Step 1: Separate the two questions — representation and inclusion

The single most common measurement mistake is treating "diversity" as one number. It is really two different questions that need two different kinds of data. Representation asks how many people from each group you employ — a headcount question. Inclusion asks whether those people feel they belong, are treated fairly and can speak up — an experience question. An organisation can look impressively diverse on paper while different groups quietly have very different day-to-day experiences.

You need both, because each is meaningless without the other. Representation with no inclusion data tells you who is in the building but nothing about whether it is a good place for them to be. Inclusion scores with no demographic breakdown tell you the average mood but hide the groups whose experience is far worse than that average. Measuring both, and cross-referencing them, is what turns a diversity dashboard from a vanity exercise into something you can act on.

Step 2: Measure representation against the Equality Act 2010

Start with the quantitative baseline: the proportion of your workforce in each demographic group. Base these categories on the nine protected characteristics set out in the Equality Act 2010 — age, disability, gender reassignment, marriage and civil partnership, pregnancy and maternity, race, religion or belief, sex, and sexual orientation. For ethnicity, use the Office for National Statistics categories so your numbers are comparable with national and sector benchmarks rather than a bespoke list nobody else can read.

The metric people forget to track here is the disclosure rate — the share of employees actually willing to answer each demographic question rather than selecting "prefer not to say". Disclosure is consistently lowest for disability, sexual orientation and, increasingly, socio- economic background — precisely the characteristics where the data is most valuable. A representation figure built on a 40% disclosure rate is not a measurement; it is a guess. Raising disclosure is mostly a function of trust, which is why genuine anonymity is a measurement issue, not just an ethical one.

Step 3: Measure inclusion and belonging — not just headcount

Representation is the easy half. The harder, more revealing half is measuring how the organisation actually feels to work in for each group. We capture this through experience questions we call Lenses: belonging, fairness, psychological safety, whether people feel able to be themselves at work. These are the metrics that predict retention, engagement and whether your diversity survives contact with reality.

Measure inclusion with a consistent, well-designed scale so scores are comparable between groups and over time — the choice of agreement scale genuinely affects the result, as does whether each question is worded inclusively. Pair every experience question with the demographic data from Step 2, because an inclusion score only becomes useful the moment you can split it by group. Our guide to measuring inclusion in complex organisations goes deeper on how to structure this when teams span multiple locations and reporting lines.

Sign-up to measure diversity in your workplace...

Step 4: Collect the data anonymously — or the numbers will be wrong

This is the step that determines whether everything above is accurate. People answer honestly about disability, ethnicity or sexual orientation only when they are certain the answer cannot be traced back to them. When they are not certain, disclosure collapses to 20–30% and your smaller groups vanish into the "prefer not to say" column — so anonymity is not a nicety, it is the thing that makes the measurement valid.

Real anonymity is an engineering property, not a promise in an email. On the Divrsity platform there are no cookies, no IP logging and no link between an answer and an email address: each response is tied only to a random 128-bit identifier, the participant's email is purged the moment they finish, and not even Divrsity staff can see how an individual answered. Because people trust it, disclosure rises and short, well-communicated surveys routinely reach 70–80% participation instead of the 20–30% a generic form achieves — the difference between data you can stand behind and data you cannot.

Step 5: Analyse by group and intersection — not headline averages

Headline averages hide the very inequities you are measuring diversity to find. The insight lives in the gaps between groups — and, crucially, at their intersections, where for example the experience of younger women differs from that of women overall. A company-wide "belonging" score of 72% is comfortable and useless; the same score split by ethnicity, disability and age is where the real story appears.

Our most uncomfortable finding makes the case. In one organisation we asked whether employees agreed that racial slurs were tolerated. As a single average the result looked unremarkable. Split by ethnicity it was stark: the white community strongly disagreed that slurs were tolerated, while the non-white community strongly agreed they were — two groups describing the same workplace and living in two different realities. No headline number would ever have surfaced that. On Divrsity this segmentation is automatic: an interactive dashboard, filterable by team, location and demographic with intersectional breakdowns, is generated within about 10 minutes of the survey closing.

Step 6: Benchmark against UK sector and regional norms

A number on its own rarely tells you whether to worry. Is 61% ethnic-minority representation in your early-careers intake good or bad? It depends entirely on your sector and region. Benchmarking your representation and inclusion metrics against comparable UK organisations turns a raw figure into a judgement you can act on — and stops you either panicking over a healthy number or celebrating a poor one. Our note on regional and industry benchmarking explains why local comparison matters far more than a global average.

Step 7: Measure on a cadence and turn the metrics into action

A single measurement is a snapshot; the value is in the trend. Measure on a regular cadence — once or twice a year — using the same questions each time so the numbers are genuinely comparable, then track whether the gaps you found are closing. This is how you move from measuring diversity to benchmarking your own progress and prove, with data, that an action plan worked rather than hoping it did.

Measurement only earns its keep when it changes something. Pick two or three specific commitments off the back of the data, deliver them, and measure again — and share what you found, including the uncomfortable parts. A prioritised, AI-generated action plan follows every Divrsity survey within 24 hours, so the step from "here are the numbers" to "here is what we are doing about them" does not stall in a spreadsheet.

Measure diversity with an anonymous, automated survey

Which diversity metrics actually matter

If you strip the exercise back to the numbers worth reporting to a board, five metrics carry most of the weight:

• Representation by protected characteristic — the proportion of your workforce in each group, overall and by level, mapped to the Equality Act 2010.
• Disclosure rate — the share of people willing to answer each demographic question. A low disclosure rate invalidates everything below it, so track it first.
• Inclusion / belonging score — the experience metric, always reported split by group rather than as a single company-wide figure.
• Retention and progression by group — whether people from each group stay and get promoted at the same rate. Diversity that leaves within a year is not really diversity.
• Pay gaps across every characteristic — not only the gender pay gap you may be required to publish, but ethnicity, disability and beyond, which is where Divrsity reports gaps out of the box.

Why measuring diversity matters in the UK right now

Measurement is no longer optional or purely reputational. McKinsey's 2023 research found companies in the top quartile for ethnic diversity on their executive teams were significantly more likely to outperform their peers financially — but you cannot manage, or prove, what you do not measure. The UK regulatory backdrop is tightening in the same direction: mandatory gender pay gap reporting for employers with 250 or more staff, ethnicity pay gap reporting moving up the agenda, confirmed disability pay gap reporting, and equality action plans that require an evidence-based analysis of why a gap exists. Every one of those obligations rests on demographic data you can only get by measuring diversity properly — and anonymously — in the first place. The Equality Act 2010 gives you the framework; a well-run measurement programme gives you the evidence.

Frequently Asked Questions

How do you measure diversity in the workplace?

Measure diversity on two axes. First, representation: how many people from each group you employ, measured against the nine protected characteristics in the Equality Act 2010 and reported using Office for National Statistics ethnicity categories. Second, inclusion: how fairly those groups experience the organisation — belonging, fairness and psychological safety. Collect both through an anonymous survey and analyse by group, not headline average.

What metrics should you use to measure diversity?

The core metrics are representation by protected characteristic, disclosure rate (the share of people willing to answer each demographic question), an inclusion or belonging score, retention and progression rates broken down by group, and pay gaps across every characteristic — not just gender. Representation tells you who is in the room; the experience metrics tell you whether being in the room feels the same for everyone.

What is the difference between measuring diversity and measuring inclusion?

Diversity is a headcount question — how many people from each group are represented. Inclusion is an experience question — whether those people feel they belong, are treated fairly and can speak up. An organisation can look diverse on paper while different groups have very different day-to-day experiences, so measuring only representation misses half the picture. You need both numbers to know where to act.

Why are anonymous surveys important for measuring diversity?

Because disclosure rates collapse when people fear being identified. If employees suspect an answer about disability, sexual orientation or ethnicity can be traced back to them, many select "prefer not to say" and your data becomes unreliable — especially for the smaller groups you most need to hear from. Genuine, technical anonymity is what makes the numbers accurate in the first place.

How often should you measure diversity in the workplace?

Measure on a regular cadence — typically once or twice a year — using the same questions each time so the numbers are comparable. A one-off measurement is a snapshot; the value is in the trend. Tracking the same representation and inclusion metrics over time is the only way to know whether an action plan is actually working rather than guessing.

How many people do you need to measure diversity meaningfully?

There is little value in formally measuring diversity below about 20 people. Beyond the headline, a breakdown for any single group is only meaningful — and only safely anonymous — where that group contains at least five respondents. Below that threshold a good platform automatically suppresses the figure so no individual can be identified. On Divrsity this suppression floor defaults to five and is configurable.

Conclusion: measure both halves, or you are guessing

Measuring diversity well comes down to refusing to settle for a single number. Measure representation against the Equality Act 2010, measure inclusion as a genuine experience score, collect both anonymously so the data is honest, and read every result by group and intersection rather than by average. Do that on a regular cadence and benchmark it, and you move from a diversity statement on the careers page to evidence you can act on and defend.

Divrsity was built to make each of those steps the default: an Equality Act-aligned question set we keep current, obsessive UK-based anonymity that lifts disclosure rates, and automated analysis that turns a closed survey into an interactive, segmented dashboard within minutes rather than a week of spreadsheet work. If you want to know exactly how the survey itself runs end to end, our companion guide on how to run a DEI survey walks through it step by step.

Sign-up to measure diversity in your workplace

Related Articles See All Blog Articles

• Ready to measure? See how to run a DEI survey end to end, and the latest set of DEI survey questions
• Since accurate measurement depends on honest answers, learn how Divrsity is obsessed with anonymity and how to maximise participation using applied psychology
• Go deeper on measuring inclusion in matrix organisations and the science of agreement scales
• Turn numbers into context with regional and industry benchmarking and benchmarking your own progress over time
• Understand the reporting drivers: ethnicity and disability pay gap reporting, and equality action plans