The £51 Billion Question

The number comes from Deloitte. Fifty-one billion pounds every year, paid by UK employers as the annual cost of mental ill health at work. Presenteeism accounts for roughly half of it. The rest is turnover, absenteeism, and the quieter knock-on effects of staff who are nominally at their desks but functionally elsewhere.

It is a figure that has been quoted so often it has lost the ability to shock. Which is a problem, because shock is the appropriate response. Fifty-one billion pounds is roughly three times the entire NHS mental health budget. It is spent by organisations that would scrutinise a six-figure procurement decision for weeks. And it disappears every single year, reliably, with almost nothing to show for it.

That is the puzzle worth sitting with. Because the same research report that produced the fifty-one billion figure also produced the four pounds seventy. A return on investment that, in any other corner of corporate finance, would cause a stampede.

So where is the stampede.

The working theory inside The Threshold Collective is that the stampede is not arriving because most of what is being sold as workplace wellbeing does not actually return four pounds seventy. It returns nothing. It returns a procurement tick-box. Employers buy it the way they buy insurance they hope never to use, and employees use it the way they use a fire exit they hope never to need.

Somewhere between the spreadsheet and the shop floor, the four pounds seventy evaporates.

The largest share, presenteeism, is also the hardest one for any employer to see. It is the cost of a person who shows up, logs on, sits through meetings, and produces work that is measurably worse than the work they were producing six months ago. Presenteeism is invisible on sick-day dashboards. It is invisible on attendance reports. It shows up only in the slow erosion of output quality and, eventually, in the turnover line. Where another twenty billion pounds disappears.

A system that cannot see its largest cost cannot reduce it. Which is how a country ends up spending a decade of conversation about workplace wellbeing and, at the end of the decade, a bigger bill.

The rest of this issue is about what a system that could see would look like. What it would measure, what it would ignore, who it would be built by, and why it does not yet exist at scale. Section 2 begins with the thing the market is actually selling.

Walk into the procurement conversation at any mid-sized UK employer and the shortlist is more or less fixed. A meditation app. An employee assistance programme. A mental health first aider course. Perhaps a webinar series delivered by a consultancy whose name nobody inside the building will remember by Friday. All of it pitched as wellbeing. All of it bought with good intentions. None of it built to change the numbers on the previous page.

This is not because the vendors are dishonest. It is because they are selling products that serve a different purpose. A meditation app is a consumer comfort tool. An employee assistance programme is a compliance and liability instrument. A first aider course is a training product. These are all legitimate things. They are simply not the things that reduce presenteeism, turnover, or absenteeism at a population level inside an organisation.

The word wellbeing did a lot of damage here. It is a soft, broad, agreeable word, which is why it spread so quickly through human resources departments in the 2010s. But broad words make bad procurement categories. When a finance director is asked whether the organisation has invested in wellbeing, the correct answer is almost always yes. When the same finance director is asked whether the investment is reducing the drivers of the fifty-one billion pound bill, the answer is almost never on the desk.

The category collapsed the question. Wellbeing became the word, and the word ate the outcome.

Usage data tells the next part of the story. Employee assistance programmes in the United Kingdom report utilisation rates that typically sit in single digits. Meditation apps licensed at organisation level show open rates that collapse within weeks of rollout. Mental health first aiders, once trained, often report that they are rarely approached. This is not a failure of the individual products. It is a signal that the products do not sit close enough to the actual moment of need to be reached for when the need arrives.

The moment of need, in most cases, is ordinary. Not a crisis. A Tuesday afternoon three weeks into a project that has drifted off track. A Monday morning after a weekend of not sleeping well. The moment when a person is still functional and still at their desk, but the margin between coping and not coping has narrowed. That is where presenteeism lives. And almost nothing currently on the market is engineered to notice it.

A tool used by seven percent of a population cannot meaningfully shift outcomes across the other ninety-three. The mathematics does not allow it. Which means the organisations paying for these services are, on the whole, paying for the presence of the service rather than the effect of it. And the employees inside those organisations are, on the whole, unaware that the service exists, distrustful of it, or confident that reaching for it would be read as weakness by someone in the building.

None of this is new. The interesting question is why the market has not corrected.

The buyer does not yet know what to ask for because the category was defined by the vendors, and the vendors defined it around the products they already had. This is the normal shape of a young market. Section 3 is about what the category will look like once the buyer knows the question.

Start with what such a system would refuse. It would refuse to identify individuals to management. It would refuse to predict who is about to leave in a way that enables pre-emptive performance action. It would refuse to hand aggregated wellbeing data to anyone whose job is to optimise for output first and people second. These refusals are not ornamental. They are the load-bearing walls of the whole structure, because without them nobody on the shop floor would ever tell the system the truth.

A workplace wellbeing system that cannot be trusted by the workforce it measures is, by definition, producing garbage data. Garbage data produces garbage predictions. Garbage predictions produce exactly the category of tool the market is already full of.

Predictive rather than reactive is the part that sounds, on first hearing, like the thing the current market already claims to offer. It is not. Most existing wellbeing products are reactive by design. A person is struggling, so they open the app. A person is in crisis, so they ring the helpline. The intervention arrives after the signal has already become a problem. A predictive system works earlier. It reads the shape of the pressure while it is still just pressure. It acts before the signal becomes a sick day, a resignation letter, or a quiet deterioration in the work.

Governed rather than optimised is the part the existing market finds hardest to swallow. Every other category of software is rewarded for optimising. More engagement, more data, more upsell. A workplace mental health system has to refuse that logic, because the natural direction of optimisation in this category is surveillance. Governance is the brake. And for some buyers the brake is the whole product. The reason they will pay for it is because it is the only thing in the room they can trust to stop.

Lived-experience led is the part nobody else can copy. The other two properties are, in principle, buildable by any sufficiently funded team. The third is not. Lived experience is the thing that tells you which questions the system must refuse to answer. It is the thing that tells you the difference between a supportive tone and a patronising one. It is the thing that makes a refusal sound like respect rather than a legal disclaimer. You cannot hire it, and you cannot pattern-match it from research papers. It has to be in the room.

Which brings the question back to the fifty-one billion pounds. A system with all three of these properties would not be cheap to build. It would require clinical oversight, engineering discipline, and a governance structure that a normal startup would find inconvenient. It would take longer to ship than the current category averages. It would refuse certain paying customers. It would disappoint some procurement departments by being unwilling to produce the reports they most want to see.

And it would still return four pounds seventy on every pound spent. Probably more, because the current figure is averaged across a market that is largely selling the wrong products.

The gap between fifty-one billion pounds spent and two hundred and forty billion pounds of return not realised is the shape of the market that does not yet exist. Somebody is going to build it. The interesting question is who, and under what governance, and whether the people building it have been close enough to the problem to understand why the current market cannot see its largest cost.

That question is why Momentum exists. Each issue will take one corner of this landscape and examine it without the usual softness that the category defaults to. Not because softness is wrong. Because the fifty-one billion pound bill will not be paid off with softness alone.

Issue 02 will look at the four-person model. Why a workplace mental health consultancy that does not include a lived-experience leader, a consultant psychiatrist, and a substance-use specialist is structurally incapable of doing the work the category requires. Thank you for reading Issue 01.