Diversity Review Guide - Turn Insights into Action

Diversity reviews are only useful when they change decisions, not when they produce a neat chart and a vague promise. In a UK public-sector setting, I think the real job is to show where representation, progression, service access, or day-to-day experience is uneven, and then explain what to do about it. This guide walks through what the review should cover, which metrics matter, how to collect data safely, and how leaders can turn findings into action.

What a diversity review should answer

When I look at a diversity review, I start with a simple test: does it help a leader make a better decision this quarter? If the answer is no, the exercise is probably too broad, too shallow, or too detached from reality. In public services, the point is not just to count people; it is to understand where the organisation is creating unequal outcomes. The UK public sector also has a legal and practical reason to do this well. The Public Sector Equality Duty asks public bodies to think about how their policies, services, and decisions affect people with different protected characteristics, and to monitor the impact of what they do. That means a credible review should tell you three things: who is represented, who is progressing, and who is experiencing barriers.

I find it helpful to frame the work around three questions:

• Who gets through the door, and through which routes?
• Who stays, develops, and moves into stronger roles?
• Who benefits from services, and who still finds them hard to access or navigate?

That framing keeps the work practical. It also stops the review from becoming a tidy HR exercise that ignores service delivery, which is where many public-sector organisations miss the real story. Once the scope is clear, the next step is deciding what to measure.

What to measure if you want a useful picture

The strongest reviews combine workforce data, service-user evidence, and qualitative feedback. I would not begin with dozens of indicators; I would begin with a small set that exposes where the system leaks. Standardising the categories matters too, because consistent data is what lets you benchmark over time or compare one service area with another.

Metric	What it shows	Common trap
Representation by grade or level	Whether people from different groups are present at entry, middle management, and senior leadership	Looking only at overall headcount and missing the bottleneck higher up
Recruitment funnel data	Where applicants are filtered out, from application to shortlist to offer	Assuming a weak final result means the whole process was fair
Pay and progression	Whether people are moving through pay bands and promotion routes at the same pace	Confusing equal pay with equal opportunity
Retention and exits	Whether certain groups leave faster, transfer less, or stagnate longer	Ignoring exit reasons and treating turnover as normal churn
Inclusion and service experience	How safe, respected, and well served people feel, whether they are staff or service users	Assuming silence means satisfaction

For most public bodies, the relevant reference point is the nine protected characteristics under the Equality Act 2010, but I would not stop there. Contract type, location, shift pattern, and manager behaviour often explain patterns that the headline categories alone cannot. That is especially important when the numbers are small and easy to misread. With the right measures in place, the more difficult task becomes gathering the data without damaging trust.

How to run the review without losing trust

Data collection is where good intentions often go wrong. If people do not trust the process, they either do not respond or they give answers they think are safe. In practice, I try to make the review feel useful and safe at the same time.

1. Define the decision you want to improve. I would not start with a survey form; I would start with the problem, such as low representation at senior grades or weak service access for a particular group.
2. Separate monitoring from selection. Equality forms should stay apart from application forms, CVs, and interview decisions so the information cannot influence hiring or promotion.
3. Use voluntary and anonymous channels where needed. If confidentiality cannot be guaranteed, a short anonymous staff survey is usually better than forcing a formal form that people do not trust.
4. Collect both numbers and comments. Monitoring data tells you where the gap is; focus groups, open-text comments, and exit interviews tell you why it exists.
5. Protect sensitive data properly. If you are handling special category data, document the lawful basis and the Article 9 condition, and carry out a DPIA where the processing is likely to be high risk.
6. Read the data by group and by intersection. A team can look fine on ethnicity and disability separately while still disadvantaging disabled women or older minority staff.

That process sounds procedural, but it is what protects the credibility of the review. Once people believe the data is handled carefully, they are more likely to answer honestly and more likely to accept the findings when the results are uncomfortable. From there, the real value comes from reading the patterns correctly.

How to read the patterns behind the numbers

Raw data rarely tells the whole story. I have seen organisations celebrate strong diversity at entry level while ignoring the fact that the same groups disappear before they reach management. I have also seen service teams assume that a group is well represented because it appears in overall usage figures, only to discover that access problems are hidden in one channel, one location, or one step in the journey.

When representation looks fine but progression does not

This is the classic bottleneck problem. You may have decent recruitment into junior roles, yet the same people are not moving into stretch assignments, acting-up opportunities, or promotion pools. That usually points to sponsorship, informal networks, appraisal bias, or a narrow definition of “ready”.

When service users are present but not well served

Public-sector organisations sometimes measure who turns up and miss who gets what they need. A group can be represented in service data and still face longer waits, more refusals, or worse outcomes. That is why service design, accessibility, and feedback loops belong in the review, not as an afterthought.

Why intersectionality changes the reading

Intersectionality simply means people do not experience work or services through one identity at a time. A review that treats race, disability, sex, or age as separate silos can miss the people who are most affected. I would always test whether the same barrier shows up across multiple groups or whether it concentrates in a specific intersection.

Reading the data this way turns a descriptive report into a diagnostic one. The next risk is not under-reading the data, but doing the wrong thing with it.

Common mistakes that make the exercise pointless

The fastest way to waste a review is to treat it as compliance theatre. The second fastest is to collect data that nobody intends to use. I see the same mistakes again and again, and they are avoidable.

• Measuring only what is easy to count instead of what actually shapes outcomes.
• Using the review to produce a report, but not a decision.
• Letting managers see monitoring forms that should stay separate from selection decisions.
• Assuming low response rates mean there is no problem.
• Ignoring small sample sizes and over-claiming from weak data.
• Failing to connect workforce patterns with service-user experience.
• Reporting gaps without naming an owner, deadline, or next step.

The common thread is simple: the organisation wants reassurance, not insight. Good practice does the opposite. It surfaces uncertainty, then asks what evidence would reduce it. That is why the final stage matters so much: leaders have to convert findings into action.

How leaders turn findings into action

I usually recommend a three-horizon response. It keeps the work grounded and stops leaders from trying to solve everything at once. A useful plan has quick fixes, structural changes, and a review point for accountability.

Horizon	What to do	What good looks like
First 90 days	Fix obvious barriers, brief managers, and tighten data collection	People can see immediate changes and understand why they matter
3 to 6 months	Review promotion criteria, recruitment wording, induction, and manager capability	The process becomes fairer, not just the messaging around it
6 to 12 months	Track whether the gap is shrinking, then adjust policy or leadership behaviour	The organisation can prove progress instead of promising it

Where the evidence shows a persistent disadvantage, I would also consider whether targeted outreach or positive action is appropriate, but only with proper legal and HR review. In public-sector organisations, credibility matters as much as ambition. People will forgive a slow improvement plan more readily than a flashy one that never changes the numbers.

What strong practice looks like once the review is done

The strongest diversity reviews are the ones that become part of how the organisation thinks, not a one-off event. I look for four signs: the findings are discussed by senior leaders, the actions have owners and dates, the same metrics are tracked again, and staff can see what changed because of the review.

• The review combines workforce, service, and culture data instead of relying on a single dashboard.
• The organisation uses consistent categories so trends can be compared over time.
• Managers are held accountable for the actions, not just the report.
• People can tell that the process led to something concrete, such as a fairer recruitment step or a more accessible service route.

That is the point of the work: not to prove the organisation is perfect, but to make unequal outcomes harder to ignore and easier to fix. If the review is honest, well governed, and followed by action, it becomes one of the most practical leadership tools a public body can use.