Guide · Comp policy

Who approves a pay exception?

An exception nobody owns becomes the rule within a year.

Short answer: Pay exception approval should sit with someone who is not the person asking and who can see what peers earn. Each exception needs a reason from a short written list recorded at the time. Where the reason will pass, it needs an end date. Count exceptions once a year. A rising count means the structure needs fixing, not more approvals.

Thirty-one one-offs

Picture a 260-person payments software company in Raleigh, with a team of 40 in Dublin. It has salary bands, a merit guide and a written rule that offers sit inside the band. Its first proper pay equity review turns up 31 people whose pay sits outside the band or well above peers at the same level. None of them is a mistake in the usual sense. Each one was a decision.

A hiring manager went above the band to close a senior engineer who had a competing offer. A director matched an outside offer for a product manager in March. The CEO approved a promotion raise twice the size of the guide because the person had been "doing the job for a year". A sales leader agreed a retention payment that quietly became a base increase at the next review. Each decision was approved by someone, usually in a chat thread or an email. Almost none of them recorded why.

Then a new engineer, hired at the same level as a colleague who joined a year earlier, finds out the colleague earns about $18,000 more. She asks HR why. HR can find the approval but not the reason. The manager who approved it has left. The Dublin team lead asks a harder question: Under the EU pay transparency rules coming into force, would the company be able to explain any of the 31?

This is a composite of a common situation at companies this size, not a client. The figures are illustrative. The pattern is not rare. Pay exception approval rarely fails because people approve bad requests. It fails because nobody owns the question of who approves which exception, on what grounds and what happens to it afterward.

Why pay exceptions are hard to govern

A pay exception is any pay decision outside the written rule: An offer above the band, a raise outside the review cycle, an increase above the merit guide, a retention payment or a counteroffer. Every pay structure needs some. Rules written for the typical case will meet cases they did not foresee. The question is not whether to allow exceptions but how they are decided.

Two good goals pull against each other. The first is speed and local judgment. The manager closest to the situation knows the candidate, the competing offer and the cost of losing the person. A process that takes two weeks loses the candidate. The second is consistency. Every exception is a precedent. The next person in the same role, with the same competing offer, will expect the same answer. Colleagues compare notes. An exception granted to one person is, within a year, the price of the job.

The people best placed to judge the situation are also the people with the strongest reason to say yes. A hiring manager with an open role and a deadline will rarely decline a candidate over $10,000. A director facing a resignation will rarely let a strong performer walk. Neither sees the peer group across the company. That is not a character flaw. It is where they sit.

The pressure is real. In an Australian survey of 500 hiring managers published in June 2026 by a recruitment firm, 85% of employers said they had made a counteroffer to an employee with an outside offer in the past year. Of those employees, 32% still left within 12 months and another 7% turned the counteroffer down [1]. In that market at least, counteroffers are routine. Routine decisions need a rule.

And the stakes have risen. Pay discrimination law in the US and the EU asks employers to show that a pay difference rests on a legitimate factor that the employer actually applied and applied the same way to everyone [4] [5] [6]. An undocumented exception is a pay difference with no stated reason. It is hard to defend for that reason alone.

Five traps in approving pay exceptions

Most of these traps share a root. Each exception is decided as a one-off and nobody looks at what the exceptions add up to.

"The hiring manager knows the candidate best"

What it looks like: The person who wants to make the offer also approves it. Offers above the band need only the hiring manager's sign-off, or a recruiter's. Over a year, new hires in competitive roles land steadily higher than the people already doing the same work.

Why smart teams fall into it: Speed matters in hiring and the manager does know the candidate. Adding an approver feels like bureaucracy at the exact moment the company is competing for someone.

What it costs: Offer-stage discretion is where many pay gaps start. A peer-reviewed study of more than 700,000 initial salary offers made in the US from 2017 to 2020 found a 5.5% gap between offers to women and men, after controls for job, employer, occupation, industry, location and qualifications [3]. Those gaps sit in the first number the company puts on the table, before any negotiation. When the person making the offer is also the only person approving it, nobody compares it against the peer group. The gap shows up a year later as inversion, with new hires paid more than experienced staff, or in the equity review.

"We'll just match the outside offer"

What it looks like: A valued employee brings in an offer from another company. The manager matches it, often within a day. There is no rule about who can counter, by how much or whether the role's band should move instead.

Why smart teams fall into it: Losing a strong performer is expensive and the offer is concrete. Matching it feels like a market correction, priced by the market itself.

What it costs: Counteroffers reward the people who interview elsewhere and say so. Research from Sweden makes the cost visible. Using national register data, economists found that better outside job opportunities raised men's pay in their current jobs but not women's, even though men and women changed jobs at similar rates. The researchers estimate that this kind of renegotiation accounts for roughly half of the gender pay gap among workers in the same workplace and occupation. Where pay was set by collective agreements, neither men nor women gained [2]. A company that counters by reflex is paying for willingness to bargain, not for the work. And in the Australian survey, nearly a third of counteroffered employees left within a year anyway [1].

"Approval by whoever is in the thread"

What it looks like: There is no named approver and no threshold. An exception is approved when someone senior enough replies "fine" in a chat, says yes in a hallway or forwards the email to payroll. Different exceptions of the same size go to different people.

Why smart teams fall into it: At 60 people, everyone knows everyone and the founder sees every decision. Informal approval works until the company passes the size where one person can hold the whole picture. Nobody notices the moment it stops working.

What it costs: No record, so no explanation later. US equal pay guidance asks whether a factor other than sex, applied consistently, in fact explains a pay difference and whether the stated reason is a pretext [4]. California goes further: A factor used to justify a pay difference must be applied reasonably and must account for the entire difference [5]. An approval with no recorded reason makes both tests hard to meet. In the EU, the directive shifts the burden of proof to the employer once a worker shows facts that suggest discrimination [6]. "Someone approved it" is not a reason.

"It's only temporary"

What it looks like: An exception is approved for a reason that will pass: A hot market, a critical project, a counteroffer during a launch. It goes into base pay. Nobody sets a review date. Three years later the reason is gone and the premium is still there, compounded by three merit cycles.

Why smart teams fall into it: Base pay is the easiest lever to pull and the one people value most. A one-time payment feels less convincing to the person being retained. Agreeing to review it later feels like enough.

What it costs: The exception becomes the band. The next person hired into the role asks for the same pay and points to a colleague who has it. A temporary premium in base pay is permanent in practice. When it pushes pay above the band maximum, the company faces the harder question of what to do with pay above the band, with no record of why it got there.

"The CEO approves everything"

What it looks like: To stop the drift, every exception now goes to the CEO. Requests pile up. The CEO approves almost all of them, because there are no criteria to decline against and no view of the peer group in front of them. Or the opposite happens: The CEO approves quickly for people they know and slowly for people they don't.

Why smart teams fall into it: Central approval feels like control and the CEO does care about payroll. It also avoids an awkward conversation about who else can be trusted with the decision.

What it costs: Speed is lost and consistency is not gained. An approver without information is a signature, not a check. A CEO who decides alone also concentrates the risk: If their judgment is uneven, every exception carries the same pattern. Central approval only helps when the approver sees what the request does to the peer group and has written grounds for saying no.

What sound exception approval looks like

A sound approach does not try to eliminate exceptions. It makes each one a deliberate decision that the company could explain later, to the employee, a colleague or a regulator. Several qualities set it apart.

Everyone knows what counts as an exception. The written rule says plainly which decisions need extra approval: Pay outside the band, raises outside the cycle, increases above the guide, counteroffers, retention payments. If managers have to guess, the rule will be applied unevenly.

The approver is not the person asking. Approval sits at least one step away from the manager making the request, with someone who sees across teams. In most growing companies that means HR or finance alongside the manager's leader. The bigger the deviation, the further from the request the approval sits. Seniority of the role alone is a poor guide.

The approver sees the peer group. Before saying yes, the approver can see what others in the same job and level earn, how the request compares and whether it would put the person above people with more experience. A request reviewed against a single number is not reviewed.

The reason is chosen from a short list and recorded at the time. Scarce skills, a documented competing offer, a role that has grown, a temporary assignment. A short named list makes exceptions comparable. "Strong performer" on its own is not a reason. The record lives where pay data lives, not in a chat thread.

Temporary exceptions have an end. Where the reason will pass, the premium is paid in a form that can end, or carries a review date that someone owns.

The next person test is passed. A useful question for every approver: Would we pay the next person in this role the same, for the same reason? If the answer is no, the decision is a favor, not an exception.

Exceptions are reviewed as a set. Once a year, someone counts them: How many, for which reasons, in which roles, by gender and other groups the company tracks. A rising count in one job family usually means the band is wrong, not the people. A pattern by gender means the process is producing a gap. Either finding is far cheaper to act on early.

A quick check: Pick three exceptions from the past year. If you can name who approved each one, why and whether the reason still applies, the system works. If you can't, it does not.

The calls only you can make

Some of the choices behind exception approval are not process questions. They are statements about how the company pays.

Do you counter at all? Some companies never counter, on the view that pay should be right before anyone resigns. Others counter for a small number of critical roles. Both are defensible. Countering for whoever asks is not, for the reasons above.

How much speed will you trade for consistency? A two-day approval loses some candidates. A same-hour approval loses some consistency. The right balance depends on how competitive your hiring is and how much inversion you can live with.

Who holds the final say? The CEO, the CFO, the head of HR or a small group. The answer should be one that still works when the company is twice its size.

How open will you be about the rule? Telling staff that exceptions exist, why they are granted and who decides can feel risky. It is also what makes the rule credible. Your compensation philosophy is the natural place to say it.

When do exceptions mean the structure is wrong? Decide in advance what share of a role paid outside the band would make you reopen the band itself, rather than approve one more exception.

The research and the law

  • Counteroffers are routine. A recruitment firm's survey of 500 hiring managers in Australia, run in October 2025 and published in June 2026, found 85% of employers had counteroffered in the past year. 46% said the employee stayed, 32% said the employee left within 12 months and 7% said the employee declined and left [1]. These are the firm's own figures for one country.
  • Renegotiation and the gender gap. A study using Swedish register data found better outside options raised men's within-job pay but not women's. The researchers estimate renegotiation accounts for roughly half of the gap between men and women in the same workplace and occupation, about 8% on average. Under collective agreements, neither group gained [2].
  • Gaps start at the offer. A study of more than 700,000 US initial salary offers from 2017 to 2020 found a 5.5% gap between women and men after detailed controls, published in a peer-reviewed journal in May 2025 [3].
  • US federal guidance. EEOC guidance on the Equal Pay Act asks whether a gender-neutral factor, applied consistently, in fact explains a pay difference. It states that prior salary cannot by itself justify a disparity and tells investigators to consider whether the employer bargains differently with men than with women [4].
  • California. An employer relying on a bona fide factor must show it is applied reasonably and accounts for the entire wage difference. Prior salary does not justify any disparity. Wage records must be kept for three years [5].
  • EU. The pay transparency directive requires employers to make the criteria for pay and pay progression easily accessible to workers and to keep them objective and gender-neutral (Article 6). Where a reported gap of 5% or more in a category of workers is not justified by objective, gender-neutral criteria and not remedied within six months, a joint pay assessment follows (Article 10). In a pay discrimination claim, the burden of proof can shift to the employer (Article 18) [6]. Our guide on which pay gaps you can defend covers why a recorded reason is not the same as a defensible one.

This is general information, not legal advice. Equal pay rules differ by state and country and change over time. Check the rules where your people work.

Our view

We think every pay exception needs three things: An approver who is not the person asking and who can see the peer group, a reason from a short written list recorded at the time and an end or review date when the reason will pass. That is enough structure for a company of 100 to 500 people. It does not need a committee.

We would counter rarely and by the same rule for everyone, because a habit of matching outside offers pays for willingness to bargain and, on the evidence, widens gender gaps. We would rather move a band that the market has left behind than approve the tenth exception in the same role.

Most of all, we would count exceptions once a year. A company that knows how many it granted, why and to whom can explain its pay. One that doesn't will find out from the next person who asks.

Sources

  1. Robert Half, "The rise and risk of counteroffer culture", press release on the 2026 Salary Guide (16 June 2026; 500 hiring managers in Australia, surveyed October 2025), checked 2026-10-26
  2. Rockwool Foundation Berlin, "Outside job opportunities raise men's pay, but not women's" (20 July 2026), on Fredriksson, Gülümser and Hensvik, Outside Job Opportunities and the Gender Gap in Pay, RFBerlin Discussion Paper 71/25, checked 2026-10-26
  3. Sterling and Wang, Setting Up the Gap? Gender Differences in Initial Salary Offers in Hiring, Organization Science (19 May 2025), checked 2026-10-26
  4. US EEOC, Compliance Manual Section 10: Compensation Discrimination, 10-IV.F.2 and 10-IV.F.2.g, checked 2026-10-26
  5. California Labor Code § 1197.5, equal pay, checked 2026-10-26
  6. Directive (EU) 2023/970 on pay transparency, Articles 6, 10 and 18 (EUR-Lex), checked 2026-10-25

How should you split the salary budget?

When one budget pays for promotions, market fixes and range moves, merit gets what is left. Five traps and what a sound budget split looks like.

What triggers a joint pay assessment?

A 5% gap in any category of workers can trigger an EU joint pay assessment. Five ways mid-size employers get caught and what a sound position looks like.

One global rate or pay by location?

One rate everywhere and a rate for every city both break. Five ways location-based pay goes wrong and why a few clear zones usually work better.

See Resources for every guide and the self-check.