Not every risk needs a Monte Carlo model. Not every risk can be handled with a traffic-light matrix and hope. The skill is knowing which analysis earns its keep — and refusing to fake precision you do not have.
Qualitative analysis: the matrix that is allowed to be honest
Qualitative analysis scores likelihood and impact on a simple scale — often 1–5 or Low/Medium/High — then multiplies or maps them into an exposure rating. The point is ranking, not prophecy. You are deciding where attention goes this month.
Calibrate the scale in the room. If everyone rates every risk High/High, the matrix is a panic button, not a tool. If everything is Medium, it is a comfort blanket. Define what High impact means in this project: missed go-live, six-figure overrun, regulatory breach, or reputational damage. Vague scales produce vague priorities.
A matrix is enough when the decision is 'which five risks do we actively manage?' It is not enough when someone is asking whether to buy insurance, dual-source a vendor, or cancel a workstream. Those decisions need money attached, or at least a range.
When quantitative analysis is worth the hours
Quantitative analysis puts units on the uncertainty: days of delay, dollars of cost, or expected monetary value (EMV). EMV is probability × impact. A 30% chance of a $100k hit is a $30k exposure. Ugly, simple, and suddenly comparable across risks.
Use it when the decision cost is high: contingency sizing, go/no-go, or a change that trades one expensive risk for another. Do not use it to decorate a steerco pack. A three-decimal EMV built from a guess is still a guess. It just looks like it went to university.
If you lack data, say so. Analogous figures from the last similar project beat invented percentages. 'We do not know' is a professional sentence. 'Likelihood is 37%' when nobody measured anything is how PMs lose the room.
Do not mix the signals
Keep qualitative scores for triage and quantitative figures for the few risks that change a decision. Mixing them on one slide — red boxes next to fake-precise dollar figures — teaches executives to distrust both.
Re-score when the world changes. A risk that was Low/Low before the vendor went into administration is not still Low/Low because the spreadsheet cell has not been edited. Analysis is a timestamp, not a personality trait.
Watch for false comfort. A beautiful heatmap can coexist with zero responses. Scoring is not mitigation. If the highest-scoring risk has no owner and no action, you have produced art.
Summary & next steps
Use qualitative scoring to rank attention. Use quantitative analysis — including EMV — when a decision needs a number. Do not invent precision. Re-score when facts change.
Next: the response strategies themselves — avoid, transfer, mitigate, accept — and the equally important skill of picking one on purpose.
Practice this
Risk Register
Turn theory into a real deliverable. Generate a risk register with the AI tool and adapt it to your own project.
Open the generator