The strongest DEI work is measurable, local, and tied to service quality
- In public bodies, DEI should sit inside policy, procurement, recruitment, and service delivery, not outside them.
- Structured hiring, clear promotion criteria, and easy access to adjustments usually move outcomes faster than awareness campaigns alone.
- Leaders need named accountability, regular review, and consequences for poor behaviour, not just visible support.
- Useful measurement combines workforce data, service-user data, and lived experience, because numbers alone miss too much.
- Inclusive design means testing decisions with the people affected early enough to change the outcome, not after complaints arrive.
Why UK public sector DEI needs a different approach
I usually start with the legal frame, because in UK public-sector settings this work is usually discussed as equality, diversity and inclusion, or EDI, and it is not optional branding. The EHRC is clear that the Public Sector Equality Duty is a legal requirement, and GOV.UK guidance treats it as part of good decision-making: public bodies must consider how policies, programmes, services, recruitment, and procurement affect people with protected characteristics.
The Equality Act 2010 protects nine characteristics, so public bodies need to think beyond a single audience. The duty has three practical aims: eliminate unlawful discrimination, advance equality of opportunity, and foster good relations. In plain English, that means you do not just ask whether a decision is lawful; you ask who benefits, who is left behind, and what effect the choice will have on trust and access. The guidance also expects a proportionate approach, so a major policy change deserves a deeper equality assessment than, say, buying office furniture.
- Eliminate harm by removing discrimination, harassment, and victimisation.
- Open up opportunity by reducing barriers for underrepresented groups.
- Check the wider effect on relationships, confidence, and participation in public life.
Once that frame is in place, the next question is where the everyday barriers actually sit inside the employee journey.
Build fairness into recruitment and progression
I would not start with another generic training course. The fastest gains usually come from fixing the points where decisions are made: job design, shortlisting, interviews, and promotion. When those steps are structured, people are judged more consistently and managers have less room to rely on instinct.
| Stage | Weak practice | Stronger practice | Why it helps |
|---|---|---|---|
| Job design | Long lists of essential criteria and jargon-heavy adverts | Plain-English role profiles with only the skills that really matter | It widens the pool and reduces unnecessary self-selection barriers |
| Shortlisting | Unstructured reading of CVs | Scored criteria and calibrated review by more than one person | It makes bias easier to spot and challenge |
| Interviews | Loose conversations that vary by panel | Consistent questions and work-sample tasks | It improves fairness and usually predicts performance better |
| Progression | Opaque promotion routes and informal sponsorship | Visible criteria, regular talent reviews, and recorded decisions on stretch work | It reduces the “we just know who is ready” problem |
I also think reasonable adjustments should be treated as normal operating practice, not a special favour. Reasonable adjustments are changes to process, timing, format, or equipment that remove barriers for disabled people or those with access needs, and if a candidate or employee has to negotiate for access every time, the system is already creating friction. The same logic applies to internal mobility: if people cannot see how to move forward, they stop believing the organisation is fair.
That is why the practical work is usually less glamorous than people expect and much more effective.
Make leaders accountable for outcomes, not slogans
DEI fails quickly when it becomes something HR owns. In public organisations, leaders set the tone through the targets they choose, the behaviour they tolerate, and the trade-offs they approve. Government project guidance is blunt on this point: senior owners and managers carry specific accountability, and everyone else follows the standard they set.
I look for four things in a serious leadership model: measurable objectives, regular review, visible behaviour standards, and consequences when those standards are ignored. Psychological safety matters here too; it simply means people can question a process, raise a concern, or ask for an adjustment without fear of retaliation.
- Set a small number of meaningful objectives. One or two outcomes are easier to manage than a long list of aspirations.
- Put them into performance conversations. If inclusion never appears in appraisal or delivery reviews, it will drift.
- Equip line managers. Most inclusion problems are experienced locally, so managers need scripts, escalation routes, and examples.
- Act on behaviour early. Microaggressions, exclusion from meetings, and biased allocation of opportunities rarely fix themselves.
- Reward the right things. Promotion should not go only to the loudest or most visible people if that pattern undermines trust.
The main mistake I see is confusing visible support with actual accountability. A leader who posts a statement but never checks promotion data, adjustment delays, or team climate is signalling that the work is optional. Once leaders own the outcomes, the next step is to prove whether those outcomes are changing.
Measure what changes behaviour and service quality
If you do not measure it, you usually end up managing anecdotes. The trick is to track the metrics that connect directly to decisions, not vanity indicators such as training completion alone. In England and cross-border bodies, the specific duties include equality objectives at least every four years, annual gender pay gap data, and annual information on compliance; Scotland and Wales have their own equivalents. The exact reporting regime differs across the UK, but the principle is the same: review, publish, improve.
| What to track | What it tells you | Common trap |
|---|---|---|
| Representation by grade | Whether access improves or stalls higher up the organisation | Looking only at overall headcount |
| Applicant, shortlist, and offer rates | Where the hiring funnel narrows unfairly | Assuming the final offer tells the whole story |
| Promotion velocity and pay gap data | Whether progression is equitable over time | Ignoring promotion and reward once people are hired |
| Retention and exit feedback | Why people leave and whether certain groups leave sooner | Focusing only on recruitment |
| Engagement and belonging scores | How safe and valued people feel day to day | Using one annual survey without following up |
| Service outcomes by group | Whether users experience the same quality of access and response | Measuring internal culture but not public impact |
That mix matters because inclusion is partly cultural and partly operational. If the numbers look stable but staff still report avoidance, silence, or slow adjustments, the culture has not caught up with the dashboard. The real test is whether data changes a decision, a process, or a deadline.
When the evidence stack is honest, the conversation moves from opinion to action, and that is where service design becomes the obvious next step.
Design policies and services around real users
This is where public-sector DEI becomes visible to citizens, not just employees. A policy can look fair on paper and still fail badly if the service is hard to reach, the language is opaque, or the only channel assumes digital confidence and free time. I always ask who gets excluded by the default design, because the default is rarely neutral.
- Involve the people affected early. Service users, frontline staff, and community voices will surface problems that internal teams miss.
- Test the whole journey. Forms, letters, appointments, call handling, website content, and follow-up messages all affect access.
- Design for accessibility from the start. Alternative formats, interpreters, reasonable timeframes, and clear language should be built in, not bolted on.
- Use equality impact work as a live tool. It should shape options, not merely justify a decision after it has already been made.
- Carry inclusion into procurement. Suppliers influence delivery, so contracts and service standards should reflect the same expectations.
The best public organisations I have seen do not wait for complaints to reveal a weak process. They test early, publish clearly, and keep checking who still cannot get through. That leaves one final question: what should you stop doing before you start the next wave of work?
The reset I would start with over the next 90 days
If I had to reset a programme quickly, I would start by cutting the habits that waste time and credibility.
Common mistakes
- Launching training before fixing the process that creates the bias.
- Collecting demographic data without a clear decision attached to it.
- Leaving staff networks to carry the work without authority, time, or budget.
- Measuring diversity at entry level and ignoring progression, pay, and retention.
- Talking about inclusion in communications while managers still run inconsistent practices.
- Using one generic plan for workforce culture and public service delivery.
Read Also: Diversity Review Guide - Turn Insights into Action
A 90-day reset
- Pick one workforce gap and one service gap that matter most to the people you serve.
- Map the current numbers, the lived experience, and the policy or process behind the problem.
- Change one high-friction process end to end, such as shortlisting, a booking flow, or a complaints route.
- Assign a named owner, a deadline, and a review rhythm that leaders actually attend.
- Publish progress in plain language and keep the next objective specific enough to judge.
That is usually enough to move DEI from aspiration to management discipline. In the UK public sector, the organisations that make lasting progress are the ones that connect fairness to delivery, measure the right things, and stay honest about what still needs work.
