The method: reading the test bench's signals

Every Hub case comes out of a real run of the Kapari engine: a panel of simulated voices reacts to a decision, and the test bench returns how the panel responds on five steps (from outright rejection to clear support), a reception risk (Low, Moderate, High), the range of reactions and a series of computed signals. This page defines each signal and says what a decision-maker does with it. These are not polls or predictions: simulated voices, to see the frictions coming. Deciding always happens under uncertainty; Kapari sheds light on the decision, it does not make it.

The composition

where each family's weight comes from

Where each family's weight in the panel comes from. Hub cases run in the 'by the question' mode: the families the decision names, or the documented frame anchors, weigh one and a half to two times the others. A composition choice, shown before the reading, never an opinion rate: weights are composition dials, not a measurement.

A composition weight, never an opinion rate: the test bench says where each family's weight comes from before saying who carries weight. The next signals are read with this composition in mind.

The dissenting voices

the profile that breaks with its group's line

On the simulated panel, a voice whose sentiment crosses its own group's line: its camp leans one way, it reports the other. The signal only fires on a clear gap, never on nuance.

This is where a camp's line is weakest: this voice shows where the group's support or refusal cracks on this panel.

Common ground

the friction opposing camps share

Two opposed camps of the panel (one won over, one lost) share the same dominant friction: they stumble on the same point for different reasons.

When supporters and opponents stumble on the same point, addressing it speaks to both camps at once.

The noise

who makes noise, who carries real weight

The computed cross of a group's weight in the composition and what its voices report they would do: a carrying bloc, an opposing bloc, dormant weight, or loud snipers without weight.

Volume is not influence: weight comes from each family's share of the panel, not from how loudly it reacts. A loud group without weight can saturate the debate without changing the outcome; a heavy bloc, even a quiet one, can carry or block the decision.

Internal vs External

the gap between internal and external voices

The gap between how internal voices (the house: employees, management, governing bodies) and external ones (the street: customers, public, partners) receive the decision, when each sphere is represented enough to be read.

A gap between internal and external signals that one message cannot serve both audiences.

The blind spot

the documented reaction no voice carried

The psychosocial lever documented in the knowledge frame for this type of decision that NO panel voice expresses: the simulation's blind spot.

The panel's blind spot: a reaction documented for this type of decision that nobody voices here. Either the panel is missing this voice, or the decision truly defused it: treat it as the first objection to test in the real world.

The yardstick

a verified benchmark, external to the simulation

A verified benchmark from a cited public source (statistics offices, studies), external to the simulation: the only number on the page that does not come from the panel.

The only number on this page that does not come from the panel: a real anchor for orders of magnitude.

Stability

does the response keep its step from one re-run to the next

The engine replays the simulation several times: if the response moves to another step from one re-run to the next, it is shown as split and reads as an open question.

The panel is replayed several times. Same step each time: the result holds. Neighbouring steps but tight positions: the bet lives at the boundary between two steps, read as the step shown under conditions. Scattered positions, or three steps crossed: a genuinely open question.

The tipping point

the light family that would change the response at twice its weight

Each light family of the panel is set to twice its weight, one at a time, and the response step is re-read: the families that change it are named. A response that hinges on a 10% family reads as a boundary bet, not a verdict.

The step shown depends on its composition weights. When one light family is enough to change it, the reading is a boundary bet: what you choose to weigh decides as much as what the panel says.

Why the process, and not the analysis alone

A 2010 McKinsey survey of 1,048 decisions, on outcomes self-reported by the executives themselves, finds that how the decision was made explains six times more of the outcome gap than the quality of the analysis. The question that best captured process quality was whether viewpoints contradicting the most senior leader had been discussed. That is what Kapari puts on the table, before the meeting.

Two limits, stated up front. These are self-reported outcomes, not measured by a third party. And Kapari does not measure the quality of your decisions: it surfaces plausible reactions, which is not the same thing.

The test bench for decisions. Simulated voices, never a measure of opinion.

Your next decision deserves the same scrutiny.

Run it through the test bench before you announce it: a panel of voices reacts, you read the range and you see the frictions coming.

Start free