In short: When a committee decides, the minority view is smoothed out of the record. Over time, recorded dissent is often the most predictive signal an organisation has, and it costs one line to keep.
Key points
- Collect each participant's view and probability before discussion, because views converge once the most senior person speaks.
- Record dissent in one attributed line with a probability, kept separate from assigned devil's-advocate challenges.
- Review dissent against outcomes at each falsifier check date.
- Track each person's dissent hit rate by decision class over several cycles.
- Do not use dissent records in performance evaluation in the first year.
Decision records usually store the result of a discussion: approved, with conditions. Minutes record consensus because that is what minutes are for. The analyst who argued against it, and the specific risk they named, disappears into the memory of the people who were in the room.
That is a loss twice over.
First, the dissent is frequently where the real falsifier lives. The person who disagreed was usually pointing at the risk that, if anything, would later materialise. When it does, the organisation discovers it was warned only if someone remembers.
Second, dissent histories are the clearest view you will ever get of who reads a class of situation well. Agreement tells you little about individual judgment. Disagreement that turns out right tells you a lot.
How to capture it well
Independent views first. Ask each participant to record their view and probability before discussion. Once the most senior person has spoken, views converge, and the dissent you capture afterwards is already diluted.
One line, attributed, with a probability. "Against. Customer concentration risk understated; 35% this account is lost within a year." That is enough.
Genuine, not assigned. A designated devil's advocate produces weaker signal than real disagreement, because everyone knows it is a role. Record assigned challenges separately from genuine dissent.
How to use it
At the falsifier check date, review the dissent alongside the outcome. Was the minority right? About the thing they named, or something else?
Over several cycles, track each person's dissent hit rate by decision class. After a while you can weight a dissent from someone with a strong record in that class differently from one without. That is the closest thing to institutional wisdom a system can hold: not a rule, but a memory of whose doubts deserve attention.
Protecting dissenters
A visible record of who predicted what can feel like exposure, and in some cultures it will suppress candour rather than encourage it. Decide deliberately who can see dissent: the panel, the function head, or only the review gate. Do not use dissent records in performance evaluation in the first year. The aim is to make disagreement safer, not riskier.
In practice
- Collect independent views and probabilities before discussion in one committee.
- Record dissent in one attributed line with a probability, separate from assigned challenge.
- Review dissent against outcomes at each falsifier date.
- After a year, show each person their own dissent hit rate, privately.
What would prove this wrong. Recorded dissent proving no more predictive of bad outcomes than majority views over a meaningful number of decisions. Then dissent is noise, and the record is clutter.
Common questions
Why should committees record dissenting views?
The dissenter is often pointing at the risk that later materialises, and dissent histories show who reads a class of situation well. Minutes usually record only consensus.
How do I capture dissent in decision meetings?
Ask each participant to record their view and probability before discussion, then keep dissent as one attributed line with a probability, separate from assigned challenges.
Does a devil's advocate count as dissent?
It produces weaker signal than genuine disagreement because everyone knows it is a role, so record assigned challenges separately.
How do I stop dissent records from discouraging candour?
Decide deliberately who can see dissent, do not use the records in performance evaluation in the first year, and show people their own hit rate privately.