ModulesIncident InvestigationReport & Learn
Find why, not who.One investigation, from the first fact to the verified action.
For investigation leads and EHS teams: work the causes as a tree, a fishbone or an outline, and test each barrier against what your HAZOP, hazard identification or permit relied on. The same causes in every view.
Coupling CPL-305 came apart under pump pressure
- Cause
- No hold point between connection and pump start
- From
- Timeline · 10:33 Pump started
- Category
- Method
- Root cause
- Root cause
- Actions
- AT-0232 · Done
4 of 4 root causes have an action
The same causes, three views. Switch the view: nothing is retyped. An example investigation; the fields are the product’s, the values illustrative.
Where it connects
On the same loop: Incident Reporting, Safety Talks, Hazard & Risk Register and 38 more modules in one workspace.
Method note
Five whys, fishbone or barriers? Each one sees a different part.
Five whys is quick and easy to teach. Its critics say a single chain forces one root cause on an event that has many.
Leandro · Founder, UProcess
A fishbone spreads the search across categories. Barrier analysis starts from what should have stopped the event. Experienced investigators use more than one, and still argue about which comes first.
So the causes live in one model, and every view edits the same tree.
- Branching whysA tree where any cause can have more than one why, and any node can be a root cause.
- FishboneThe same causes on 6M bones. Drag a factor to another bone and its category changes everywhere.
- Barrier analysisEach failed or missing barrier starts a cause chain of its own.
Where Incident Investigation usually breaks
Four places a good investigation goes wrong, and what the record does about each.
One chain, one root cause.
A critique in BMJ Quality & Safety shows that five whys pushes the team down a single path to one distal cause, while a causal tree of the same event finds far more contributing factors.
Source: Card, BMJ Quality & Safety (2017)
In UProcess Any cause can have several whys, and any node can be marked as a root cause. Tree, fishbone and outline show the same causes, so switching views never loses one.
Causes · INV-2026-014
Coupling came apart
Four root causes on one event.
The finding stops at the operator.
HSE’s investigation guide says a conclusion that operator error was the sole cause is rarely acceptable, and that training, supervision and design must be examined.
Source: HSE, HSG245
In UProcess Barrier analysis records each barrier as effective, failed or missing, linked to the HAZOP or hazard identification safeguard, permit, isolation certificate or register control it came from. A failed or missing barrier starts its own cause chain.
Barriers · INV-2026-014
Each failed or missing barrier starts a cause chain.
Actions fix the surface.
After Texas City, the refineries review panel found root cause analysis so weak that corrective actions addressed surface causes, and process safety deficiencies were not tracked to completion.
Source: US DOE Safety Advisory 2007-02
In UProcess Actions are raised per root cause, and the investigation shows coverage: how many root causes have an action. Each action can be set to close only after someone checks that it worked.
Action Tracking · AT-0231
Replace bay couplings with pinned cam-lock couplings on all three bays
- Owner
- Daniel Kowalski
- Due
- 30 Nov 2026
- From
- Incident Investigation · root cause “Bay 3 couplings have no locking pins”
Pinned couplings fitted and pull-tested on bays 1–3 · checker Aisha Bello
4 of 4 root causes have an action
The same event, years later.
The CSB found that an action item from 1994 to review the Texas City blowdown system, raised after two serious incidents, was never completed.
Source: U.S. Chemical Safety Board
In UProcess Lessons learned become a safety talk in one step, and the investigation shows the talk and its status. A hazard the investigation uncovered can open a line in the Hazard & Risk Register.
From INV-2026-014
- 4 Nov
- Lessons learned recorded · Step 6
- 12 Nov
- Safety talk TK-0057 published · Hand-tight is not tight
- 18 Nov
- HRR-0063 opened · Hazard & Risk Register · From INV-2026-014
Safety talk · Published
Investigation INV-2026-014Lead Lucas MoreauClosed 4 Nov 2026
| Root cause | Action | Owner | Due | Status |
|---|---|---|---|---|
| The unloading checklist has no pin check | AT-0232 · Add pin check and second-person sign to the checklist | Grace Okafor | 15 Nov 2026 | Checked |
| Bay 3 couplings have no locking pins | AT-0231 · Pinned cam-lock couplings on bays 1–3 | Daniel Kowalski | 30 Nov 2026 | Done |
| No hold point between connection and pump start | AT-0232 · Same action | Grace Okafor | 15 Nov 2026 | Checked |
| XV-305 closure time never tested | AT-0233 · Test and repair XV-305; add to the test schedule | Daniel Kowalski | 20 Nov 2026 | Open |
4 of 4 root causes have an action
The record an auditor asks for.
An investigation is audited on what it changed.
- The timeline, with each event and condition, and the ones marked as possible causes.
- The barriers: effective, failed or missing, linked to the study, permit or control they came from.
- Root causes, each with its actions and their status.
- Lessons learned, and the safety talk made from them.
What’s in an investigation
- Origin
- An incident report, a concern or a safety observation, shown at the top. Equipment tags in the header.
- Timeline
- Events and conditions in sequence, as a chart or a list. Any entry can be marked as a possible cause.
- Causes
- Branching whys and root causes marked on any node, in tree, fishbone and outline views.
- Barriers
- Each effective, failed or missing, linked to a HAZOP or hazard identification safeguard, a permit, an isolation certificate or a register control.
- Evidence
- Attached to a timeline entry, a cause or an action. Photos from the report, the concern or an inspection are used by reference, not copied.
Questions
Can we use five whys and a fishbone on the same event?
Yes. They are views of the same causes, so an edit in one shows in the other.
Can a barrier point to our HAZOP?
Yes, to the safeguard of a scenario in your study, with what the study credited.
Do we have to complete every step?
No. The steps guide the work without blocking it. You can close with gaps, and the coverage line shows them.
What happens to the lessons?
They can become a safety talk, which appears on the investigation with its status.