Annual merit cycle mechanics — running it without breaking trust

60-Second Summary

Annual merit cycles run on a 12-16 week timeline from kickoff to communication. Compressing it below 10 weeks usually means skipping calibration; stretching above 18 weeks erodes credibility.
Budget allocation choice matters more than budget size — flat 3% across the board feels fair, but a 0/3/5/7% performance-tiered distribution produces measurably higher top-performer retention.
Calibration sessions surface and correct rater leniency more reliably than any training program. Skip calibration and your raters' personality becomes your comp policy.
The communication failure that breaks trust: an employee discovering through Slack/Glassdoor that peers got bigger raises despite same/lower performance. Pre-empt with structured manager scripts.

The annual merit cycle is the most operationally complex thing the HR function does in any given year — and it's the single largest single-event trust risk. Done well, employees feel paid fairly. Done badly, you've handed your top performers reasons to interview elsewhere.

The 12-week timeline

Week	Activity	Owner
1–2	Comp committee sets total budget + performance distribution targets	CFO + CHRO
3–4	Manager training + cycle kickoff communication to employees	HRBP
5–7	Managers submit recommendations through comp tool	Managers
8–9	Calibration sessions by org / function	HRBP-facilitated, manager-attended
10	Comp committee final approval; exec team review of edge cases	CHRO + CEO
11	System updates; comp letters generated	Comp ops
12	Manager 1:1 communication to employees; effective-date payroll change	Managers

Budget allocation models

Two ways to spend the same budget

Flat distribution (e.g., 3% across the board)

Politically simple
Feels fair on the surface
Top performers leave because differential signaling is zero
Bottom performers learn they pay zero price for low performance
Typical at risk-averse mature companies

Performance-tiered (e.g., 0 / 3 / 5 / 7%)

Same total budget
Top quartile gets meaningfully more
Low performers get explicit signal
Requires honest performance distribution — no all-meets ratings
Higher top-performer retention; harder middle-of-pack conversation

Calibration mechanics

Calibration is where comp recommendations get sanity-checked across a peer group of managers. The mechanics: each manager presents their recommendations, peer managers question outliers, the facilitator surfaces leniency or harshness patterns. The output is recommended adjustments, not mandates — the manager keeps decision authority but with structured peer pressure.

On forced ranking

Forced ranking (mandating a specific distribution like 20/70/10) was popularized by GE Jack Welch and is now mostly abandoned at major employers. The honest version of calibration: target distributions are a sanity check, not a quota; outliers require justification, not auto-correction.

The communication script

What every manager needs from HR

1
The number
Base change, bonus, equity refresh — all in one document, with effective date.
2
The rationale
1–2 sentences tying the raise to specific performance themes. Generic language ('great year') is worse than no rationale.
3
Position in band
New compa-ratio and a sentence on what it means: 'You are now at the midpoint of your level — this is the market rate for fully proficient L4 engineers in our SF tier.'
4
Forward signal
Honest read on promotion timeline, growth opportunities, or — if applicable — that the trajectory is flat and they should know.
5
How to ask questions
Explicit invitation to follow up, with response timeline (e.g., 'I'll answer any question within 24 hours').

Frequently asked questions

Should managers know the company-wide distribution?

Yes, at least at the aggregate level. Otherwise managers calibrate against their personal team norm — which is how leniency drift compounds over years.

Can we do off-cycle comp adjustments?

Yes, for: pay-equity remediation, retention risk, role change/promotion, market re-benchmark of hot-skill roles. Document the policy explicitly; ad-hoc raises that aren't policy-driven become the precedent everyone references next cycle.

What if our budget can't cover what we'd want to give?

Be honest with employees that the budget didn't support the recommendation you would have made. Then either fix the budget next cycle or change the comp philosophy. The third option — pretending the recommendation was the right size — destroys manager credibility.

The 12-week timeline

Budget allocation models

Calibration mechanics

The communication script

Frequently asked questions

Read next