Most Equality Impact Assessments fail before the assessment proper begins. Not because the aims are wrong or the evidence is thin — those are problems too, and they are covered in the companion resource on what every EqIA must answer. They fail because three structural questions that sit upstream of the EqIA were never asked, or were asked vaguely and never recorded. When the assessment is later scrutinised — by a tribunal, a regulator, a complainant, a journalist — the absence of those three answers is what makes the whole document collapsible. It was never a defensible record. It was form-filling.
The three questions are derived from the reasoning in LS v NHS England (ET Leeds, Case No 1802318/2024), the first reported Employment Tribunal decision applying the For Women Scotland ruling to an employer’s trans-inclusive facilities policy. The employer lost on indirect sex discrimination. The tribunal identified three specific failures, each of which maps to a gate that should have been closed before the EqIA proper was written. Each gate is answered specifically and on the record — not by implication, not by assumption, and not after the fact.
Gate 1: Who did you consult — and was it everyone materially affected?
The question is not “did you consult?” but “which materially-engaged groups did you consult, and where is that recorded?” The Equality Act 2010 requires public bodies to have due regard to eliminating discrimination and advancing equality of opportunity under the Public Sector Equality Duty — and that duty runs across every protected characteristic the policy engages, not just the one the policy is about. A trans-inclusion facilities policy is not a policy about trans staff alone. It engages sex, gender reassignment, religion or belief, and potentially disability. Consultation has to reach each of those groups, and the record has to show it did.
Why this is upstream. Consultation shapes the evidence base the EqIA then analyses. If you only consulted the group the policy is designed to support, you have a one-sided evidence base before you start. The assessment built on it is not wrong in its reasoning — it is incomplete in its inputs, and that incompleteness is itself the procedural flaw.
What a defensible answer looks like. A table listing each materially-engaged group, whether they were consulted before the policy was issued, where the consultation is documented, and what they said that changed the draft. “Women’s network: yes, minutes dated X, led to revised signage schedule” is a defensible answer. “We spoke to people” is not.
What an indefensible answer looks like. “We consulted the trans staff network.” That is consultation with one group — the group the policy benefits. The women’s network, the faith network, the disability network: none consulted, none on the record. In the LS v NHS England judgment, the tribunal treated that gap as the failure that made the policy indirectly discriminatory. The policy did not need to be hostile to be unlawful. It needed to be unconsulted.
Gate 2: Did you consider the Workplace (Health, Safety and Welfare) Regulations 1992 reg 20?
If the policy touches sanitary, changing, or single-sex facilities — and any trans-inclusion facilities policy does — the Workplace (Health, Safety and Welfare) Regulations 1992 reg 20 obligation must be considered on the record. Regulation 20 requires employers to provide suitable and sufficient sanitary conveniences with separate provision for men and women, unless each convenience is in a separate room with a lockable door. That is a health and safety obligation, not an equality one — but it intersects directly with any policy that changes who can access which facilities.
Why this is upstream. An EqIA that analyses equality impact without acknowledging the underlying statutory facilities regime is reasoning in a vacuum. The reg 20 obligation is the legal floor the policy operates on top of. If the EqIA never mentions it, the assessment looks like it was written without understanding the regulatory context — and a tribunal or regulator will notice.
What a defensible answer looks like. A named position on whether reg 20 applies to the facilities in scope, what it requires, and how the policy operates within it. Where the position is complex — for example, where the policy changes access to shared facilities and the reg 20 interaction is contested — the EqIA records that the question was identified and refers it to specialist legal advice. The point is that the question was asked, not that it was answered by the EqIA itself.
What an indefensible answer looks like. Silence. The facilities policy is the subject of the EqIA, and reg 20 — the primary statutory obligation governing those facilities — is nowhere in the document. In the LS v NHS England judgment, the tribunal identified the employer’s failure to consider reg 20 as a specific failure contributing to the proportionality finding. The omission was not a minor drafting gap. It was a structural one.
Gate 3: What lesser alternatives did you consider — and why did you reject them?
Any policy that restricts access, participation, or provision on the basis of a protected characteristic has to clear a proportionality test. That test, drawn from Bank Mellat v HM Treasury (No.2) [2013] UKSC 39, asks whether a less intrusive measure could achieve the same aim. The EqIA needs to show this was actually tested — not assumed away. The preflight version of that test is the lesser-alternative gate: name at least three alternatives you considered, say whether you adopted or rejected each one, and record the reason.
Why this is upstream. If the EqIA reaches the proportionality section and the lesser-alternative analysis is being written for the first time, it is too late. The alternatives need to have been weighed before the policy was fixed, because the reasoning has to reflect a genuine choice — not a post-hoc justification for a decision already made. Recording the alternatives at the preflight stage is what makes the later proportionality reasoning credible.
What a defensible answer looks like. A table with three or more named alternatives, each marked adopted or rejected, with a one-line reason on the record. “Gender-neutral facilities on every floor: rejected — cost and listed-building constraints. Single-occupancy lockable facilities in two locations: adopted. Status quo mixed provision: rejected — does not address dignity concerns raised in consultation.” That is a defensible record. The numbers matter too: where a small protected group can be accommodated by a lesser measure, a blanket policy affecting the larger group is harder to justify.
What an indefensible answer looks like. No alternatives considered. The policy as drafted is the only option the decision-maker looked at, and the EqIA offers no evidence that anything else was weighed. In the LS v NHS England judgment, the tribunal noted that directing trans staff to gender-neutral facilities was an available lesser measure — one the employer never considered. With a very small number of trans staff across a large workforce, a blanket policy was disproportionate when a targeted measure was available. The absence of the lesser-alternative analysis was not a gap in the reasoning. It was the failure.
Putting the three gates together
Each gate is independent, but together they form the structural foundation the rest of the EqIA stands on. If consultation was one-sided, the evidence base is incomplete. If reg 20 was never considered, the facilities analysis is ungrounded. If no lesser alternatives were weighed, the proportionality reasoning is a justification written backwards. Any one of those failures can make the policy indefensible regardless of what the rest of the EqIA says — because the tribunal, regulator, or complainant will read the document from the top, and the top is where the gaps are.
The good news is that all three gates are closable before the assessment proper begins. They do not require the full EqIA to be written first. They require a structured conversation with the decision-maker, a table for each gate, and the discipline to record what was said. That is the difference between a form-filling EqIA and a defensible one — not more pages, but earlier questions.
Take this further
The EqIA/DPIA Wizard walks each preflight gate interactively, surfacing gaps before you draft the full assessment. For policies that touch single-sex facilities or carry litigation exposure, the full EqIA Companion delivers forensic-depth analysis with practitioner sign-off — including the reg 20 reasoning where the facilities question requires specialist legal input.
For the four questions that sit inside the EqIA proper — aims, evidence, proportionality, and review — see the companion resource on what every trans-inclusion policy EqIA must answer. This resource covers the gates upstream. That one covers the assessment that follows. A defensible EqIA needs both.