Compression, Inversion and Stale Market Data
A 320-employee software company last refreshed its salary structure three years ago and its market data is now 18 months old. The merit budget is 2.5%; local CPI has run near 4% for two years. In engineering, two incumbents with two years' tenure earn $70,000 while new graduates in the same job family were just hired at $72,000. Regretted attrition in engineering is 14%, and six of the last nine exit interviews mention pay. Payroll is roughly $30 million. The CFO has ruled out increasing the base budget but will consider a one-time pool of up to $400,000. The CHRO wants a proposal for the next compensation cycle in five weeks.
1. Diagnose what is actually happening here. Name the distinct problems and the analysis you need before recommending a dollar of spend.
Read the worked answer
Three different problems are hiding behind the word 'pay.'
- Inversion. New graduates earning $72,000 in the same job family as two-year incumbents at $70,000. This is a specific, visible, morale-destroying defect. It is cheap to fix in absolute terms and expensive to leave.
- Compression. Tenure differentials are too small to reward progression. Compute the compression ratio - the pay difference between adjacent levels divided by the difference in their midpoints. If it is drifting toward 1.0, promotion stops being worth anything.
- Market lag. Merit at 2.5% against CPI near 4% means roughly 1.5% of real erosion per year, compounding. And the market data is 18 months old; if the market moved with CPI over that period, the structure's midpoints could sit several points below where the market actually is. Refresh the data before allocating anything - spending against an 18-month-old benchmark is how organizations overpay one group and underpay another.
Analysis to run first: compa-ratios and range penetration by job family and tenure cohort; a proper inversion/compression report; separation of regretted from total attrition; and - critically - destination data on the nine engineering leavers. Six citing pay is not proof that pay is the cause. People who leave for a better manager, a better project or a title bump often say 'pay' because it is the unarguable reason. Check where they went and what they were paid there. If they went sideways on pay, the money is not your problem.
2. Allocate the 2.5% merit budget plus the $400,000 one-time pool. Justify the split and explain what one-time money must not be used for.
Read the worked answer
Sizing. Payroll of $30 million means the 2.5% merit budget is roughly $750,000, plus $400,000 one-time - about $1.15 million of total flexibility, of which only two-thirds is recurring.
Recommended allocation.
- $20,000 off the top: refresh the market data. Non-negotiable. Every other decision depends on it, and it is the cheapest line in the proposal.
- Carve ~0.7 points out of the merit budget (≈$210,000) into a targeted market-adjustment pool. General merit spreads money evenly and therefore fixes nothing structural; it is the mechanism that created the compression. Redirect part of it to named individuals whose compa-ratio falls below a defined threshold.
- $150,000 to close the engineering inversion. Decide the policy question first: do you lift incumbents to the new-hire rate, or hold the entry rate and stop creating new inversions? Lifting incumbents without holding the entry rate means you will be doing this again in twelve months. Do both.
- Remaining merit (~$540,000) distributed on performance, with a floor at the bottom of the band.
- The residual one-time pool (~$230,000) as retention lump sums to the specific flight-risk group identified from the separation data - not spread across the population.
What one-time money must not do: correct structural base-pay lag. A lump sum expires, does not fix the compa-ratio, does not affect pension or bonus calculations, and creates an expectation of renewal that becomes a recurring cost you did not budget. Use one-time money for time-bound problems only. Using it to patch base pay is borrowing a problem forward with interest.
Say the uncomfortable thing: if refreshed data shows the structure is materially below market, a 2.5% budget cannot fix it, and the proposal should say so with a number rather than implying this cycle solves it.
3. What equity and communication risks attach to this plan, and how do you manage them?
Read the worked answer
Equity risk - the classic failure mode. Targeted market adjustment tends to flow to the loudest, most market-visible group. Engineering is male-dominated and highly mobile; if the entire pool goes there, you have widened the gap with female-dominated job families that are equally below market but less mobile and less likely to be poached. Before allocating, run the same compa-ratio analysis across all job families and check whether the pattern is market-driven or bias-driven. In most Canadian jurisdictions this is also a pay-equity exposure, and the adjustment you make now becomes the baseline for the next review.
Inversion risk. If incumbents are lifted to $72,000 and new hires keep arriving at $72,000, the inversion simply moves up a level and recurs. Fix the entry rate at the same time or you have purchased one cycle of peace.
Transparency risk. In a 320-person company, pay is discussed. Assume the $72,000 new-hire rate is common knowledge. Communicate the adjustment logic - not individual amounts - in terms of market position and band, before people discover it through a colleague.
Manager capability. Managers will be asked 'why did she get more?' Give them the answer in one page: the individual sat below the market threshold for the band. If the answer is 'she was a retention risk,' you have an equity problem you cannot defend.
Do not promise what is unfunded. The most damaging outcome is announcing a market review, then delivering a 2.5% merit pool. Announce the review, the threshold, and the timeline - and be explicit that not everyone will receive an adjustment.