Employee referral: the cheapest hiring channel and its blind spot
Employee referrals hire faster and cheaper than any advert. How to run a scheme, what to pay and when, and the homogeneity problem that comes with it.
Most incident reports are written to close the file. The useful ones separate what happened from who is to blame — and organisations that cannot make that separation stop hearing about incidents at all.
An incident report is the record of something that went wrong: an injury, a near miss, a spill, a security breach, an outage. Its purpose is to establish what actually happened while people still remember, so the same thing is less likely to happen again. Almost every organisation writes them and a striking number gain nothing from it, because the reports are written to satisfy a process rather than to be read.
The distinction that decides whether the system works is between a report and a judgement. A report states facts, times and sequence. A judgement assigns cause and responsibility, and belongs in a later stage, done by someone with time to look. Merging them produces documents that are defensive, vague, and useless as evidence.
A falling number of incident reports is ambiguous and usually misread as good news. It means either that fewer things are going wrong or that fewer people are telling you — and the second is far more common after an investigation that felt like a search for someone to blame. Watch near-miss reports specifically: they are voluntary, so they fall first and fastest when trust goes. In a workplace some incidents become reportable by law, and riddor sets which ones and by when.
A report that closes on filing has wasted the reporting. The sequence that works is: report the facts, triage by severity and recurrence, investigate the ones that warrant it, decide corrective actions with owners and dates, and check later whether the action worked. The middle of that chain is covered in root cause analysis and the end in corrective action; the underlying question of what could have caused it in the first place belongs to risk assessment.
Ettex Forms is the reporting end — a short form anyone can reach, on a phone, without an account — and the submissions land as records in Ettex Records where they can be filtered by type, area and severity, and where the investigation and its actions attach to the original report rather than living in a separate document.
Stating the limits plainly: there is no anonymous reporting channel with the guarantees a whistleblowing scheme requires, no regulatory notification — several jurisdictions require certain injuries and data breaches to be reported to an authority within a fixed period, and that remains your obligation and your deadline — and no incident-management platform features such as on-call escalation. What we hold is the record and the trail from it to the change.
Reporting is the first half. Where the incident involves a serious injury or a dangerous occurrence, an external report may be legally required within days, and the internal form should flag that rather than leaving someone to remember it. Problem management is the other half: one report closes an event, the pattern across reports closes a cause.
Time, place, people involved and witnesses, the sequence of events in plain language, immediate actions taken, and the outcome. Cause and responsibility belong to the investigation, not the report.
Yes — they are the cheapest safety information available, arriving without the injury. They are also the first reports to disappear when people stop trusting what happens next.
Same day, while memory is accurate and evidence intact. Some categories carry legal deadlines for notifying an authority, which depend on your jurisdiction and are worth knowing before you need them.
Almost always because an earlier report led to blame rather than change, or to no visible response at all. Acknowledging every report and publishing what changed does more than any reminder.
Employee referrals hire faster and cheaper than any advert. How to run a scheme, what to pay and when, and the homogeneity problem that comes with it.
Screening decides who gets a conversation. How to set criteria before reading applications, what to screen on by phone, and the shortcuts that quietly become discrimination.
Conversion tracking records when a visitor does what the page existed for. Which events to define, why platform numbers disagree, and the double-counting nobody checks.