On-site massage as tier-2 provision: the brief
For a professional firm the exposure here is not an HSE inspection of a factory floor. It is a tribunal, a civil stress claim, or a regulator, and what each tests is whether you took reasonable steps to manage a foreseeable risk. The provision most firms rely on to clear that bar has quietly stopped clearing it.
The cost is already on your books
Work-related stress costs UK employers around £51bn a year, of which roughly £24bn is presenteeism, the cost you absorb without it ever showing as a sick day. The per-employee figure is highest in finance and insurance at £5,179, with professional services at £2,357. In 2024/25 the HSE recorded 22.1 million working days lost to work-related stress, anxiety or depression, and 964,000 affected workers. The bill lands on the employer regardless of where the stress originated.
Figure A
What the sessions cost, against the payroll they sit in
Annual figures for a 40-person firm at about £32,000 of payroll a head (wages plus employer National Insurance and pension). The programme is roughly 0.2 to 0.5% of total payroll. It sets one cost the firm already carries against another, and makes no claim about what the sessions achieve.
The reasonable-steps position most firms rely on has failed
For two decades firms leaned on Sutherland v Hatton (2002): offer a confidential counselling service with referral, and you are unlikely to be found in breach. The Court of Appeal qualified that within five years. Daw v Intel (2007) held counselling is "not a panacea"; Dickins v O2 (2008) warned it is "dangerous to apply guidance given by the court as though it were a statutory provision." The EAP trade body still cites the original 2002 reading as the foundation of its market.
The HSE issued the University of Birmingham a notice of contravention. Birmingham presented sickness absence below the UK average, EAP usage data, and occupational-health referrals. The HSE did not dispute the figures and found the breach anyway: the institution "cannot demonstrate sufficient understanding of the risk to your staff from work-related stress."
That is a deliberative-quality test, not an outcome test. Good numbers are not the defence. A documented understanding of the risk is. (The first notice of this kind was served on the East of England Ambulance Service in April 2025.)
Read together: an EAP on its own is no longer a safe place to stand.
The EAP misses exactly the people who matter
Provision is not uptake. Counselling requires disclosure, and the disclosure becomes a record that outlasts the call: special-category data, reachable in a family court or an employment tribunal. The more career a person has to protect, the less they use it. Peer-reviewed work finds surgeons and senior leaders use these services at materially lower rates than the average workforce. Around one in ten UK workers holds a role where a mental-health record is a regulatory matter rather than a private one (FCA-certified, SRA-regulated, GMC, DVLA Group 2 and others). The PM-commissioned Stevenson and Farmer review found only 11% of employees had discussed a recent mental-health problem with their line manager, and 8 in 10 employers report no disclosures at all. Provision that depends on someone raising their hand cannot reach the person who will not.
Why massage specifically
NICE Guideline NG212 sets a three-tier architecture for workplace mental health, and is explicit (§1.6.1) that the tiers are additive: individual-level provision sits on top of the organisational foundations, it does not replace them. An EAP discharges tier 1. Tier 2 is usually empty.
Massage is the one tier-2 option that asks nothing of the disclosure-averse. It requires no disclosure, the recipient is passive rather than performing a practice, attendance signals nothing to colleagues, it fits a meeting room, and it creates no mental-health record. No crystals, chakras, bells or gongs, just a therapist and a table. It is not the only tier-2 option and is not a substitute for the organisational work; it is the option that reaches the population the EAP structurally cannot. The evidence is real and worth stating plainly: independent meta-analysis of 137 trials puts adult effect sizes for pain, stress and anxiety in the medium-to-large band, in the same class as CBT. Transfer to a self-selecting workplace population is directional rather than a precise figure.
Figure B
Where massage sits: the tier most firms leave empty
NICE NG212 makes the tiers additive, not substitutes. An EAP discharges tier 1; tier 2 is where massage sits, and where most firms have nothing.
What it costs, and the trail it leaves
A 12-week pilot runs £3,000 to £4,800 and reaches 30 to 50 staff, below the threshold at which most procurement engages. Against the per-employee stress cost already being absorbed, that is single-digit percentages. A single resignation in a senior or specialist role costs £40,000 to £100,000 to put right.
The pilot also produces the artefact the deliberative-quality test asks for: an anonymised booking log, a quarterly aggregate uptake report, and a written provision policy. That trail exists at 5% uptake or at 50%. Even where take-up is modest, the firm holds documented evidence that it identified a foreseeable barrier and provisioned for it. That document sits in the reasonable-steps file before any inspection, not in response to one.
The headline savings rest on an effect-size transfer the full model flags as directional, not proven ROI. The durable case is the compliance one, which does not depend on uptake. And this is a tier-2 supplement, not a replacement for the stress risk assessment, workload monitoring and manager training NG212 puts at tier 1.
Informed commentary, and in places informed legal opinion, by an author with a paralegal background and a direct commercial interest in the conclusion. It is not legal or tax advice. Take qualified advice before acting on the regulatory or compliance points.