FROID NR-1/ISO-45003 — anchored in ISO 45003:2021
Most tools survey your workforce and hand back a score. The harder obligation — showing that a preventive measure actually reduced exposure — needs a baseline, a second cycle, and the discipline to report no improvement when the evidence does not support one. FROID measures that, and refuses to claim success when the confidence interval will not carry it.
The obligation
Not always as a named rule about "psychosocial risk" — more often as the general duty to identify and control every hazard arising from work, which psychosocial factors have never been exempt from.
The reference guideline for managing psychological health and safety within an occupational health and safety management system. It is guidance rather than a certifiable standard — which is precisely why buyers should ask a supplier what method it actually implements, and against what evidence.
The Framework Directive obliges employers to evaluate all risks to workers' safety and health. Several member states have since issued specific guidance on work-related stress and organisational factors. The exposure is rarely a missing rule — it is a missing assessment.
Brazil wrote psychosocial factors expressly into the risk inventory, with a start date and an enforcement date. That produced something unusual: a jurisdiction where the method had to survive inspection. It is the regime this platform was built and tested against.
The method
Both are checked before any result is released, and both are enforced inside the database rather than in the interface — because a rule in the interface is a rule someone can route around.
Fifteen substantive responses in a campaign, ten in any breakdown. These are absolute counts, and they protect the person: below them, knowing a group's average is close to knowing what each member answered.
No amount of extra participation fixes this one — the floor looks at how big the group is, not at how many replied. A team of six will never publish its own breakdown, and any supplier promising otherwise is promising something the arithmetic forbids.
The responding cohort has to be able to speak for the declared headcount: a sample for a proportion with finite population correction, at 95% confidence and a 5-point margin.
This one participation does fix, and the dashboard shows exactly how many responses are still missing. The two gates therefore call for opposite remedies — which is why a report that fails to say which one blocked it sends you down the wrong road for a whole cycle.
The sample requirement flattens as the organisation grows. Ninety-eight people need seventy-nine responses; three thousand people need three hundred and forty-one. Below roughly a hundred, the required sample reaches the entire workforce and the exercise becomes a census.
Said before the contract rather than after the collection window closes — it is the promise that most often breaks at the end of a cycle, with the money already spent.The differentiator
Almost every supplier asserts effectiveness in a report. Very few can demonstrate it, and the difference is the whole argument in front of an auditor or an expert witness.
The comparison is always a unit against its own baseline, on the same instrument, before and after the measure — never against a market average or a generic population, which is what makes a comparison contestable.
We compute the standardised difference between the two moments and its margin of error, which depends on how many people answered. Classification uses the conservative bound of the interval: only the magnitude the whole interval supports gets asserted.
The measure is recorded as ineffective, the risk is not marked down on paper, and it goes on the list of what has to be corrected. A supplier that only ever reports improvement is not measuring — it is confirming what it already wanted to say.
Real engine output on demonstration data. The first two are assertions; the third is the refusal to make one.
| Dimension and unit | Before → after | Effect | Verdict |
|---|---|---|---|
| Excessive demands · Support | 4.39 → 2.94 | +3.05 ±0.51 | effective |
| Harassment · Support | 1.85 → 3.17 | −2.86 ±0.49 | worsened |
| Harassment · Logistics | 2.50 → 2.14 | +0.60 ±0.60 | no change |
The third line is the one that matters. The average fell, the effect is positive — and the system refuses to say it improved, because with twenty-two responses the margin of error reaches zero. That is the difference between measuring and asserting.
What no one else delivers
Participation that fell short, a team too small, a site with twenty people: ordinary situations, and the market's usual answer is a blank page.
An empty panel is not neutral. Anyone who opens a report and sees nothing concludes there is no risk there — and that is the one conclusion the absence of data never authorises. A document that shows three units and says nothing about the fourth asserts, by silence, that the fourth is fine.
Suppressing is concealing. Declaring insufficiency is documenting.A breakdown that fails a gate does not disappear from the report. It enters the same inventory — never an appendix, never omitted — as a declared-insufficient line, naming which gate blocked it, what the remedy is, and stating explicitly that this does not mean there is no risk.
An inventory line either carries all four numbers — cohort, mean, severity and likelihood — and no gate, or it carries a gate and an escalation note with those four left null. There is no in-between state: the database refuses it. A half-filled line is exactly what an auditor reads as low risk, and that is the mistake the design makes impossible.
The boundary
This is not an access setting someone can loosen. It is how the database is built, and we will demonstrate it live in thirty seconds.
The individual-response tables carry no read policy at all, and the role the system runs under has no permission over them. The only way data leaves the database is an aggregation function that has already applied the minimum group floors.
This is not a retention policy — the column does not exist. You provide a payroll number; the system stores a cryptographic pseudonym of it and a digest of the invitation link, never the pair between the two. Who received which link is known only to your own HR.
Someone who suspects the answers are readable replies with what is safe, not with what is true — and then the organisation pays for a portrait that does not match it. The guarantee protects the employer at least as much as the worker.
Scope
Pricing
FROID NR-1/ISO-45003 assesses working conditions, anonymously and in aggregate. It does not assess individuals, does not produce a diagnosis and does not hand clinical data to the employer.
| Component | Amount |
|---|---|
| Platform base, per establishment | R$ 200 / month |
| Band 1 — from 1 to 100 workers | R$ 15.00 / worker / month |
| Band 2 — from 101 to 300 | R$ 12.50 / worker / month |
| Band 3 — from 301 to 1,000 | R$ 9.30 / worker / month |
| Band 4 — above 1,000 | R$ 6.55 / worker / month |
Bands are cumulative and apply to the company's total headcount: with 300 workers you pay R$ 15 for the first 100 and R$ 12.50 for the next 200. The base is per establishment — not per department: a company with five departments at the same address has one base, not five. Amounts are in Brazilian reais. The proposal simulator computes the figure for your headcount.
The calculation comes from FROID's pricing engine, not from this page: the amount carries the version and the fingerprint of the commercial table that produced it.
Estimate based on the table in force. The contracted amount is the one in the Commercial Proposal.
Observation of the real activity, dialogue with workers and the execution of prevention measures belong to the contracting company. FROID does not conduct field assessment, does not perform an AET and does not implement any measure — neither with its own team nor through third parties.
What FROID delivers is the instrument: the measurement, the record of each piece of evidence with its declared method, and the verification of the effectiveness of the actions the company decides and carries out, in the form the rule requires.
The reason is technical, not commercial. The company is the one that knows the activity, the organisation of work and the concrete conditions — and it is to the company that the rule assigns the decision on measures, their implementation and the signing of the documents. A supplier that proposed and executed its own measures would be measuring the result of its own work.
Next step
A single site, criteria documented from the first day, and exposure characterised in a form your assessment can absorb. If the method holds up there, it scales; if it does not, you found out cheaply.