The Federal Aviation Administration once admitted something about itself in writing. It wanted pilots to report the events that had not yet turned into accidents, the misread altitude, or the runway somebody rolled onto without a clearance. Pilots were never going to hand that to the agency holding their certificates. So the FAA went looking for an objective third party and asked NASA, an agency with no regulatory or enforcement authority of its own, to take over receiving the reports, processing them, and stripping the reporter's identity out. Filing also carried confidentiality and limited immunity from FAA enforcement.1 The system began operating in April 1976 and still runs today. Fifty years of aviation safety improvement sits on top of a regulator concluding it had to get itself out of the mailroom.

The gap everyone agrees exists

Benchmark Gensuite, which sells environmental health and safety software and therefore has a stake in there being a visibility problem, surveyed more than 260 EHS professionals for a report it released in February. Ninety percent of them said incidents, hazards, or near misses are going underreported inside their own organizations, against 79 percent the year before.2 Forty-five percent guessed that as much as a quarter of their workforce files nothing at all, and 39 percent said they are missing the early indications of risk that would let them head off something worse.2

A number of write-ups turned this into "90 percent of incidents go unreported," which is a far bigger claim than the survey can carry. What moved eleven points in a year was the share of safety professionals who believe they are not seeing the whole picture. That is worth knowing on its own, and it is worth stating accurately, because the version circulating is the kind of number that gets pasted into a board deck.

Then comes the part I keep turning over. Asked why the reports are not arriving, nearly half of those same leaders said workers avoid reporting because the process takes too long, the systems are inadequate, or the steps feel tedious.3 The diagnosis is the form.

The other survey, asking the other people

Earlier this year MHFA England, a social enterprise that sells mental health training and published this research to launch a workplace campaign, put a related question to 2,000 working adults in the UK. Forty-five percent said they feel unable to raise mistakes or risks at work.4 Thirty-five percent said they do not feel safe asking for help.4 The figure that stopped me was smaller and much more expensive: 15 percent, roughly one in seven, said they have made preventable mistakes because they felt unsafe speaking up.4

Not one of those answers is about the form.

Two populations were asked to explain the same missing information, and they gave incompatible answers. One group described a user-experience problem. The other described what it costs them personally to be the person who says something. Both groups are being honest. The safety professional really has watched people abandon a twelve-field web submission halfway through, and shortening it really would help. That explanation also happens to be the only one available from where he is standing, because he is not the one who has to file the report and then sit across from a plant manager who reads it as an accusation.

Throughput in a reporting system is set by what happens to the reporter afterward. The length of the form is a rounding error next to that.

Why the tooling answer wins anyway

Buying a better tool is a decision a leader can make on a Tuesday, fund from an existing budget line, and point to in a quarterly review. Changing what happens to the messenger is slower, has no purchase order attached, and requires a senior person to concede that the current arrangement has been producing silence on purpose without anyone intending it. Offered a problem solvable by procurement alongside one solvable only by behaving differently in front of witnesses, most organizations find the procurement version more compelling. Then the reports still do not arrive, and next year's survey reads 91 percent.

In a leadership session I ran, a mid-level manager put her finger on the mechanism better than any research paper has. Her employer, she said, kept telling people to bring their authentic self to work, and then treated the people who did as problems to be managed. Another participant in the same group described an organization he belonged to where the person in charge called people out in front of everyone, and said his own response had been to stop volunteering anything at all. He was not disengaged. He had run the calculation and gotten a clear answer.

Something similar happens with employee surveys, where the instrument cannot find what leaders are hunting for. I covered that in why engagement surveys miss toxic managers and will not relitigate it here. Reporting is the more concrete case anyway. A hazard report is a fact somebody was holding and decided not to hand over.

What the FAA actually did

Notice what the aviation fix left alone. Nobody redesigned the form to be friendlier. The FAA changed who received the information and what could be done to you for having sent it, then paid another agency to hold the mail so the guarantee would be believable. Expensive, awkward, institutionally embarrassing for a regulator, and the reason the data started flowing.

So the question I would put to a leadership team staring at a reporting gap has nothing to do with how long the submission takes. It is what happened to the last three people who filed one. Does anyone know their names? If nothing happened to them, good, and does the workforce know that, or do they only know the single story about the person it went badly for? That story travels farther than any policy you publish, and it is doing more work than your software.

The information already exists, and one in seven people are paying for it with errors they watched coming. What would it take for it to reach you?