A good risk workshop is a facilitated argument with a written output. A bad one is a brainstorm that produces thirty vague rows, two dominant voices, and a lingering sense that you 'did risk.' The difference is design.
Design the session before you book the room
Decide the outcome: a ranked top ten with owners, not a complete inventory of the universe. Send a short pre-read: scope, constraints, and any existing register. People who walk in cold will only name the obvious risks, then check email.
Invite the people who can see different failure modes — delivery, tech, commercial, operations, and at least one sceptic. A workshop of only enthusiasts produces a cheerful register. You want the person who will say the quiet part: the dependency that is already late.
Timebox. Ninety minutes is plenty if you have a structure. Three hours is how you get diminishing returns and a parking lot the size of a car park. If you need more, run a second session on a specific category, not a longer first one.
In the room: silence, then structure
Start with two minutes of silent writing. It stops the first speaker from setting the entire agenda. Then collect, cluster, and rewrite into cause–event–effect before you score anything. Scoring unfinished sentences is how you get a confident matrix of nonsense.
Facilitate the power dynamics. The senior person who says 'that's not a real risk' will shut the room down. Your job is to park the dismissal, finish the capture, and come back to challenge with evidence. You are not hosting a consensus hug. You are protecting the list from politics long enough to write it down.
Score as a group, but do not average your way to Medium. If engineering says High and the sponsor says Low, that disagreement is the finding. Record both views and the rationale. Fake agreement is how workshops manufacture false confidence.
After: the part most teams skip
Within 48 hours, send the cleaned register: top risks, proposed responses, owners, and open questions. Ask owners to confirm or refuse. An owner who never confirmed is not an owner.
Book the first review date before you leave the session. Workshops that end with 'we'll tidy this up' are how tidy never happens. The workshop is initiation. The cadence is the product.
If the session produced more heat than light — a fight about scope disguised as a risk debate — that is useful. Write it down as a stakeholder or alignment risk and take it to the sponsor. Pretending the room was 'aligned' helps no one except the minutes.
Course complete — what's next
Design for a ranked, owned output. Silent capture, then structured rewrite, then honest scoring. Confirm owners within two days and lock the first review date before the workshop ends.
You now have a working risk practice: framework, analysis, response, control, and facilitation. Use it on a live project this week — even a small one. A register with five real risks beats a template with twenty ghosts.