Merit Increase Matrix: How to Build One That Works

Direct answer: A merit increase matrix converts two decisions into one salary-increase guideline: how strongly an employee performed and where the employee's current salary sits within an approved range. A useful matrix creates meaningful differences between outcomes, fits the approved merit pool when applied to employee-level salaries, and gives managers clear rules for exceptions.

The grid itself is the easy part. The difficult work is deciding whether the performance ratings are credible, whether the salary ranges are current, how employees at or above range maximum should be treated, which pay actions stay outside merit, and how much manager discretion the organization can govern consistently.

This article focuses on the allocation method. The separate guide to building a 2026 merit budget covers how the overall pool is sized and approved.

Last reviewed: August 4, 2026.

What Is a Merit Increase Matrix?

A merit increase matrix is a compensation-planning framework that recommends a base-salary increase by combining:

• Performance outcome, such as Needs Improvement, Successful, or Exceptional
• Salary-range position, usually expressed as compa-ratio or range penetration

The matrix translates the organization's compensation philosophy into manager guidance. It should make the following decisions visible:

• How much more salary movement stronger performance receives
• How the recommendation changes for employees low or high in the range
• Which outcomes receive no merit increase
• Whether the matrix provides one guideline or a permitted range
• What managers may change without additional approval
• How the modeled recommendations stay within the approved pool

A merit matrix recommends a merit action. It does not determine promotions, market adjustments, equity remediation, statutory increases, retention actions, or whether a pay decision complies with applicable law.

Why a Matrix Can Look Differentiated but Behave Like a Flat Increase

Consider an illustrative matrix built around a 3.2% budget. It gives a strong performer 3.5%, a successful performer 3.1%, and a developing performer 2.7%. The labels differ, but the economic distinction is narrow. Managers may move recommendations toward the center, and rounding may reduce the visible spread even further.

The result is an across-the-board increase with performance terminology attached.

Four design choices usually create that outcome:

1. The difference between performance columns is too small.
2. Most employees receive the same rating.
3. Manager discretion is wider than the difference between matrix cells.
4. Market, promotion, equity, and retention actions are mixed into merit.

A matrix only differentiates pay when the organization is prepared to make and communicate real distinctions.

Before Building the Grid, Confirm the Foundations

Do not start with percentages. Start by confirming that the inputs can support a governed decision.

The supporting structure may include governed job architecture, documented job evaluation, current salary ranges, and reviewed market benchmarking. None of those inputs should be treated as automatically correct merely because they exist in a system.

Step 1: Choose a Performance Scale the Organization Can Actually Use

A five-point scale offers more theoretical differentiation than a three-point scale. It also creates more opportunities for managers to interpret adjacent ratings differently.

Review the prior rating distribution by manager, function, location, and level. The cost of the matrix will be driven by the actual distribution, not by the number of columns displayed.

For example, if four out of five eligible employees fall into the same performance category, the recommendation in that column will drive most of the merit spend. The remaining columns may have little financial effect even though they occupy most of the design discussion.

Rating calibration should happen before the final matrix is released. A guide on preparing managers for merit season can support the communication and training layer, but the organization still needs clear performance definitions and accountable calibration owners.

Step 2: Choose Compa-Ratio or Range Penetration

The second axis normally reflects salary position.

Compa-ratio = Employee salary ÷ Range midpoint

A compa-ratio of 1.00 means the employee's salary equals the midpoint. The compa-ratio guide explains how the measure is used across compensation decisions.

Range penetration = (Employee salary − Range minimum) ÷ (Range maximum − Range minimum)

Compa-ratio is often easier to explain because it compares salary with one reference point. Range penetration shows how far salary has moved through the complete range and can be useful when range widths vary.

Neither measure proves that an employee is overpaid, underpaid, or treated fairly. The measure only describes position within the approved range. The range, job match, performance evidence, and relevant employee circumstances still require review.

Step 3: Write the Policy Rules Before Assigning Percentages

Describe the intended decisions in plain language first. A typical policy may state that:

• Higher performance should generally receive a meaningfully higher recommendation.
• Employees lower in a credible range may receive more salary movement than equally performing employees higher in the range.
• The lowest performance outcome normally receives no merit increase, subject to applicable law, agreements, and policy.
• Employees at or above range maximum require separate treatment rather than automatic base-salary movement.
• Promotions, market adjustments, equity remediation, statutory actions, and retention decisions remain separately identified.
• Every manager exception requires an approved reason and may require additional review.

These statements become tests for the grid. A cell that contradicts the written policy should be changed, even when the overall average still fits the budget.

Step 4: Build the First Matrix as a Planning Model

The following 3×3 matrix is illustrative. It is not a market recommendation or a substitute for modeling the organization's workforce.

This grid makes two distinctions. Moving across the columns changes the recommendation for performance. Moving down the rows changes the recommendation for salary position.

Before accepting the design, ask:

• Do the performance categories reflect the actual rating process?
• Are 0.85 and 1.05 meaningful boundaries for these ranges?
• Does the highest recommendation occur rarely enough to remain affordable?
• Is the difference between adjacent cells large enough to remain visible after rounding?
• Would the matrix work differently for hourly, sales, executive, union, or international populations?
• What happens when the recommendation would move salary above range maximum?

Worked Example: Three Employees in the Same Grade

Assume three employees are in the same grade with a $90,000 midpoint. Their ratings and salaries differ.

The example shows why percentage and dollar views are both necessary. Employee C receives the lowest percentage but still receives more dollars than Employee A would receive from a 3.0% flat increase. Employee B receives the largest dollar action because the higher performance guideline applies to a larger salary.

The recommendation is not automatically final. Review whether Employee C is at or above range maximum, whether the range is current, whether any salary issue is actually a promotion or market concern, and whether the performance evidence supports the rating.

Step 5: Model the Cost at Employee Level

A matrix has no reliable cost until it is applied to actual eligible salaries.

Recommended merit dollars = Eligible salary × Matrix guideline × Proration factor

Budget utilization = Total recommended merit dollars ÷ Approved merit pool

The model should include:

1. The dated eligible population and salary snapshot
2. Current range minimum, midpoint, and maximum
3. Expected or calibrated performance outcome
4. The applicable matrix cell
5. Proration for hire date, status, or policy where applicable
6. Currency and rounding treatment
7. Range-maximum treatment
8. Separate fields for promotion, market, equity, statutory, and retention actions
9. The resulting annualized salary increase and current-year financial effect

Run at least three distributions:

• Expected: The most likely rating and range-position distribution
• Higher cost: More employees in stronger performance cells or lower range positions
• Lower cost: Fewer high ratings or more employees high in range

Do not ask managers to absorb a modeling error through informal restraint. If the expected recommendations exceed the pool, change the matrix, the budget, or the policy before launch.

Step 6: Normalize the Matrix Without Removing the Signal

One starting method is to apply a common normalization factor to positive cells.

Normalization factor = Approved merit pool ÷ Modeled recommendation total

For example, if the modeled recommendations total $3.6 million and the approved pool is $3.2 million:

$3.2 million ÷ $3.6 million = 0.889

An illustrative 4.0% guideline becomes approximately 3.56%, while a 6.0% guideline becomes approximately 5.33% before final rounding.

Then inspect the matrix again. A mechanically scaled grid may fit the budget but fail the differentiation test. Watch for:

• Adjacent cells that round to the same manager-facing number
• Top and core performance outcomes that are too close
• Range-position differences that disappear
• Manager discretion bands that overlap several cells
• Minimum increase rules that override the matrix

The final grid should be simple enough to operate and distinct enough to explain.

Step 7: Decide What Happens at Range Maximum

An employee at or above range maximum creates a policy decision. Automatically adding more base salary may weaken the range. Automatically denying any recognition may create a retention or communication problem.

Possible treatments include:

• Capping base salary at the approved maximum
• Providing a non-base lump-sum award where permitted and appropriate
• Reviewing whether the job has changed enough to require formal reevaluation
• Reviewing whether the range itself is outdated
• Using another approved reward that does not permanently increase base salary

The correct treatment depends on compensation philosophy, plan design, local law, contracts, collective arrangements, payroll capabilities, and employee circumstances. A high compa-ratio alone is not proof that the employee is overpaid or placed in the wrong job.

Step 8: Keep Other Pay Actions Outside the Merit Recommendation

Merit answers one question: how should base salary change based on contribution within the current role and the approved merit policy?

Other actions have different reasons:

• Promotion: The employee moves into a job with greater scope. See merit increase versus promotion.
• Market adjustment: Salary is misaligned with the external market or approved structure. See market adjustment versus merit increase.
• Equity adjustment: A reviewed internal difference requires remediation.
• Statutory or contractual adjustment: Law, agreement, or policy requires a change.
• Retention adjustment: The organization addresses a specific documented talent risk.

An employee may receive more than one action in the same cycle. Keep the reason, calculation, budget, approval, and employee communication separately visible.

Step 9: Define Eligibility and Edge Cases Before Managers Enter Recommendations

Document the treatment of:

• Recent hires and employees without a complete review period
• Employees on leave
• Employees with recent or pending promotions
• Employees below range minimum or above range maximum
• Employees with active performance actions
• Part-time employees and salary conversions
• Transfers between managers, entities, countries, currencies, or pay frequencies
• Employees expected to terminate before the effective date
• Employees covered by collective agreements or separate pay programs

Use a consistent data date and preserve later changes. Compensation integrations can reduce conflicting employee, job, and performance records, but every source, refresh date, and exception still needs an accountable owner.

Step 10: Control Manager Discretion and Calibration

A matrix should guide judgment rather than remove it. The governance design should specify:

• Whether a cell is a fixed guideline or a range
• How far a manager may move from the guideline
• Which standard reasons are available
• When a written explanation is required
• Which recommendation routes to skip-level, HR, or Finance approval
• Whether managers may move unused budget between teams
• When ratings and recommendations are calibrated
• When the recommendation becomes locked

Free-text comments alone are weak controls. Standard exception reasons make patterns reviewable. Examples may include sustained exceptional impact, documented compression, critical skill, approved retention action, data correction, or another reviewed circumstance. The reason should not convert a non-merit action into merit.

Monitor override frequency, direction, reason, manager, and employee population. A high override rate may indicate manager behavior, rating inconsistency, stale ranges, poor cell design, or a workforce issue the matrix does not address.

Step 11: Review Pay Equity Before and After Manager Input

A matrix can reduce arbitrary variation, but it can also reproduce differences already present in ratings, ranges, job placement, and manager decisions.

Before launch, test the projected recommendations across comparable jobs and relevant employee groups where analysis is lawful and appropriately protected. After manager input, review:

• Average and median increase percentages
• Increase dollars
• Rating distributions
• Range-position distributions
• Zero-increase outcomes
• Override rates and reasons
• Budget consumption
• Results within comparable job groups

A difference does not establish discrimination. It identifies where compensation, HR, legal, or specialist reviewers may need to evaluate the evidence before final approval. The pay equity audit process covers the broader analytical and remediation workflow.

Step 12: Report Whether the Matrix Worked

The cycle-close report should go beyond average increase percentage. Review:

• Approved pool, final spend, and variance
• Spend by performance outcome and range-position cell
• Average and median recommendation by cell
• Percentage of employees inside, below, and above guidelines
• Override volume and approval outcome
• Employees capped at range maximum
• Separate promotion, market, equity, statutory, and retention spend
• Post-cycle compa-ratio distribution
• Unresolved data, structure, or policy issues

Compensation Analytics can support ongoing internal monitoring, while Compensation Reporting can support fixed, versioned outputs for the review date. The organization remains responsible for the methodology, interpretation, access, and final decisions.

When a Merit Matrix May Not Be the Right Tool

A matrix should not be forced onto every population. Another method may be more appropriate when:

• Performance outcomes are not reliable enough to support differentiated pay
• Salary ranges are missing or materially outdated
• The population is governed by a collective agreement or centrally determined increase
• The organization is making a one-time structural market correction rather than a merit decision
• The workforce is too small for several matrix cells to have meaningful populations
• Sales, executive, hourly, or international pay programs use materially different decision factors

The absence of a matrix does not require an ungoverned process. The organization can still define eligibility, decision criteria, budgets, approvals, documentation, and equity review.

Common Merit Matrix Failures

• Building the percentages before reviewing the rating distribution
• Using stale ranges or incorrect grade assignments
• Assuming compa-ratio proves fairness
• Creating cell differences that disappear after rounding
• Giving managers discretion wider than the matrix differentiation
• Combining merit with promotions, market adjustments, and equity remediation
• Ignoring range-maximum treatment until manager review
• Modeling percentages without calculating employee-level dollars
• Changing the matrix during the cycle without preserving the original version
• Reporting average increases without showing overrides and cell outcomes

How CompBldr Supports the Merit Matrix Process

CompBldr can connect the governed records surrounding the merit decision:

• Job architecture, grades, and salary-range context
• Reviewed market matches and compa-ratio
• Performance data received through approved integrations
• Merit recommendations and separate pay-action fields
• Manager workflows, exception reasons, and approvals
• Budget visibility and cycle reporting
• Analytics and versioned decision records

The matrix remains a human-approved compensation policy. CompBldr does not independently determine performance ratings, legal eligibility, pay equity, the appropriate increase, or whether a manager's explanation is valid. Authorized compensation, HR, Finance, legal, and leadership stakeholders retain responsibility for policy and final approval.

Review the platform's current security information when compensation and demographic data are included in the workflow.

Merit Increase Matrix Checklist

• The matrix has one clearly defined purpose.
• The performance scale is understood and calibrated.
• Jobs, grades, and salary ranges are current enough for the cycle.
• The salary-position measure and boundaries are documented.
• Written policy rules exist before percentages are assigned.
• The proposed matrix has been tested against employee-level salaries.
• Expected, higher-cost, and lower-cost distributions have been modeled.
• Normalization preserves meaningful differentiation after rounding.
• Range-maximum treatment is approved.
• Merit is separated from other pay actions.
• Eligibility and workforce-change rules are documented.
• Manager discretion, exception reasons, and approvals are configured.
• Pre-cycle and post-manager equity review is planned.
• Cycle-close reporting can show cell outcomes, overrides, and budget variance.
• The original matrix, revised versions, approvals, and final decisions are retained.