Skip to Content
Knowledge baseBowtie analysis

Bow-tie analysis with example

A bow-tie analysis is a risk method in which you place one unwanted event in the centre, with all causes on the left and all consequences on the right. Between cause and event you draw the preventive barriers, and between event and consequence the recovery barriers. The diagram looks like a bow tie, which is where the name comes from. At a glance, you see which measures stop an accident and which limit the damage.

The bow tie is described as a technique in IEC 31010:2019 . It is worked out in the reference work Bow Ties in Risk Management  by CCPS and the Energy Institute. Within certification, you mainly use the bow tie for serious risks under hazards and risks in ISO 45001, in VCA and the Veiligheidsladder (Dutch safety ladder), and increasingly for scenarios in the risk analysis for ISO 27001.

Where does the bow-tie method come from?

The earliest known mention is training material from the British chemical company ICI, dating from 1979. In the 1990s, Shell made the method its company standard for safety risks, after which other sectors followed. Today the bow tie is used in the process industry, aviation, healthcare, construction and information security.

The parts of a bow tie

A bow tie has seven fixed parts. Use these names, so that every auditor or safety officer understands your diagram.

PartMeaningExample (roofing)
HazardSomething with the potential to cause harmWorking at height
Top eventThe moment you lose controlPerson falls from the roof
Causes (threats)What can lead to the top eventSlipping, tripping over material, roof edge gives way
Preventive barriersMeasures on the left: stop it from happeningEdge protection, fall protection, LMRA (Last Minute Risk Assessment)
ConsequencesWhat can happen after the top eventSerious injury, death
Recovery barriersMeasures on the right: limit the damageFall arrest with harness, BHV (Dutch company emergency response team), rescue plan
Escalation factorsWhat can make a barrier failRushing, bad weather, harness not inspected

The difference from a risk matrix: the matrix gives one score per risk. The bow tie shows which measures keep that risk low, and whether you depend too heavily on one barrier.

Example: bow-tie analysis for falls from height

A roofing company with VCA and ISO 45001 makes a bow tie for its biggest risk. The diagram looks like this in text.

CAUSES PREVENTIVE TOP RECOVERY CONSEQUENCES EVENT Slipping on wet -> anti-slip footwear, -> -> fall arrest harness -> serious injury roof stop work in rain Person and anchor point Tripping over -> tidy workplace, -> falls from -> rescue plan and -> permanent injury material LMRA at start the roof trained BHV member Roof edge gives way -> edge protection -> -> quick alarm -> death per plan 112 with location

The team works out each barrier in a table. This turns the bow tie into a working document rather than a picture.

BarrierSideOwnerHow do you know it works?Escalation factorExtra measure
Edge protectionPreventiveSite supervisorWorkplace inspection at startTemporarily removed for materialsOnly removed with a permit and harness
LMRA at startPreventiveEvery fitterCompleted LMRA in appRushing, routineSpot check by site supervisor
Fall arrest with harnessRecoveryFitterAnnual harness inspectionHarness not wornToolbox talk, sanctions policy
Rescue planRecoveryProject managerDrill once a yearNobody knows where the plan isPlan in the van and in the app

The analysis reveals a weak spot. Three of the four barriers depend on the behaviour of the fitter. The company therefore also chooses technical measures, such as permanent edge protection. This matches the occupational hygiene strategy in the Dutch Working Conditions Act (Arbowet) and the hierarchy of control measures in ISO 45001.

A barrier is only a barrier if it can stop the chain on its own. “Training” and “procedure” are support, not barriers. They make a real barrier, such as a harness, more reliable.

Bow-tie analysis for information security

The bow tie also works outside safety. A SaaS company with ISO 27001 makes a bow tie for the top event “attacker gains access to customer data”.

SideElementExample
CauseStolen passwordPhishing email
PreventiveBarrierMulti-factor authentication on all accounts
CauseUnknown vulnerabilityOutdated library in the application
PreventiveBarrierMonthly patching and vulnerability scan
RecoveryBarrierLogging with an alarm for unusual download behaviour
RecoveryBarrierIncident procedure and data breach notification within 72 hours
ConsequenceDamageData breach, fine, loss of customers

This shows which measures from Annex A belong to each scenario. It is useful when you justify your statement of applicability.

Step by step: making a bow-tie analysis

Choose the hazard and the top event

Take a risk that scores red or orange in your risk matrix. The top event is the moment you lose control, not yet the damage.

Name the causes and consequences

Work with people from the field. Five to eight causes and three to five consequences are usually enough.

Place the barriers

Start with the measures that already exist. For each barrier: who is the owner, and how do you check that it works?

Look for escalation factors

For each barrier, ask: when does it fail? Think of an extra measure for that case.

Use the bow tie

Link barriers to inspections, toolbox talks and audits. Take the bow tie out after an incident: which barrier failed?

Bow tie after an incident

After an accident or near miss, the bow tie is a strong tool for the accident investigation. You look at which barriers failed and why. To find the deeper cause for each failed barrier, combine it with a root cause analysis.

Common mistakes in a bow-tie analysis

  • The top event is already the damage. “Death” is a consequence. The top event is “person falls”.
  • Barriers that are not barriers. “Awareness” does not stop a fall.
  • Too much in one bow tie. One top event per diagram. Make more diagrams instead.
  • No owner per barrier. Then nobody checks whether it still works.

Further reading