How to run a merit cycle

Once a year, most organizations take a fixed pot of money from finance and turn it into a base pay increase for every employee. It sounds like arithmetic. In practice it is the single most scrutinised thing HR does all year, and the process that most often gets run in a tool that was never designed for it.

This is the whole cycle, in order, written for someone who has to actually run one.

The eight stages

1. Set the budget. Agree a merit budget with finance, expressed as a percentage of eligible payroll, and decide whether it is one pool or several.
2. Clean the census. Assemble one row per employee with job, salary, rating, tenure, manager, currency, and salary range. Fix inconsistent job codes and missing ratings before anything else.
3. Build the merit matrix. Define how increase percentage varies by performance rating and position in range, so higher performers below midpoint move furthest.
4. Allocate down the organization. Push the budget from company level to division, to manager, and finally to individual recommendations that sum exactly to the pool.
5. Correct unexplained pay gaps. Compare employees in comparable roles with comparable tenure and rating, and prioritize closing differences your policy cannot justify.
6. Open manager planning. Give each manager their own population, their own budget, live tracking, and guardrails, then set a hard submission deadline.
7. Route approvals. Send submitted plans up the management chain for review, with a full audit trail of who changed what.
8. Communicate and load. Issue reward statements to employees and load approved increases back into the HRIS ahead of the effective payroll date.

1. Setting the budget

The cycle starts with a number from finance — typically expressed as a percentage of eligible payroll. In recent US cycles most organizations have planned somewhere between 3% and 4%, but your figure should come from your own market data, not a benchmark article.

The more consequential decision is whether that number is one pool or several. A single global percentage is simple and quietly wrong wherever markets move at different rates: if salaries in India are climbing far faster than in the US, applying one blended figure underpays one population and overpays the other. Separate pools by country, and often by division, reflect reality.

You can work out your own pool with our free merit budget calculator.

2. Cleaning the census

The census is one row per employee: identifier, job title and code, department, manager, current base salary, currency, country, performance rating, time in role, and salary range or grade. Every HRIS will export it.

Almost none of them export it cleanly. Expect inconsistent job codes, missing ratings for recent hires, salaries stored as text, and employees whose manager left in October. Fix this first. Every downstream problem in a merit cycle is a data problem wearing a disguise.

3. Building the merit matrix

A merit matrix converts two inputs into an increase percentage: performance rating and position in salary range, often expressed as a compa-ratio. The logic is straightforward once stated: a strong performer sitting near the bottom of their range is both underpaid and worth keeping, so they should move furthest. A weak performer already above the top of their range should move least, or not at all.

Getting this grid right is where pay-for-performance stops being a slogan. If your top performers receive 4% and your average performers receive 3.5%, you do not have a pay-for-performance culture; you have a cost-of-living adjustment with extra paperwork.

4. Allocating down the organization

The pool now has to travel from company level, through divisions, to individual managers, and finally to each employee. Two constraints apply simultaneously: every manager needs a budget proportionate to their population, and the individual recommendations must sum exactly to the pool.

This is the step that breaks spreadsheets. Allocation by hand across dozens of workbooks reliably lands a few percent off, and you find out during reconciliation — after managers have had conversations with their people.

5. Correcting unexplained pay gaps

Somewhere in your data are two people in the same job, hired around the same time, performing at the same level, being paid materially differently. Nobody decided that. It accumulated — through starting salaries, a promotion timed well, a year someone negotiated and someone else did not.

Most organizations treat this as a separate remediation project requiring its own budget and its own executive conversation, which is precisely why it keeps getting deferred. The better approach is to fold the correction into the merit allocation, so the gap closes using money you were already spending. More on this in our guide to pay equity in the merit cycle.

6. Manager planning

Managers make the final call, and they should — they hold context no dataset contains. What they need is a strong recommendation with its reasoning attached, a bounded range for adjustment, and a live budget counter so that any increase they add visibly comes from somewhere else.

What they usually get is an emailed spreadsheet tab with no guardrails, no visibility of their remaining budget until they add it up themselves, and no way to see whether they are treating two similar employees consistently. Then someone spends a week collecting the files back in.

7. Approvals

Submitted plans route up the management chain. The requirement here is unglamorous but non-negotiable: a complete audit trail of who changed what and when. When an executive asks in March why a particular increase is what it is, the answer needs to be on a screen, not in somebody's memory of a February conversation.

8. Communication and load

Employees receive a reward statement explaining their increase, and approved figures load back into the HRIS ahead of the effective payroll date. Work backwards from that payroll date when you build the timeline — it is the one date in the whole cycle that genuinely cannot move.

A realistic Q1 timeline

Roughly 70% of organizations run merit in Q1. Working back from increases effective 1 March:

• October / November — agree the budget with finance; decide pool structure.
• November — pull and clean the census; confirm the merit matrix and proration rules.
• December — configure the cycle; generate and sanity-check recommendations.
• January — open manager planning; hold a hard submission deadline.
• Early February — approvals routed and closed.
• Mid February — reward statements issued; increases loaded to payroll.

Where software actually helps

Not everywhere. If you have 150 employees, five managers, and one budget pool, a spreadsheet is a reasonable tool and you should keep using it.

The threshold is not employee count, it is manager count — because a spreadsheet cycle scales by adding one workbook per planner. At 20 planners it is uncomfortable. At 100 it is a logistics project. Add multiple budget pools, several countries, or a reorganization mid-cycle and it arrives sooner. We laid the full comparison out in spreadsheets vs. HyperEquity.