims. Either implement at the start of the next plan year with a documented acceleration to it, or run the current plan through its term and add a one-time correction for the specific mechanical defects. Communicate the why through a total rewards and detailed compensation statement - the company currently has no statement at all, which is why a 130% payout felt to sales like deserved success rather than a plan failure.