Ian Hargreaves.
Rigour for strategic decisions
Section № 03 · Decision Tools
№ 03
Decision Tools

Decision
Tools.

Built for commercial decision-makers in consumer markets. Four working instruments for reasoning under uncertainty. Each takes the kind of question people usually argue about and turns it into something you can compute. Built deliberately to show how to ask a hard question well, not to dress up a confident answer.
№ 03.1 / Instrument

Conversion A/B

When you have the data, but a higher rate could just be luck.
A Bayesian comparison of two conversion rates. Treats each true rate as a range of plausible values, then asks how often one campaign genuinely beats the other. Computed exactly in the browser, no simulation.
Inputs: conversions and total leads, per campaign Outputs: chance B beats A, plus chance B wins by a meaningful margin Honest about uncertainty: shows where the two ranges overlap
Open the instrument →
№ 03.2 / Instrument

Elicit

When you don’t have data, you have judgement.
Three honest answers per option, low / typical / high, turned into distributions you can compare, no more certain than the answers you gave. Covers four common quantity types, from probabilities to skewed positives, and tells you when your three numbers don’t sit together in the shape you picked.
Inputs: low, typical, high values for two beliefs Outputs: chance A beats B, plus the believable range of each Fit quality: warns when judgement and shape don’t hang together
Open the instrument →
№ 03.3 / Instrument

Simulate

When the answer is a chain of uncertain numbers, not one.
Runs your back-of-envelope model thousands of times with every input varying inside its range (a Monte Carlo). Give each input a centre and a 95% range, from data or from Elicit, and read the full spread of outcomes, which input drives the risk, and the chance of missing your target.
Inputs: 2–5 variables, each a centre and a 95% range Outputs: spread of outcomes, which input moves the result most (the tornado), chance of missing target Pairs with Elicit: turn judgement into the ranges you feed in here
Open the instrument →
№ 03.5 / Instrument

Branch

When the decision comes before the uncertainty.
Draw the choice as a one-step tree: two or three options, each a gamble with its own odds and payoffs. Folds back to the expected value of every option, keeps the best, and shows the downside behind the average. Computed exactly in the browser, no simulation.
Inputs: two or three options, a probability and two payoffs each Outputs: expected value per option, and the best by fold-back Reads at a glance: the payoff and the risk in one tree
Open the instrument →
A note on these
Each is deliberately small. They demonstrate one method, end to end, rather than dress up a generic answer. Each has a plain-English companion: the A/B test pairs with Significant, and still wrong, Elicit with Elicitation methods, and Simulate with Simulate, in plain English. More instruments will appear here as the underlying methods get written up. If you find one useful, or surprising, or wrong, I’d like to know: tell me on LinkedIn.