The difficulty in making the case for anything in this area is that the problem has no invoice. Nobody receives a bill for a conversation that did not happen. The cost turns up months later, in categories that are already budgeted for and already blamed on something else.
That makes it easy to defer indefinitely. It also makes it worth modelling — carefully, and with the limits admitted up front.
Where the cost actually lands
Three places, mostly:
- Absence. Pressure carried long enough tends to become time off, usually later and for longer than it would have been if something had been said earlier.
- Replacement. People leave. Recruiting and getting somebody productive again has a cost most organisations can put a number on, even if they rarely attribute it to this.
- Incidents and investigations. Fatigue and distraction are not abstractions on a working site. An investigation is expensive in a way that is very hard to argue was unrelated.
None of these are recorded as mental-health costs. All of them are affected by whether somebody felt able to say something in February.
How to model it without overclaiming
The honest way to do this is to build it from figures your own organisation already holds, and to be explicit that the model is an illustration rather than a forecast.
Take one site. You need four numbers, and three of them you already have: your day rate for absence cover, what it costs you to replace somebody, and roughly what an investigation costs when one happens. The fourth — how much of that is attributable to people not speaking earlier — is the one nobody can give you honestly, which is exactly why it should be shown as an assumption rather than buried in a total.
A model that hides its assumptions is a sales tool. One that shows them is an argument you can take into a room with a finance director.
What the comparison is actually between
It is tempting to present this as "cost of the platform versus cost of doing nothing", but doing nothing is not free and is not static, and pretending to know the difference precisely is where these models lose credibility.
The more defensible framing is narrower: on one site, over one year, with the licence counted as a cost against it, what is the range in which this changes anything? If the honest answer is that it only takes one avoided replacement and a shorter absence to cover the licence, that is a stronger argument than a large number nobody believes.
Our own pricing, so the model is checkable
EasierAsk is sold as an annual licence banded by employee headcount, starting at £3,000 a year up to 99 employees and rising by band, with the largest organisations negotiated. Workers never pay: there is no paywall in front of help.
That is stated here so a model can be checked rather than taken on trust. If a supplier will not tell you the price until late in the process, you cannot build the case, which is usually the intention.
Any modelled figure you see from us — here, in the experiences, or in the walkthrough pack — is a sector-level illustration built on stated assumptions. It describes a plausible shape, not your workforce, and it is not a guarantee of outcomes. Run it on your own numbers and treat any figure you cannot source as an assumption.
The one number worth getting first
Before any of this: how many times has anybody approached your Mental Health First Aiders in the last twelve months?
Most organisations have never counted, and the answer usually reframes the conversation faster than a cost model does. If the trained help you already pay for is being approached a handful of times a year, the question stops being whether to spend more and becomes why the existing spend is not reaching anyone.
