Redundancy selection criteria decide who stays and who leaves when roles are cut. They are also the first part of a redundancy exercise a tribunal takes apart. A genuine redundancy with careless selection is still an unfair dismissal. 

The standard is not perfection. It is a fair pool, objective criteria, scores backed by evidence, and a record that proves all three. Most UK SMEs can meet that standard. Most fail on the record, not the intention. 

This article covers the pool, the criteria and the scoring matrix. It covers the evidence behind each score. It also covers the discrimination traps hidden inside common criteria. The consultation meetings that run alongside selection are a separate discipline. Our guide to individual redundancy consultation covers those in full. This guide covers England, Scotland and Wales. Northern Ireland has its own redundancy legislation, separate discrimination law and different pay caps. 

Get the redundancy pool right before you score anyone 

Most SME redundancy exercises start scoring too early. The redundancy pool comes first, and it is where more selections fail than most employers expect. 

The pool is the group of employees whose roles are at risk and who will be scored against each other. ACAS guidance on selecting employees for redundancy is clear on the shape of it. Include all roles that are the same or similar. Consider adding roles that use similar skills, even if the job titles differ. Check any agreement with a recognised trade union, and any existing redundancy policy, before you settle the pool. 

A tribunal will not redraw your pool for you. It asks a narrower question: did the employer genuinely apply its mind to who should be in it? Case law, including Capita Hartshead v Byard, shows how that question bites. An employer who never considered alternatives to a pool of one loses on the pool alone, before scoring is even discussed. 

That points to the cheapest insurance in the whole exercise. A one-page pool memo, written before scoring starts, does the job. It records which pools were considered, which were rejected and why, and who sits in the chosen pool. Ten minutes of writing, and the "did you apply your mind" question now has a documented answer. 

One warning. A pool drawn around a person, rather than around roles, is the fastest route to a finding of pre-determination. If the pool happens to contain exactly the one employee the business wanted to exit, expect a tribunal to notice. 

Bumping and the pool of one 

Two pool questions come up in almost every SME exercise. 

The first is bumping. Bumping means moving an at-risk employee into a colleague's role, and making the colleague redundant instead. No rule requires an employer to bump. Tribunals do expect the option to have been considered. That expectation is strongest where the at-risk employee has long service and transferable skills. The obligation is modest: consider it, record the outcome in the pool memo, and move on. 

The second is the pool of one. Sometimes a role genuinely stands alone, and a pool of one is the honest answer. It is still the most challenged pool shape there is. The test does not change. Were alternatives genuinely considered? A pool of one picked for convenience looks like a decision about a person. A pool of one backed by a memo naming the wider pools considered, and why each was rejected, looks like a decision about a role. 

What fair redundancy selection criteria look like 

The principles have been settled since Williams v Compair Maxam in 1982. Fair redundancy selection criteria are objective, capable of being checked against evidence, and applied consistently across the pool. Criteria that rest on one manager's opinion fail all three tests at once. 

ACAS puts it plainly: criteria should be measurable, based on facts, and not affected by personal opinion. Common redundancy selection criteria examples that meet the standard include: 

1. Performance against the objectives of the role, drawn from appraisal records. 
2. Skills, qualifications and expertise, drawn from training and certification records. 
3. Disciplinary record, counting only live warnings. 
4. Attendance record, adjusted as covered in the next section. 

Vague criteria are where defensibility leaks away. "Attitude", "flexibility" and "commitment" are opinions wearing suits. If a criterion cannot be defined in a sentence and evidenced from a document, it does not belong in the matrix. Criteria should also be shared with affected employees during consultation, before scoring, so challenges arrive while they can still be answered. 

How many criteria is enough? Four to six is the working range. Fewer than three looks arbitrary. More than eight dilutes the signal and multiplies the evidence burden. Every criterion added is another set of documents to produce for every person in the pool. 

The discrimination traps hidden in common criteria 

Selection is where discrimination claims in redundancy are usually born. The criteria look neutral. The effect is not. Selecting an employee because of any of the nine protected characteristics under the Equality Act 2010 is unlawful. The ACAS guidance also warns against criteria that discriminate indirectly. Five traps catch employers most often. 

Unadjusted attendance scoring. Attendance is a legitimate criterion. It only works after absence related to disability, pregnancy or maternity has been removed from the numbers. Scoring a disabled employee down for disability-related absence converts a redundancy exercise into a discrimination claim. 

Flexible working as a marker. Scoring down employees who work flexibly, or part time, risks indirect sex discrimination. Women still carry more of the caring responsibility, so the criterion lands harder on them. ACAS uses exactly this example. Part-time status is separately protected in its own right. 

Length of service on its own. Last in, first out feels safe because it is objective. Used alone, it risks indirect age discrimination against younger staff. As one weighted criterion among several, it is easier to defend. As the whole method, it is likely to be age discrimination (GOV.UK). 

Salary cost as a proxy. Selecting the most expensive people in the pool can operate as a proxy for age. If cost pressure is driving the exercise, deal with it through the pool and the business case, not through the scores. 

Automatically protected reasons. Some selection reasons are automatically unfair, whatever the scores say. They include trade union membership or activity, acting as an employee representative, fixed-term status, and raising whistleblowing or pay concerns. Pregnant employees, and employees on or recently returned from maternity, adoption or shared parental leave, carry extra protection. From the point pregnancy is notified until 18 months after the birth or placement, any suitable alternative vacancy must be offered to them first, ahead of other employees at risk. 

Two more facts sharpen the point. Discrimination compensation is uncapped. And discrimination and automatically unfair claims need no minimum service, so short-serving employees are not the safe scores some employers assume. Our guide to unfair dismissal risk under two years' service explains that gap. That gap is closing. From 1 January 2027 the Employment Rights Act 2025 cuts the ordinary unfair dismissal qualifying period from two years to six months, so redundancy dismissals taking effect after that date face far wider unfair dismissal exposure. 

Building a redundancy scoring matrix that holds up 

A redundancy scoring matrix is a table. Criteria run down one side. Each carries a defined score range and a weighting that reflects what the business needs going forward. ACAS publishes a worked example with weighted scores, and the structure below follows the same logic. 

The weighting is a business decision, and it is yours to make. A firm that needs to keep scarce technical skills can weight skills at three times attendance. What matters is discipline. The weighting is set before scoring starts. It applies to everyone in the pool. Each weight should be explainable in one sentence. Different pools can carry different criteria, provided each pool is scored on one consistent matrix. 

The arithmetic should be visible on the sheet. A 4 on performance at weight x3 contributes 12 points. A 2 on attendance at weight x1 contributes 2. Totals rank the pool from top to bottom, and the lowest totals are provisionally selected. Show the working on every score sheet. A total with no visible working invites the suspicion that the total came first. 

Now the part that separates a defensible matrix from a decorative one. Define what each number means, in writing, before anyone is scored. "4 means the employee exceeded objectives in both of the last two appraisals" is a defensible band. A bare 1-to-5 scale with no definitions is opinion with decimal points, and cross-examination will treat it as exactly that. 

Three process rules complete the matrix. Two scorers score independently, then moderate their differences. Any manager who is inside the pool never scores it. Where independent scores differ by more than one band, the moderation discussion is noted. The note records the reason for the final figure. 

Scoring evidence: the part tribunals actually test 

British Aerospace v Green settled how tribunals treat scores. They do not re-mark the exercise or substitute their own numbers. They test whether the system was fair and honestly applied. That moves the battleground from the marks to the evidence behind them. 

The working rule is short: no document, no score. Every figure on the sheet should trace back to a record that existed before the exercise began. Appraisals, absence data, training logs, live warnings. A score a manager cannot connect to a document becomes an opinion the moment it is challenged. Opinions score poorly at tribunal. 

This is where thin HR records catch up with smaller employers. If appraisals have not happened for two years, the performance criterion has no fuel. There are honest fixes: widen the evidence window, use measurable work records instead, or reduce the weight on performance. There is one dishonest fix, which is writing retrospective appraisals to justify scores already decided. Tribunals recognise reverse-engineered evidence, usually from dates and from scores that bunch suspiciously around one individual. 

Keep everything. Each scorer's original sheets, the moderated final sheet, and notes of the moderation meeting. Where a score changed, the note says why. In a claim, original scoring sheets are evidence. A single tidy final version is assertion. 

Sharing scores and handling challenges 

Provisionally selected employees should see their own scores during consultation. They should also see the criteria, the band definitions, and where they sit against the lowest retained score. They should not see colleagues' names or individual scores. Explaining relative position does not require breaching anyone's data. 

Expect challenge, and treat it as part of the process rather than an attack on it. An employee may point out that a score missed a qualification, or counted disability-related absence. That employee is doing the review a tribunal would otherwise do later, at far lower cost. Where a challenge lands, correct the score, re-run the ranking, and record both. A corrected score demonstrates a genuine process. A process that never changes anything demonstrates the opposite. 

Assume every document will be read twice. Once by the employee, through a data subject access request. Once by a tribunal, in disclosure. Scoring sheets, moderation notes and emails about the exercise are all disclosable. Each one should be written as if the person scored will read it, because they may. 

Offer an appeal against selection, heard where possible by someone not involved in the original scoring. The cost of skipping these steps compounds quickly once a claim is issued. Our guide to unfair dismissal compensation awards sets out what the downside looks like in numbers. 

Where selection sits in the wider redundancy process 

Selection is one link in a statutory chain. The chain has trip points either side of it. 

If you propose 20 or more redundancies at one establishment within 90 days, collective consultation rules apply on top of everything above. For 20 to 99 proposed redundancies, consultation runs at least 30 days before the first dismissal. For 100 or more, it is 45 days. The Redundancy Payments Service must be notified on form HR1 before consultation starts, and failing to notify carries an unlimited fine. The thresholds are on GOV.UK's collective consultation guidance. Our article on collective redundancy and the 45-day rule walks through the mechanics. 

After selection comes money and notice. Employees with two or more years' service qualify for statutory redundancy pay. The rate is half a week's pay per full year worked under age 22. It is one week per year from age 22 to 40, and one and a half weeks per year from 41. Service is capped at 20 years. For redundancies from 6 April 2026, a week's pay is capped at £751. The maximum statutory payment is £22,530. Both figures are on GOV.UK's redundancy pay guidance. Our free redundancy calculator produces the figure for each employee in minutes. Notice periods and notice pay carry their own calculation rules, covered in our guide to redundancy notice and notice pay. 

None of that machinery rescues a bad selection. Get the pool, the criteria and the evidence right, and the rest of the process has something solid to stand on. 

Frequently asked questions

What are fair redundancy selection criteria?

Fair criteria are objective, capable of being checked against evidence and applied consistently across the pool, as settled in Williams v Compair Maxam. ACAS says they should be measurable and based on facts. Examples include performance against role objectives, skills, live disciplinary warnings and adjusted attendance. Vague terms such as attitude do not belong.

How does a redundancy scoring matrix work?

A scoring matrix is a table with criteria down one side, each carrying a defined score range and a weighting. A 4 on performance at weight x3 contributes 12 points. Totals rank the pool and the lowest are provisionally selected. Define what each number means in writing before anyone is scored, and show the working.

Can I use attendance as a redundancy selection criterion?

Yes, but only after absence related to disability, pregnancy or maternity has been removed from the numbers. Scoring a disabled employee down for disability-related absence turns a redundancy exercise into a discrimination claim. Attendance is a legitimate criterion once adjusted. Discrimination compensation is uncapped, and no minimum service is needed to bring such a claim.

How do I choose a redundancy pool?

The pool is the group of employees whose roles are at risk and who are scored against each other. ACAS guidance says to include all roles that are the same or similar, and to consider roles using similar skills even if titles differ. Check any union agreement and redundancy policy first, then record the reasoning in a one-page pool memo.

Can an employee challenge their redundancy selection score?

Yes, and you should expect it. Provisionally selected employees should see their own scores, the criteria and the band definitions during consultation. Where a challenge shows a score missed a qualification or counted disability-related absence, correct it, re-run the ranking and record both. Offer an appeal against selection, heard where possible by someone not involved in the scoring.

What records should I keep for redundancy selection scoring?

Keep the pool memo, each scorer's original sheets, the moderated final sheet and notes of the moderation meeting, with the reason for any changed score. Every score should trace back to a record that existed before the exercise began, such as appraisals, absence data, training logs and live warnings. No document, no score.

Find the gaps before a redundancy exercise finds them 

Selection exposes the state of your underlying HR records. Appraisals, absence data, training logs and warnings either exist when you need to score against them, or they do not. 

The free HR Health Check takes under a minute. It lets you know where your HR processes stand before a redundancy exercise tests them. Start it at HR Health Check.