Merit increase guidelines are the rules that decide how raises get handed out: who qualifies, how much each performance level earns, and what managers can and can’t do on their own.
Good guidelines make raises consistent and defensible. Weak ones turn every cycle into a negotiation, and the difference usually comes down to one adjustment that separates guidelines that feel fair from guidelines that only look fair.
This is written for teams of roughly 100 employees and up, with defined roles and someone who owns compensation. If you’re smaller than that, the flat-percentage table below is usually all the structure you need.
What Should Merit Increase Guidelines Include?
A complete set of merit increase guidelines answers six questions before the cycle opens. If any of these lives only in someone’s head, a manager will eventually ask about it mid-cycle and get an inconsistent answer.
1. Eligibility
Who qualifies this cycle? The common default is anyone active on both the recommendation and payment dates, in their role for at least 90 days, with a current performance rating on file.
Most organizations also exclude employees on a performance improvement plan and prorate or defer increases for those who started partway through the review period.
The cases that derail cycles are the ones nobody decided in advance: the employee on parental leave, the one who gave notice last week, the internal transfer who changed managers in October. Write your answer to each before the cycle, not when a manager emails you in the middle of it.
2. Performance tiers and ranges
Managers reference this part more than any other, so it carries the most weight: the number of performance levels you use and the percentage range that maps to each. The typical ranges are below.
3. Budget envelope
Set the total merit pool, usually a percentage of base salary, and decide how it splits across teams. Your ranges have to be fundable inside this number, which is harder than it looks and is the single most common way cycles go wrong. The “Three Mistakes” section covers what happens when the two don’t reconcile.
4. Discretion limits
Decide what a manager can approve alone and what needs sign-off. The common default is a soft target range managers can move within freely, with a hard ceiling (often the salary band maximum) that requires HR or second-level approval to exceed. Settle this before the cycle, because managers will assume they have more latitude than you intend if you don’t say otherwise.
5. The exception process
How a manager requests something outside the guidelines, who approves it, and what justification is required. Even strict guidelines need a release valve for genuine retention risks. Require a written reason for every exception. It keeps the volume down and gives you a record when someone asks why two similar employees were treated differently.
6. Off-the-table cases
What happens with employees already at or above their salary band maximum. The standard practice is a one-time lump sum rather than a base increase, so you reward the performance without permanently inflating a salary that’s already above market. Spell out the calculation so managers don’t improvise it.
Typical Merit Increase Ranges
The percentage ranges tied to each performance tier need to do two things: create real separation between performance levels, and hold up when an employee asks why they got 2.5% and a colleague got 4.5%.
A standard five-tier structure looks like this:
| Performance Rating | Typical Range | Approximate Share of Population |
|---|---|---|
| Exceptional | 5.0% to 7.0% | 5 to 10% |
| Exceeds Expectations | 3.5% to 5.0% | 15 to 20% |
| Meets Expectations | 2.0% to 3.5% | 50 to 60% |
| Partially Meets | 0% to 1.5% | 10 to 15% |
| Does Not Meet | 0% | Not eligible |
WorldatWork’s 2025 Salary Budget Survey put the median merit budget at roughly 3.5% across most industries, and Mercer’s 2025 data landed in the same range. Treat those as market references, not targets. Your ranges should reflect your own budget, your market positioning, and what your performance distribution actually looks like.
One warning sign: if your top tier allows 5.0% to 7.0% but almost everyone in it lands at exactly 5.0%, the upper range is decorative. Either managers don’t realize they have room to move, or the budget never supported the top end. Worth knowing before the next cycle, not after.
A bigger problem hides inside even a well-calibrated table, though, and it’s the one most guidelines never address.
The Adjustment That Makes Guidelines Actually Fair
Here’s where most guidelines quietly go wrong. They apply the same percentage to everyone at a given performance level, ignoring where each person sits in their pay range.
Don’t have salary bands yet? This section won’t apply to you, and that’s fine. Use the flat-percentage table above for now, and treat building a salary band structure as the prerequisite project. Everything below depends on it.
Compa-ratio measures where someone sits: their salary divided by the midpoint of their band. Someone earning $70,000 in a band with a $77,000 midpoint has a compa-ratio of about 91%, meaning they’re paid below the middle of the range for their role.
When you give the same 3% increase to someone at 82% of midpoint and someone at 114%, it looks even-handed. Over a few cycles it isn’t.
The person below midpoint keeps losing ground to the market every year. The person above midpoint, already paid well, pulls even further ahead. The raise that was supposed to reward performance ends up rewarding whoever negotiated the better starting salary.
The pattern shows up in exit interviews before it shows up in your data. A senior analyst rated “exceeds expectations” three years running, hired at 81% of midpoint, watches a lateral hire start at 97% for the same role. She doesn’t file a complaint. She takes the recruiter’s call. By the time you see the resignation, the flat-percentage math has already done the damage.
A compa-ratio-adjusted matrix fixes this by giving larger increases to strong performers who are underpaid relative to their range, and smaller ones to those already above midpoint:
| Performance Rating | 80 to 89% of Midpoint | 90 to 109% of Midpoint | 110% and Above |
|---|---|---|---|
| Exceptional | 6.0% to 8.0% | 4.5% to 6.5% | 3.0% to 4.5% |
| Exceeds Expectations | 4.0% to 5.5% | 3.0% to 4.5% | 1.5% to 3.0% |
| Meets Expectations | 2.5% to 4.0% | 2.0% to 3.0% | 1.0% to 2.0% |
| Partially Meets | 0% to 1.5% | 0% to 1.0% | 0% |
When you adjust this way, a manager will eventually ask why one employee got a smaller raise for the same rating.
The answer belongs in your guideline document, not in a tense one-on-one: compensation manages to a range, and an employee already paid above the market midpoint for their role is being kept competitive, not shortchanged.
A raise that holds their position in the band is still a raise.
Three Mistakes That Break Merit Cycles
Most cycles that go sideways fail for one of three reasons, and all three are preventable before the cycle opens.
The first is promising ranges you have to walk back. Checking that your ranges fit your budget is a math problem, and it’s covered above. The damage when you skip it is a people problem.
Once managers have entered recommendations and, worse, hinted at numbers to their teams, pulling those numbers back doesn’t read as a budget correction. It reads as the company going back on its word.
You spend the next two weeks in damage-control conversations, and the trust you lose with your managers outlasts the cycle. Model the pool against real headcount before anyone touches a recommendation, because the cost of getting it wrong isn’t measured in dollars.
The second is leaving guidelines undocumented. Managers reference the matrix and the percentage table, but the eligibility rules, exception process, and discretion limits often exist only as tribal knowledge. Put all of it in a single one-page reference.
When Reworld tightened their guideline documentation and automated approval routing, their planning cycle dropped from 12 days to 3 and manager participation reached 90%.
The third is changing the framework every year. Recalibrate your percentage ranges annually against fresh market data, but keep the structure (tiers, compa-ratio logic, eligibility, exceptions) stable. Reinventing it each cycle erodes manager confidence and makes year-over-year comparison nearly impossible.
Frequently Asked Questions
What’s the difference between merit increase guidelines and a merit matrix?
Guidelines are the full set of allocation rules: eligibility, budget, discretion limits, and exceptions. A merit matrix is one piece, a grid crossing performance rating and compa-ratio to produce a recommended range. Guidelines without a matrix work. A matrix without guidelines does not.
What is a typical merit increase percentage?
For 2025, the median merit budget was about 3.5%, per WorldatWork. Strong performers commonly receive 4% to 7%, solid performers 2% to 3.5%, and underperformers little or nothing. The spread matters more than the average, which is the whole reason performance tiers exist.
Should merit guidelines change every year?
Recalibrate the percentage ranges annually against current market data and your prior year’s payout distribution. Keep the underlying structure stable, because changing the framework every year confuses managers and breaks trend analysis.
How do I set ranges when my budget is below the market median?
Build ranges around the budget you actually have, and tell managers the context before the cycle opens. People accept a tight year when it’s explained. What they don’t forgive is finding out mid-cycle that the top of the guideline was never fundable.
How does this connect to the broader merit process?
Guidelines are the rulebook; the merit increase process is how you run the cycle against them. If you’re building guidelines for the first time, read that next.
Putting Your Guidelines to Work
If you do only one thing differently this cycle, document the six decisions above before you open the planning window. The mid-cycle scrambling that comp teams treat as the normal cost of doing business mostly traces back to one thing: running a cycle on rules nobody wrote down.
The teams that avoid it made the same decisions you will, just on a quiet Tuesday in advance instead of in real time with a manager waiting on a reply.
CompLogix handles merit cycle administration with configurable guidelines, approval routing, and budget modeling that shows whether your ranges fit your pool before any recommendations go in. For organizations running different guidelines across business units or regions, it manages all of them without a separate spreadsheet for each.
Want to see how it works on your own structure? Request a demo.