Diversity, Equity and Inclusion: What Still Holds Up in 2026

Eight years running Talent Acquisition, and the fairest hiring system I ever built was never called a DEI programme.

The word itself has become unstable. To your general counsel it reads as legal exposure. To your CEO it reads as a political liability.

To a large part of your workforce it reads as a promise that was made and then quietly withdrawn. To finance it reads as a line item with no number attached. You are expected to hold all four meanings at once and still give a clean answer on Monday.

Most organisations are treating this as a naming problem. It is a wiring problem.

Programmes are being renamed rather than repaired. The slate rules, the promotion criteria, the pay bands, and the interview panels stay exactly as they were built. The organisation keeps the legal risk of the old design and loses whatever credibility the new label was supposed to buy.

Meanwhile the thing the label was meant to govern is getting worse, and it is measurable.

Translation: fewer than 4 in 10 people strongly agree on the most basic condition of an inclusive workplace. Not sponsorship. Not promotion equity. Respect.

By the end of this, you will have:

• A definition precise enough to survive a legal review and plain enough to say out loud in a leadership meeting
• One allocation rule that resolves most equity versus equality policy questions
• An honest read of the business case, including exactly where it breaks under challenge
• What the March 2025 EEOC guidance changed, what it did not, and which design patterns now carry risk
• A four-engine operating architecture and seven measurement signals that work whether the acronym survives or not

HR Insights Lab approach: every recommendation in this article changes a system rather than a statement. That is a deliberate design constraint, not a defensive one. Systems are auditable, and auditable things are much harder to delete in a budget review.

What is Diversity, Equity and Inclusion (DEI)?

Diversity, equity and inclusion (DEI) is an organisational discipline that combines three distinct things: who is represented in the workforce (diversity), whether people get the specific support they need to compete fairly (equity), and whether their contribution shapes decisions (inclusion). In practice it is a set of design choices inside hiring, pay, promotion and progression systems, not a values statement.

Here’s the difference:

The three terms are sequential, and none of them substitutes for another. Diversity without inclusion produces turnover. You hire people into a system that does not use what they bring, and they leave inside two years with a polite exit interview.

Equity without diversity has nothing to act on. Run one term and skip the other two and you get a programme that reports activity every quarter while the composition of your leadership team stays exactly where it was in 2019.

Most of the work that survives scrutiny happens upstream, in inclusive job description writing and requirement design, long before anyone reviews a shortlist. That is also the work nobody gets credit for, which is why it rarely gets done.

What Does DEI Stand For? The Three Terms, Defined for Practitioners

Every ranking page defines these three words. Almost none of them tells you which HR system each word governs, which is the only thing that makes the definitions usable. Here is that translation.

Diversity: Who is in the room

Diversity is measurable composition across visible and non-visible dimensions: race, ethnicity, gender, age, disability, veteran status, socio-economic background, neurotype, education pathway, and geography. It is a counting exercise. That is not a criticism, it is the reason diversity is the only one of the three you can audit in an afternoon.

Here’s the difference:

Aggregate headcount diversity is a near-useless number. An organisation can be 45% women overall and 8% women at director level. Those are two entirely different organisations reporting the same flattering headline.

The only diversity figure that tells you anything is composition by level and by function, read across the last four promotion cycles. Report it as a pyramid rather than a percentage. The shape shows you where the system leaks, and the leak is almost never at entry level.

That reframe matters because it moves the intervention point. If composition is a sourcing output, the fix lives in how to build a talent pipeline before the requisition opens, not in a conversation with a hiring manager three weeks before an offer.

Equity: Who gets what they need to compete

Equity is the removal of structural disadvantage from a process, so that the process measures capability rather than proximity to the default candidate. It governs process design. It does not govern the allocation of results, and that distinction is now the difference between a defensible programme and an exposed one.

How to execute:

Interview scheduling that does not assume childcare-free evenings. Job requirements written as verified capabilities rather than degree or tenure proxies. Pay bands set by role and market data rather than by prior salary.

Three artefacts, three design decisions, none of which requires knowing anything about the candidate’s protected characteristics. Equity is about the design of the contest, not the assignment of the result.

Filtering on capability signals rather than resume keywords is equity implemented at the sourcing layer. I call this Inverted Sourcing Funnels, and it is covered in full further down. It is also the closest thing to a free lunch that skills-based hiring offers, because it widens the pool and raises the evidence standard at the same time.

Inclusion: Whose contribution counts

Inclusion is whether a person’s input changes outcomes. Not whether they were invited, present, or surveyed.

Here is the operational test. In the last four significant decisions your team made, whose input changed the outcome? If the answer is the same three names every time, you have attendance, not inclusion.

Inclusion is the term most likely to be dismissed as unmeasurable soft stuff by a CFO. It is not unmeasurable. The proxies already exist inside systems you own:

• Meeting airtime distribution across the people in the room
• Who gets assigned the stretch project, and who gets the difficult client
• Promotion velocity by group, measured as median months in level
• Voluntary attrition delta between majority and minority groups in the first 18 months

Exclusion shows up in the attrition line months before it shows up in a survey. Then it gets misdiagnosed as a pay problem, and someone proposes a market adjustment.

It is rarely a pay problem. Retention failures are mostly systems failures: unclear career paths, weak manager capability, role design that fights how the work gets done, and internal mobility that exists on the intranet and nowhere else. That argument is unpacked in why retention is a systems problem, not a pay problem, and it applies here without modification.

Where Belonging Fits: DEI, DEIB and the fourth term

Many organisations now use DEIB, adding belonging as a fourth term. Belonging is a real construct: it describes whether a person believes they can succeed as themselves rather than by assimilating.

In most organisations, though, the B was added to soften the acronym rather than to add a workstream. Here is the decision rule. If adding belonging did not create a new owner, a new metric, or a new budget line, it is a rebrand.

If you do adopt the fourth term, attach it to something you already measure. Employee engagement metrics that predict attrition are the obvious candidate, because they carry a behavioural signal rather than a sentiment score.

Equity vs. Equality: The Distinction that Decides Your Policy Design

Equality gives every person the same input. Equity adjusts the input so the process measures the same thing for everyone. The distinction matters because applying the wrong principle to the wrong decision is the most common source of policy error in this area, and it is close to the line the 2025 enforcement guidance draws.

Scenario	Equality response	Equity response
Return from parental leave	Same objectives and same review timeline as everyone else from day one	Same objectives, with a defined ramp period and a re-onboarding plan so the review measures capability rather than recency
Candidate located outside a metro hub	Identical relocation package offered to every hire	Relocation support, remote-first role design, or a travel pattern chosen so location does not screen out capability
Assessment format for a neurodivergent candidate	Everyone takes the same timed live exercise	Same capability assessed, with format options such as take-home, extended time, or written response

Here’s the deal:

Equality is the correct default for anything that touches reward or outcome. Equity is the correct default for anything that touches access or process. That one rule resolves most of the policy questions sitting in your inbox this quarter.

The rule earns its keep at the interview stage more than anywhere else, because that is where access and assessment collide. The mechanics are in how to design a fair interview process, and they are entirely structural.

What DEI Looks Like in the Workplace (Across the Employee Lifecycle)

The standard answer to this question is a list of initiatives: employee resource groups, unconscious bias training, diverse interview panels. That list describes what a programme buys, not where composition is decided.

Composition is decided at five points in the employee lifecycle, and four of them sit outside the programme entirely. Here is the map.

Sourcing and pipeline

Most representation gaps are created at sourcing, not at selection. By the time anyone reviews a CV, the pool is already narrow, and every downstream intervention is arguing over the shape of a decision that was made weeks earlier.

Here’s where it breaks:

Three mechanisms narrow the pool, and each has a specific fix. Channel concentration means one or two job boards reaching the same demographic repeatedly, and the fix is a channel audit measuring source, stage-by-stage conversion, and Yield Ratio per channel.

Referral homophily means the engine reproduces the workforce you already have, and the fix is a referral design that widens rather than mirrors. Requirement inflation means degree and tenure proxies screening out capable non-traditional candidates, and the fix is rewriting requirements against verified capability.

None of that required a preference. It required knowing which channels were doing the narrowing, which is what a sourcing channel audit is for.

Interviewing and selection

Structure does more for fair selection than any training intervention available, and it does it without a single identity-based criterion entering the process. That is precisely why it survives legal scrutiny while other interventions do not.

Four elements make selection structural rather than conversational:

• A scorecard tied to the capabilities named in the role definition, not to a general impression of fit
• Interviewers scoring independently before the debrief, with scores submitted rather than discussed
• A defined evidence threshold per capability, so “strong” means the same thing to every panellist
• A debrief that opens with the lowest-confidence rating rather than the loudest voice in the room

Here’s what changed everything:

That is the argument to a sceptical COO. Structure is an efficiency intervention that happens to produce fairness, which is a much easier business case than the reverse. The build detail sits in structured interview scorecards.

Pay, Promotion and Progression

Representation gaps compound at promotion, not at hire. Most organisations have never run the analysis that would show it, and the reason is not technical.

Three analyses tell you the truth:

• Pay equity regression controlling for role, level, location, and tenure, reported as the residual gap rather than the raw one
• Promotion velocity by group, measured as median months in level before promotion
• Performance rating distribution by group, read across at least three cycles

Be direct about why this gets skipped. The findings create an obligation and a cost, so in a meaningful number of organisations the analysis is deliberately not commissioned. That is a decision, and it should be named as one.

In the EU, pay transparency obligations remove the choice entirely, which is covered further down. And progression is where the retention story and the composition story turn out to be the same story: people leave over unclear paths and weak internal mobility, and promotion velocity is the metric that exposes both. The channel view of that is in internal mobility as a sourcing channel.

Internal mobility

Internal mobility is the most underused equity mechanism in most organisations. It gives existing employees a real path, it is faster and cheaper than external hiring, and it removes the credential-screening layer that filters out non-traditional candidates on the way in.

The failure mode is predictable. Underrepresented employees leave because the only visible route upward is external, and the organisation then spends more to hire their replacement from outside.

Notice what that policy is. It is an efficiency decision and an equity decision arriving as the same intervention, which is the strongest position any programme design can occupy. The full build is in First Look policy.

Culture, ERGs and Everyday Inclusion

An employee resource group can legitimately do four things: build community, act as a feedback channel, provide input into policy design, and create visible sponsorship pathways. Those are real functions and they are worth funding.

What an ERG cannot do is substitute for a strategy, operate as an unpaid advisory board, or absorb the entire budget line while the systems that produce the outcomes stay untouched.

The Lab Way here is unglamorous: open membership, a written charter with defined decision rights, a budget line, and leadership time recognised in the performance cycle. Cultural celebrations are fine, but call them what they are. They are visibility, not equity, and the distinction matters when someone asks what the programme delivered. Related mechanics live in employee recognition frameworks.

Why DEI Matters: The Business Case, Honestly Assessed

You have probably been handed a correlation study to put in a board deck. You should know what happens when someone in that room has read the critique.

Here’s the deal:

The research linking representation to financial outperformance is real, it has been replicated, and the measured relationship has strengthened across successive reports. In the most cited study in this debate, companies in the top quartile for ethnic representation showed a 39 percent increased likelihood of financial outperformance against those in the bottom quartile.

Source: McKinsey & Company, Diversity Matters Even More: The Case for Holistic Impact (2023)

So here is the argument that does not depend on that study at all. A narrow pipeline is a smaller pipeline. A smaller pipeline produces slower hiring, weaker slates, and more compromise hires, and every one of those shows up in numbers you already own.

I have watched the size of that effect directly. Sourcing a manufacturing systems analyst the conventional way returned 14 qualified resumes. Searching for capability signals instead of credentials returned 62 candidates for the same role, in the same market, in the same week.

That is the case to make to a CFO. It runs on Time-to-Offer, slate quality, and pool size rather than on a contested regression, and it holds up when someone challenges the methodology of a study you did not conduct. The metric set is in TA metrics your CFO cares about.

The perception problem is now a delivery problem

Internal support has moved, and you have to design around that rather than message past it.

Why this matters:

The practitioner reading of that shift is specific. Most of the erosion tracks to programmes that were highly visible and low in substance: mandatory training with no measured effect, public statements with no system change behind them, and initiatives that were perceived as reallocating opportunity rather than widening access.

Your job is not to argue a workforce out of that perception. It is to build a programme whose mechanism is visible and whose fairness is demonstrable to everyone in the building, including the people most sceptical of it.

A mechanism anyone can inspect defends itself. That is also the structure that gets funded, which is the practical half of how to build a business case HR leadership will approve.

The 2025 to 2026 Legal Reality: What Changed and What’s Still Lawful

You have heard “DEI is illegal now” from one stakeholder and “nothing has changed” from another. Both are wrong, and the accurate middle position is the thing you need to be able to repeat.

What the 2025 EEOC guidance says

Most DEI programmes were designed in a period when the operating question was how to increase representation. That was the wrong question to have optimised for, and the March 2025 technical assistance makes the reason explicit.

Here’s what changed:

The central principle is that Title VII’s protections apply equally to all workers. The EEOC states plainly that there is no such thing as “reverse” discrimination, there is only discrimination. An employment decision based on a protected characteristic is unlawful even where the intent was to increase diversity.

Source: U.S. Equal Employment Opportunity Commission, What You Should Know About DEI-Related Discrimination at Work (technical assistance issued jointly with DOJ, 19 March 2025)

So the design question has changed. It is no longer “how do we increase representation.” It is “how do we widen access without making decisions on protected characteristics.”

Running Talent Acquisition across a multinational means holding two divergent regulatory pictures at once, which is covered further down. Keeping the primary sources on file is not optional in that setup, and the wider tracking discipline sits in employment law changes HR teams should be tracking.

Practices now carrying elevated risk

The instinct when guidance shifts is to pause the whole programme while legal reviews it. That costs eighteen months of momentum and fixes nothing, because the exposure sits in four design patterns rather than in the programme as a whole.

Four patterns, four structural corrections, and the rest of the programme keeps running. Where to point counsel and what to document is the sort of thing a standing HR policy review checklist should already cover.

What remains clearly lawful (and effective)

The assumption most practitioners are working under is that the choice is between running an exposed programme and doing nothing. There is a third option, and nobody has laid it out plainly.

These practices widen access without making decisions on protected characteristics:

• Skills-based role definition and capability-signal sourcing
• Structured interviews with scored criteria and defined evidence thresholds
• Broadened sourcing channels and proactive outreach into pools you do not currently reach
• Pay-band discipline, pay equity analysis, and funded remediation
• Open-membership employee resource groups with a charter and a budget
• Mentorship and sponsorship open to all, marketed deliberately
• Transparent, published promotion criteria

Why this works:

Capability-signal sourcing is what found me a former factory-floor supervisor who had taught himself Python and now leads a digital transformation function. No preference, no quota, no identity criterion anywhere in the process. The full story is a few sections down, and the implementation detail is in skills-based hiring implementation.

Why DEI Programmes Fail: Three Anti-Patterns

Competitors list challenges in the abstract: unconscious bias, lack of buy-in, resistance to change. None of those is diagnosable, and none of them tells you what to stop doing on Monday. These three are.

Anti-pattern 1: Training as strategy

Here is the mechanism behind the failure. Training moves stated attitudes measurably and durable behaviour barely, because the decision environment is unchanged when the trainee walks back into it.

The same scorecard. The same slate. The same promotion criteria and the same calibration meeting, run by the same people, in the same order. You have asked an individual to out-think a system that was built to produce a particular answer.

I apply the same logic to retention that I apply here, and for the same reason. The problem is almost always the operating model rather than the individual, and I have watched this particular pattern repeat across every organisation I have worked inside. The wider version of it is in why HR programmes fail at implementation.

Anti-pattern 2: Scaling the referral programme without fixing its physics

Referrals are usually the highest-quality, lowest-cost channel you own. They are also the single most powerful mechanism for reproducing the composition you already have, because people refer people like themselves.

Here’s the catch nobody names:

The standard response to a cost-per-hire problem is to raise the referral bonus. That works, and it accelerates the narrowing at exactly the moment you are being asked to widen the pipeline. Both things happen at once and only one of them gets reported.

Keep the engine and change its physics. Activate referrers from under-represented groups specifically, ask for referrals by capability rather than by job title, and track referral yield by source group so the narrowing is visible before it becomes a five-year composition problem. The build detail is in employee referral programme design.

Anti-pattern 3: The ownership vacuum

The programme is owned by a committee, funded from a discretionary line, and measured by an annual survey. Fourteen people attend the quarterly meeting and none of them has decision rights over any system that produces the outcomes they are judged on.

The correction is four specific things:

• A named owner at leadership level, not a committee
• Decision rights over three artefacts: the role definition template, the interview scorecard, and the promotion criteria
• A protected budget line rather than a discretionary one
• A quarterly metric review in the same forum as every other operational metric

Three artefacts is a deliberately small number. It is the smallest set that reaches hiring, pay, and progression at once, and it is small enough that one person can genuinely own it. The partnering model that makes this work is in HR business partnering frameworks.

The Lab Way: Building DEI as an Operating System, not a statement

Everything above converges here. Four engines, each one label-independent by design, each one already justifiable on efficiency grounds alone.

Here’s how to build it:

Engine 1: Market Intelligence

“There just aren’t any diverse candidates in this field” is the most common objection you will hear, and in most organisations it goes unchallenged because nobody has the data to refute it. Composition problems get diagnosed as hiring problems when they are market-knowledge problems.

If you do not know what the available pool for a skill looks like across geographies, industries, and non-obvious source sectors, you cannot know whether your slate is narrow or the market is.

How to execute:

Two frameworks do the work here. Intelligence-Led Sourcing means continuous market mapping before vacancies arise, across a defined competitor set and the adjacent industries that hold the same skills under different job titles.

Talent Market Pre-Alignment (TMPA) goes one step further. You engage identified talent ahead of demand through targeted challenges, competitions, and pooled communities, so capability, interest, and availability are pre-aligned before the requisition opens.

What gets mapped is specific: which competitors, which adjacent industries, which skills, which geographies, and how global supply compares to domestic supply for the same capability. Hiring patterns across the competitor set tell you where teams are being built and where they are being cut.

The quality effect is measurable too. Competition analysis lifted Quality of Hire by 12% in my own numbers, talent landscape work by 14%, and skill-based pipelines built on market analysis by 30%. The method is in talent mapping.

Engine 2: Structural Design

Conventional sourcing filters on proxies: degree, employer brand, job title, years in seat. Those proxies correlate with access to opportunity far more reliably than they correlate with capability.

Inverted Sourcing Funnels filter on capability signals instead. Project evidence, code repositories, portfolios, publications, certifications, and demonstrated problem-solving. The pool widens on exactly the dimensions the proxies were suppressing, and no identity-based criterion enters the process at any point.

This is what equity looks like when it is implemented as a search technique rather than as a policy statement. The role definition side of it is in how to write a skills-based job description.

Engine 3: Decision Discipline

Two interviewers assess the same candidate and reach opposite conclusions. Without a mechanism, you have no way to establish which one was right and no defensible record if the decision is ever challenged.

The decision layer needs five things: scorecards tied to the capability decomposition from Engine 2, independent scoring before any group discussion, evidence thresholds rather than impressions, calibration across interviewers, and a documented rationale for every hire and every promotion.

That last one is the piece most teams skip and the piece that matters most when a decision is questioned two years later.

Why this works:

Structure makes decisions both fairer and defensible at the same time, using one mechanism. You are not adding a fairness layer on top of a hiring process. You are building a hiring process that produces fairness as its default output.

Then there is AI, which now sits inside this engine whether you planned for it or not. Screening tools have become sourcing co-pilots that run semantic search across your own database, agentic systems are managing entire interview scheduling lifecycles rather than single steps, and skills-first assessment tooling has genuinely reduced some sources of unconscious bias. Predictive models are also being used to flag retention risk before it appears in a resignation.

“Use AI to remove bias from hiring” is a claim I would not make in a leadership meeting, and given training-data inheritance it is frequently backwards. A more honest read of where the tooling helps is in AI in recruitment, what works.

Engine 4: The Accountability Loop

The first three engines produce nothing durable without this one. A defined owner, a quarterly metric review in the standard operating forum, published criteria, and an internal-first design that gives existing employees the first look at every opportunity.

The forum choice is the part people underestimate. A metric reviewed in its own dedicated meeting is a metric nobody outside that meeting is accountable for.

• One named owner, senior enough to change the three artefacts
• Seven signals reviewed quarterly, next to Time-to-Fill and cost per hire
• Criteria published where candidates and employees can read them
• Internal-first posting as the default, not the exception

The First Look Policy is the cleanest example of an internal-first design with a measurable outcome attached. 23% of hires became internal moves, and Time-to-Productivity ran at 28 days against 67 for external hires, which is a number the operating forum understands without translation. Building the review itself is covered in HR metrics dashboard.

If you want the next framework before it goes out publicly, the Lab newsletter carries field-tested HR frameworks, real metrics, and AI-driven hiring strategies weekly. The specific signals to put in that quarterly review come next.

How to Measure DEI: Seven Signals That Move

You get asked how DEI is going and you can answer with a headcount percentage and a survey score. Neither tells anyone what to do next, which is why the question keeps getting asked.

Here’s what to put in the quarterly review:

1. Diverse slate ratio

The share of shortlists meeting a defined breadth standard. Calculate it as slates meeting the standard divided by total slates, per quarter, per function. A bad reading points upstream to sourcing rather than to the recruiter, because a slate can only be as broad as the pool it was drawn from.

2. Stage-by-stage conversion by group

Conversion rate at every funnel stage, cut by group. This is the single most diagnostic metric available to you, because it shows exactly which stage is doing the filtering. If conversion is even until the technical round and collapses there, you have found your problem in one afternoon.

3. Offer acceptance parity

Offer acceptance rate by group, against your overall rate. A gap here is not a sourcing problem, it is a candidate-experience or compensation-equity problem, and those have completely different fixes. My own high-water mark for overall acceptance is 90%, which is the benchmark I hold a funnel against.

4. Promotion velocity by group

Median months in level before promotion, by group and by function. This is where compounding failure hides, because a six-month median difference looks trivial in one cycle and produces an entirely different leadership team over eight years. Almost nobody gathers it.

5. Eighteen-month voluntary attrition delta

The difference in voluntary attrition between majority and minority groups in the first 18 months. This is the earliest reliable inclusion signal you have, and it leads the engagement survey by months. For context on what good looks like, my best 12-month cohort retention figure is 89%.

6. Internal mobility rate by group

Internal moves as a share of total hires, cut by group. It tells you whether the internal path is real or nominal, and a gap here usually explains an attrition number you have been struggling to account for. People leave when the only visible route upward is external.

7. Pay equity gap after controls

The residual pay gap after controlling for role, level, location, and tenure. Report the residual, not the raw gap, because the raw gap invites a methodology argument that goes nowhere. The residual is the number that requires an explanation or a remediation budget.

One compliance point before you build the dashboard. Collecting and processing demographic data is jurisdiction-dependent: it requires a lawful basis in the EU and UK, and in several jurisdictions certain categories cannot be collected at all. Confirm what you are permitted to hold before you design the cut. The underlying metric definitions are in yield ratio and TA metrics explained.

DEI Beyond The US: What Multinational Employers get Wrong

If you work for a multinational, you are being handed US-centric guidance that is actively wrong for a large share of your workforce. I have run Talent Acquisition across multiple jurisdictions, and the operational reality is that the regulatory picture is moving in two directions at once.

Here’s the difference:

The US direction in 2025 and 2026 has moved toward restricting identity-conscious practice. Several other major jurisdictions have moved the opposite way, toward mandated transparency and reporting.

The EU pay transparency directive is phasing in obligations through 2026. India carries statutory disability employment obligations under the Rights of Persons with Disabilities Act. The UK requires gender pay gap reporting for employers above a headcount threshold.

So a single global DEI policy cannot work. What works is a common measurement layer, the seven signals, running everywhere, with jurisdiction-specific action layers built on top of it.

The governance structure that holds this together sits in global HR governance.

The Language Shift: DEI, DEIB, Inclusion and Belonging, Talent Equity

Many organisations have moved away from the DEI label toward inclusion, belonging, talent equity, or merit-and-opportunity framing. The scale of that shift is measurable in public disclosure.

Mentions of DEI in S&P 500 annual 10-K filings fell from an average of 12.5 in 2022 to 4 in 2024.

Source: SHRM, As DEI Fades from Filings, Smart Leaders Rethink the Strategy (2025)

Disclosure changing is not the same as practice changing. Some of those organisations rebuilt the underlying systems and stopped talking about it. Others changed nothing but the intranet page.

Here’s the deal:

The practitioner test takes one line. Compare the programme’s owner, budget, and metrics before and after the rename. If all three are unchanged, it was a communications decision.

Which resolves the question in the title. The parts that hold up are the structural ones, and they hold up under any name: capability-based role definition, structured selection, broadened sourcing, pay discipline, internal-first mobility, and seven signals in the operating review. The four engines were built to be label-independent, and that is the strongest argument this article makes. How to communicate that distinction internally is covered in internal communications for HR change programmes.

Frequently Asked Questions

What does DEI stand for?

DEI stands for diversity, equity and inclusion. Diversity is who is represented across the workforce and across levels. Equity is whether people get the specific support they need to compete fairly in a process. Inclusion is whether their contribution influences decisions. Some organisations add belonging, giving DEIB.

What is the difference between equity and equality?

Equality gives every person the same input; equity adjusts the input so the process measures the same thing for everyone. The practical rule: equity governs access and process design, equality governs reward and outcome. Applying the wrong one is the most common source of policy error in this area.

What are examples of DEI in the workplace?

Practical examples include skills-based job descriptions that remove unnecessary degree requirements and structured interview scorecards with independent scoring. Others: pay bands set by market data rather than salary history, internal-first job posting windows, and open-membership employee resource groups with a defined charter and budget.

Is DEI still legal in 2026?

Yes, but the design constraints changed. March 2025 EEOC and DOJ technical assistance restated that Title VII applies equally to all workers, meaning employment decisions based on protected characteristics are unlawful even when intended to increase diversity. Practices that widen access without identity-based decisions remain lawful. Take employment-counsel advice on your specific programme.

Why are Companies Cutting or Renaming DEI Programmes?

Most are responding to a mix of legal risk, political pressure, and programmes that never produced measurable outcomes. Disclosure has fallen sharply, with DEI mentions in S&P 500 annual filings dropping from an average of 12.5 in 2022 to 4 in 2024. Renaming a programme does not change what it does.

What To Do on Monday

The organisations that still have a functioning inclusion capability in five years will not be the ones with the strongest statement. They will be the ones whose systems were built to work without one.

What separates them is unglamorous and specific. A named owner. Three artefacts that owner can change, and seven signals reviewed in the same forum as Time-to-Fill.

Everything else is positioning. Public disclosure of this work has fallen sharply, as the filings data shows, and the only question that matters now is whether the underlying systems moved with it. In most organisations they did not, which is precisely the gap you are standing in.

So pick two. One from Engine 2, one from Engine 4. Rewrite one role definition against verified capabilities, and put one signal into next quarter’s operating review. Implement them over the next 90 days, then measure the delta before you touch anything else.

If you want the next framework before it goes out publicly, the Lab newsletter carries field-tested HR frameworks, real metrics, and AI-driven hiring strategies weekly.

And if you are redesigning this inside your own organisation, particularly across multiple jurisdictions, I would like to hear how you are handling the measurement layer. Let’s connect on LinkedIn.