Decision Mining: group decisions that actually stick

A surprising number of businesses make good decisions when one person is involved, and quietly struggle with group decisions: the kind that involve more than one stakeholder, more than one viable option, and more than the brain space of any single leader. The default is to muddle through with whatever mix of meetings, emails and senior intervention is available at the time, and to find out months later whether the choice was a good one.

Decision Mining is a structured alternative to that default. I developed it in 2019 as a deliberate sibling to Idea Mining. The mining metaphor carries across neatly. The good decision already exists somewhere within the available options, the evidence and the experience of the people in the room. The question is whether anyone will do the work of extracting it, refining it and committing to it deliberately rather than by accident.

Why it is separate from Idea Mining

From the outside, the two frameworks look similar. Both are structured, both use the mining metaphor, and both produce something refined from raw inputs. They differ in one important respect, and that difference runs through the rest of the design.

Idea Mining is divergent. The aim is to produce as many viable options as the team can generate, while resisting the urge to close down on a favourite too early. The output is a set of refined candidate ideas, from which someone will later have to decide.

Decision Mining is convergent. The options largely exist by the time it starts. The aim is to commit to one of them (or combine several into a single coherent direction) in a way the whole group will actually back when it comes to execution. Doing that well needs explicit consensus mechanisms, explicit mitigation of cognitive bias and an explicit handoff from decision into action. None of these is really needed in Idea Mining. All of them are essential here.

In practice the two chain together naturally. Idea Mining produces a refined option set. Decision Mining picks from it, locks the choice in and triggers the work that follows.

The five phases

A Decision Mining cycle can run for an hour or for several days, depending on the weight of the decision and the size of the group. The phases are the same either way. What changes is the depth at each step.

Where it pays back its overhead

On paper the framework looks heavyweight. In practice the discipline is in the sequence, not the effort at any one stage. A simple decision can run through all five phases in about an hour, and the useful property is that the same conversation never happens twice. The matrix is built before the options are evaluated. The options are evaluated before the decision is made. The decision gets named owners before the room empties. Each phase locks in the one before it, so the outcome cannot silently unravel back into the original confusion the way unstructured decisions often do.

Three situations repay the overhead with interest.

The big-stakes single decision. Major investments, restructures, vendor selections, key hires and strategic pivots. The cost of getting one wrong is high enough to justify the structure. Plenty of businesses under-invest in structure exactly here, because the urgency seems to leave no room for it. That is usually when structure pays back most.

The decision that has failed to stick. If the same question keeps returning to the leadership team, the previous “decision” never achieved the right kind of agreement. Running it through Decision Mining once forces the team to confront what kind of agreement is achievable on this question, and what action genuinely follows from each level of agreement.

The cross-functional decision where ownership is unclear. The link between Phase 1 (what are we deciding?) and Phase 5 (who will own the action?) makes ownership ambiguity visible early. If the room cannot agree on the target, the chance of it agreeing on action is essentially zero.

How it fits with the rest of the kit

Decision Mining sits inside the Conscious Collaboration shell. It needs the same structured process, shared mindset and supporting technology as any productive group activity. The Cynefin framework tells you, before you start, what kind of decision you are making. A known-good-practice choice means Phase 4 can usually be quick. An expert-analysis choice means Phases 2 and 3 need more time and depth. A complex-domain choice means the “options” in the room are really experiments, not commitments, and Phase 5 sets up the next learning cycle rather than the next operational push. Idea Mining feeds the option set into Phase 3 when there isn't already a good one on the table. Together these frameworks form a pipeline that runs from sensing a change in the environment through to acting on it.

Where I use it most

I use Decision Mining most often for board-level decisions in technology-led businesses and for strategic pivots in scale-ups. Both share a profile: the cost of a poor decision is high, the people in the room have strong but different views, and without structure the default is to muddle through under time pressure. In these situations, a 90-minute Decision Mining cycle, sometimes with me facilitating and sometimes with me coaching the team to run it themselves, reliably produces better outcomes than the same 90 minutes spent debating the same options without structure.

The framework does not guarantee a good decision. Nothing does under genuine uncertainty. What it does guarantee is that if the decision turns out to be wrong, you can look back and see exactly why. In my experience, that is the main thing that makes the next decision better.

Four things worth taking seriously

For boards: if the same question keeps coming back, the last decision never reached the right kind of agreement. Ask what kind of agreement was targeted before debating the options again.

For founders: the decisions that feel too urgent for structure are the ones where structure pays back most. An hour spent on all five phases is cheaper than a pivot made by muddling through.

For anyone chairing a decision: agree the selection matrix before anyone argues for an option. Whoever controls the matrix controls the outcome, and silence in the room is not alignment.

For engineering and operations leaders: no decision leaves the room without a named owner and a timeline. Decisions without owners decay within days.

I would be interested to hear which decision keeps coming back to your leadership team, and what kind of agreement it actually reached the last time.

Related: Idea Mining · Conscious Collaboration · Cynefin and Conscious Collaboration · all writing