Why £15,000 Is the Wrong Number for Quantum Advice

Every week, practitioners advise clients on the likely value of their employment tribunal claim using a benchmark that is wrong for the majority of cases.

The average compensation award across 129,000+ employment tribunal decisions is £15,000. That figure appears in legal guides, HR risk assessments, and settlement negotiations. It is the number people reach for when they need a starting point.

The median is £2,284.

Half of all claimants who win at employment tribunal take home less than £2,300. That is not a rounding error or a data quality problem. It is a structural feature of how compensation is distributed across claim types, and it has real consequences for how clients are advised.

The Distribution Problem

The gap between £15,000 and £2,284 exists because a relatively small number of very large awards pull the mean upward. A handful of £500,000 disability discrimination claims or £1 million sex discrimination outliers skew the average significantly. The median is immune to those outliers. It reports what the person in the middle actually received.

| Statistic | Value | |---|---| | Mean (average) compensation | £15,000 | | Median compensation | £2,284 | | 25th percentile | £802 | | 75th percentile | £8,400 | | 90th percentile | £31,200 |

The 25th percentile is £802. A quarter of all successful claimants receive under £800.

This distribution is not uniform across claim types or sectors. When you filter by discrimination ground, the numbers shift substantially. When you filter by sector, they shift again. The £15,000 mean is a blended average across wildly different types of case, which makes it almost useless as a quantum benchmark for any specific case.

What Kahneman and Tversky Explain

Tversky and Kahneman's foundational work on cognitive biases, specifically the availability heuristic, explains why this problem persists in legal practice. High-value awards are memorable and widely reported. A £500,000 discrimination settlement makes the legal press. The thousands of £800 injury to feelings awards do not. When a practitioner tries to recall "what do these cases settle for," the figure that comes to mind is shaped by what is easy to recall, not by the underlying distribution.

The result is that quantum advice built on the mean is systematically optimistic for the majority of clients. A client told "average payouts are £15,000" has a fundamentally different expectation from one who knows that most successful claimants receive under £2,300. That expectation gap creates problems: unrealistic settlement demands, failed negotiations, disappointed clients when the outcome lands closer to the median than the mean.

The Practical Fix

The fix is not complicated, but it requires filtering the data rather than relying on a single headline figure.

Useful quantum advice requires:

1. Filter by claim type. Unfair dismissal cases, discrimination cases, and wage claims have very different compensation profiles. The unfair dismissal cap sits at £115,115 (2024/25). Discrimination claims have no cap. Mixing these populations inflates the mean.

2. Filter by sector. Public sector employers, particularly local authorities, face the Public Sector Equality Duty under s.149 of the Equality Act 2010. Tribunals tend to award higher injury to feelings where the employer had a statutory obligation to have done better. NHS discrimination cases average around £24,000; local authority cases average £52,000 in the TribDB dataset.

3. Filter by discrimination ground. Race, sex, disability, belief, and age discrimination cases have different distributions. Belief discrimination (driven by gender-critical cases post-Forstater) currently shows a wide spread with a mean around £15,000 and a high 90th percentile, reflecting the unsettled state of the law.

4. Look at the full distribution, not a summary statistic. Median, interquartile range, and 90th percentile tell you far more than a mean. For settlement advice, knowing that 75% of comparable cases settled under £8,400 is actionable. Knowing the mean is £15,000 is not.

Why This Is Hard to Do Manually

The reason practitioners default to the mean is not ignorance. It is that building a properly filtered dataset from scratch, across 129,000 tribunal decisions, is not feasible within the time available for any individual case.

Most practitioners know the general shape of the market. They may have a rough sense that their area of work tends to settle between X and Y. But that intuition is built from the cases they personally handled, not from the full population. And it is subject to exactly the availability bias Kahneman and Tversky identified: the memorable, high-value cases colour the estimate.

The only route to accurate quantum advice is to look at the distribution across a large, filtered dataset of comparable cases. That is not a particularly sophisticated statistical operation. It just requires the data to be searchable.

What Good Quantum Analysis Looks Like

For a mid-career employee bringing a disability discrimination claim against an NHS trust, the relevant figures are not the mean across all employment tribunal decisions. They are:

• Median disability discrimination award in the public sector
• Distribution of injury to feelings awards under the Vento guidelines for comparable cases
• 25th and 75th percentile for that specific claim type and sector combination
• Any recent trend data (compensation in this area has been rising year-on-year since 2020)

None of that requires expert statistical work. It requires the ability to filter 129,000 decisions by claim type, sector, and discrimination ground simultaneously, and then read the distribution rather than the headline.

Search employment tribunal decisions filtered by claim type, sector, and discrimination ground on TribDB. Free 14-day trial, no card needed.

Data source: 129,000+ decisions from GOV.UK and the National Archives (February 2017 to present), plus 15,000+ fitness to practise hearings from 6 UK healthcare regulators (HCPC, NMC, MPTS, GDC, GPhC, GOC). Updated weekly.

Reference: Tversky, A., & Kahneman, D. (1974). "Judgment under Uncertainty: Heuristics and Biases." Science, 185(4157), 1124-1131.