Reliability
Root cause analysis
Work an equipment failure through a structured RCA: fishbone, 5 Whys, corrective actions and sign-off. Your cases are saved in this browser as you type.
This browser is not saving anything, so your work will be lost when you leave. Private browsing usually causes this.
Your analyses
Problem statement
What failed, where, when, and what it cost you. Write this before anyone starts naming causes, so the team is arguing about the same event.
Linked incident context
The records around the event. In UpKeep these link to the real assets and people; here they are names you type, one per line.
Assets
People
Locations
Vehicles
Checklists
Contributing factors
What across the system helped this happen. Spread wide before going deep: the first cause named in the room is rarely the one worth fixing.
People
Process
Equipment
Materials
Environment
Measurement
5 Whys
Follow one thread down. If an answer names a person or a broken part and stops there, you are not at the bottom yet.
- 1
- 2
- 3
- 4
- 5
Corrective actions
What changes as a result. An analysis that ends without one of these has cost the team an afternoon and changed nothing.
Review team
Who reviewed the findings and agreed with them. You are recording sign-off on their behalf, so check before you tick.
Related cases
Link this to another analysis when the same thing is happening elsewhere. Two linked cases is where a repeat failure stops looking like bad luck.
How to run the analysis
Root cause analysis fails in two predictable ways. It stops at the first plausible cause, usually a worn part or a person who made a mistake, or it produces a finding nobody owns and nothing changes. The structure here is built against both.
Start with the event, not the cause
The problem statement should be something everyone in the room would agree happened: what failed, where, when, and what it cost. Teams that skip this spend the first twenty minutes discovering they were discussing different events.
Spread out, then go down
The fishbone is for the group. Six categories, no wrong answers, everything on the board. The 5 Whys is for one thread out of that board, followed until you reach something you can change. Belt wear is not a root cause. A PM standard that was never updated after the line was modified is.
Finish with an owner and a date
A corrective action without a name against it is a suggestion. The action list takes both, and the sign-off list records who agreed the analysis was right, which is what you will want six months later when it happens again.
What this tool does not do
It does not read your work orders, so nothing is looked up for you and nothing is written back. It does not share: cases live in this browser, and the way to get one to a colleague is Export or Copy summary. Root Cause Studio inside UpKeep does both, ranks the work orders worth investigating from your live maintenance history, and lets reviewers sign off themselves.
Frequently asked questions
Where is my data stored?
In your own browser, under a single localStorage key. Nothing is sent to UpKeep and there is no account. That means your cases are private to this browser on this device, and clearing site data removes them.
Use Export to take a copy before you clear anything, or to move your cases to another machine.
Can two people work on the same analysis?
Not at the same time. Browser storage is per person, so this tool suits one facilitator writing up a review, then exporting the result to share.
The reviewer list and sign-off here are a record of who agreed to the findings, kept by the person running the case. Root Cause Studio inside UpKeep is the version where reviewers sign off themselves.
Why do I have to type in the work order?
This tool has no connection to your UpKeep account, so it cannot look up work orders, assets or users. Inside UpKeep, Root Cause Studio reads all of that live and carries it into the case for you.
What makes a work order worth analyzing?
Reactive work with at least one of three signals: $500 or more in recorded cost, 3 or more activity updates, or high priority. Those are the same thresholds Root Cause Studio uses to rank its recommendation queue.
Scheduled work is excluded on purpose. A PM that ran as planned is not an unexpected event, so it is not what an RCA is for.
Do I have to fill in every tab?
No. The progress bar shows what is done and what is not, but nothing is required and nothing is locked. Plenty of useful analyses stop after the 5 Whys and one corrective action.
What is the difference between the Fishbone and the 5 Whys?
The Fishbone spreads out: it asks what across the whole system contributed, sorted into people, process, equipment, materials, environment and measurement. It is the right tool for a group discussion early on.
The 5 Whys goes down: it takes one thread and follows it until you reach a condition you can actually change. Most teams use the Fishbone to find the thread worth pulling, then the 5 Whys to pull it.
See this running on your own work orders
Root Cause Studio in UpKeep starts from your maintenance history: it ranks the reactive work orders worth investigating, carries the asset, location and cost into the case, and collects reviewer sign-off in the app.