Should you act now,
or gather more evidence first?

The Decision

Every investigation, pilot, audit, or experiment is an implicit VOI calculation: the information is worth gathering if and only if the expected improvement in the decision exceeds the cost of gathering it. Most organizations make this calculation qualitatively — "let’s do a pilot" — without computing whether the pilot changes anything.

A causal model makes the calculation explicit. Given the current probability distribution over outcomes, and given a specific piece of evidence that could be gathered, the VOI is the difference between the expected utility of the optimal decision before and after that evidence is obtained. If the VOI is less than the cost of the investigation, act now on current knowledge. If it is greater, investigate first.

The VOI framework also identifies which uncertainty is most worth resolving — not the largest uncertainty, but the uncertainty whose resolution most changes the optimal action. A variable that is highly uncertain but whose value does not affect which decision is best has zero VOI. A variable that is moderately uncertain but whose true value would switch the optimal action has high VOI.

How It Is Computed

For a discrete causal model, VOI is computed by comparing two expected utilities:

1. Expected utility of acting now — the utility of the optimal action given the current posterior over the unknown variable, weighted by the probability of each value.
2. Expected utility of acting after observing — for each possible value the unknown variable could take, the utility of the optimal action given that value, weighted by the probability of each value.

VOI = (2) − (1). If this exceeds the cost of gathering the evidence, gather it. The Bayesian network propagates the updated probabilities automatically — the VOI calculation is a standard inference query on the existing model, not a separate analysis.

Sequential VOI — applied iteratively after each piece of evidence arrives — is the formal basis for adaptive diagnostic sequences. See Diagnostic Reasoning.

In the Cases

Supply Chain: Before investing $5M in dual sourcing, the VOI calculation asks whether the $5M decision changes if the second supplier shares upstream dependencies with the first. If the answer is yes regardless — if the investment is correct under both scenarios — the investigation has near-zero VOI and the decision should proceed.

Climate / ESG: VOI analysis across the model identifies which adaptation investments — flood barriers, building upgrades, supply chain diversification — are robust across the range of plausible climate scenarios versus which ones only pay off under specific assumptions. The VOI of resolving the climate scenario uncertainty is highest for the investments that switch between correct and incorrect depending on which scenario obtains.

NIST CSF: The $4.2M allocation between Protect and Detect is exactly a VOI question — how much is the optimal allocation decision worth changing if the organization’s threat profile turns out to be different from the current estimate?