Diversity Targets: Has Anyone Done the Maths?

Britain’s largest employers have become enthusiastic about diversity targets. Boards announce them, HR departments promote them, and progress is reported annually. But there is a question that appears to receive remarkably little attention.

Has anyone calculated whether the targets are actually achievable?

As an actuary, I have spent much of my career modelling populations, their movements and the financial consequences. The same mathematics applies to corporate workforces. People join, leave, retire and get promoted. Their numbers can be projected, and the probability of achieving a particular outcome can be estimated.

Yet much of the diversity debate seems curiously detached from these elementary considerations.

Consider Deloitte UK.

According to its published 2025 figures, 31% of its partners were women. Its target is 37% by 2030. Meanwhile, women accounted for 38.8% of directors, the obvious internal recruitment pool for future partners.

At first glance, the target looks modest. Just six percentage points over five years. Surely achievable?

The arithmetic is rather less accommodating.

Imagine a partnership containing 1,000 people, 310 of them women. Assume its size remains constant, with 10% of partners departing annually and being replaced.

Over five years, approximately 41% of the original partnership would be replaced.

If 39% of incoming partners were women, matching roughly the current director population, female representation would rise from 31% to approximately 34.3%.

Not 37%.

To reach 37% over the same period, approximately 46% of incoming partners would need to be women.

That is seven percentage points above their representation in the director population.

The calculation is illustrative, not a forecast of Deloitte’s actual results. External recruitment, changing promotion pools, differential departure rates and partnership growth could all alter the outcome. Indeed, Deloitte reports that women now account for approximately 32% of partners, and it has made substantial progress over the past decade.

But the underlying question remains: what combination of recruitment, promotions and departures is required to deliver the target?

And how probable is that combination?

Suppose the organisation wants to accelerate progress further. There are only a limited number of mathematical levers available.

It can increase the proportion of women entering the relevant recruitment pool. It can promote a higher proportion of eligible women. It can alter retention patterns, or it can change the size and structure of the organisation.

Each has consequences. Some take many years. Others may create pressure on selection decisions.

This is where an apparently benign aspiration can become problematic.

The Equality Act 2010 permits certain forms of positive action, including limited tie-break provisions in recruitment and promotion. It does not generally permit an employer to select a less-qualified candidate simply to meet a numerical diversity target.

A target is not itself unlawful. Nor is encouraging applications from underrepresented groups. But the more demanding the target, and the shorter the deadline, the greater the temptation to interfere with selection outcomes.

Mathematics cannot establish whether discrimination has occurred. It can, however, reveal the pressure that a target creates.

There is another difficulty: probability.

Suppose a company has ten vacancies, with women constituting 40% of the genuinely eligible candidate pool. Even with completely unbiased selection, the resulting appointments will fluctuate.

Four women might be appointed. Or three, five, six, perhaps none.

Such variation is not necessarily evidence of bias. It is a normal feature of small populations.

Yet corporate diversity programmes frequently treat differences between actual and desired outcomes as problems demanding managerial intervention.

The danger is that randomness gets mistaken for discrimination, while genuine discrimination can be concealed by favourable aggregate numbers.

Neither is satisfactory.

My research into what I call Diversity Maths examines these problems using actuarial population models, probability distributions and workforce projections.

One particularly revealing finding is that organisations can become trapped in what I call “Stubborn States”: demographic compositions that change only slowly, even when recruitment and promotion are conducted fairly.

An organisation with a heavily male senior workforce and relatively few departures cannot rapidly transform its leadership merely by wishing to do so.

Conversely, a company with high turnover and a substantial female recruitment pool may achieve a seemingly ambitious target without any preferential selection.

The distinction matters enormously.

I am not arguing against diversity, or against investigating genuine underrepresentation. Nor am I suggesting that Deloitte’s target is necessarily unreasonable.

I am arguing for something considerably simpler: before announcing a target, calculate the probability of achieving it under transparent, lawful and realistic assumptions.

Boards should ask their finance directors three questions.

Is the target mathematically feasible?

What recruitment, promotion and departure assumptions are needed to achieve it?

And what happens if those assumptions prove wrong?

We would not tolerate a corporate financial forecast that ignored the arithmetic of revenues, costs and cash flows.

Why should workforce diversity targets receive less scrutiny?

Diversity policy needs fewer slogans and more mathematics.

Peter Crowley is an actuary and founder of Crowley Analytics, which develops mathematical models of workforce diversity, recruitment and pay gaps. Further information: diversitymaths.com.