(a) Surprise: What surprises will I receive from the information; and how often will I receive these surprises?

(b) Change: What decisions will I change for the better from this surprise; and how often will this occur?

(c) Payoff: What will be the payoff from making these improved decisions?

(using statistics & probability theory)

Value of Information = Expected Value of Information - Cost

Use normative decision analysis tools, incorporating expected value and Baye's theorem to judge conditional probabilities, etc.

Assume the information you have is perfect: 100% accurate and unbiased; thus you will get an optimal result

Information supports decisions; decisions trigger actions; actions have measurable results (e.g., profit, response time, accuracy of actions, etc.) Use these performance or output measurements as a surrogate for information value. Thus we're measuring information value empirically by its impact on performance.

Note: This measurement doesn't require people to be rational, as it only measures results. This, e.g., allows people to satisfice instead of coming up with the optimal solution.

How? Through experiments or prototypes, and measure their results.

Assumes:

- you can measure the results (e.g., post performance review),

- you know the cost of the information,
- experiments approximate reality; the prototype will scale up

a priori or ex post "guesstimate" based on your past experiences (if any):

---> subjective values: multi-criteria decision analysis, elimination by aspects, ranking alternatives subjectively; through what-if analysis; by asking: how much would you be willing to pay for this information or result