Decision Trees

Expected value is the sum of each outcome's probability multiplied by its payoff (1), giving the average financial result expected from an option (1).

Expected value = (0.6 × 600,000) + (0.4 × 150,000) = 360,000 + 60,000 = £420,000 (Knowledge 1, Application 2). Net gain = 420,000 − 250,000 = £170,000 (Analysis 1).

A decision tree forces BuildRight to map out both expansion options, their possible outcomes and the probabilities in a structured way (Knowledge + Application). This disciplined layout makes BuildRight consider outcomes it might otherwise overlook — such as weak construction demand leaving new capacity idle — and makes its assumptions explicit, leading to a more thorough and transparent decision (Analysis).

The tree also puts a financial value (expected value and net gain) on each expansion option (Knowledge + Application). Quantifying the capital committed to each route lets BuildRight compare them objectively — choosing the higher net gain rather than relying on gut feeling (Analysis). Setting the two options side by side on net gain gives BuildRight a clearer, more rational basis for a high-stakes investment in extra capacity (Analysis).

On the numbers, Option A — the more aggressive capacity expansion — looks best. It has the highest net gain, meaning that on an expected-value basis — averaging all outcomes weighted by probability — it offers ExpandCo the greatest financial return on the capital it commits. If ExpandCo is comfortable with risk and can absorb setbacks, choosing the highest net gain is the logical, EV-maximising decision.

However, the 40% chance of a large loss is a serious concern. Expected value assumes ExpandCo is risk-neutral, but a 40% probability of a big loss is very high, and if that loss — sunk capital tied up in idle capacity and stranded overheads — is one ExpandCo could not survive, the average EV is misleading: the firm could be ruined by the downside. A risk-averse firm would rationally prefer a lower-net-gain but safer, smaller-scale option.

Whether ExpandCo should choose Option A therefore depends on its attitude to risk and its ability to absorb the loss. If ExpandCo is financially strong and could withstand the potential loss, choosing the highest-net-gain Option A is defensible. But if the large loss would threaten its survival, the 40% probability makes Option A too risky despite its higher expected value, and ExpandCo should choose a safer, more incremental option. So the decision shouldn't rest on net gain alone: ExpandCo must weigh the highest expected return against the risk of a loss it may not survive — and for most firms, a 40% chance of a large loss argues for caution rather than simply following the highest net gain.

A decision tree offers NovaLaunch a structured way to approach a risky product-launch decision — one that commits tooling, a full production run and a marketing budget up front — but its usefulness is limited by the reliability of its inputs.

Why it's useful. The tree forces NovaLaunch to map out its options, the possible outcomes (success/failure) and their probabilities, and to attach financial payoffs — turning a fuzzy launch decision into a structured, quantified comparison. This makes NovaLaunch's assumptions explicit, helps it weigh the downside (a flop that strands its launch investment and retooling costs) as well as the upside, and gives an objective net-gain basis for choosing, rather than relying on optimism. For a risky launch, this discipline is genuinely valuable.

Why its usefulness is limited. However, the tree is only as good as its estimated probabilities and payoffs. For a new product, NovaLaunch has little historical sales data to estimate the chance of market success or the payoffs reliably — these are essentially educated guesses, and if they are wrong, the recommended option could be wrong too. The tree also assumes NovaLaunch is risk-neutral (ignoring how a big potential loss might threaten it) and ignores qualitative factors — brand fit, strategic importance, competitor reaction — that matter for a launch.

Judgement. On balance, the decision tree is a useful structuring tool but not a reliable decision-maker for NovaLaunch — and arguably less reliable for a new-product launch than for a more predictable decision, because the probabilities are so uncertain. Its usefulness depends on how well NovaLaunch can estimate the probabilities and payoffs (better with market research and test-marketing), and on whether it treats the net-gain result as one input alongside qualitative judgement and its risk appetite, rather than the final answer. NovaLaunch should therefore use the tree to structure and inform its thinking — improving its estimates with research and testing different probabilities — but should not commit to the launch simply because the tree shows the highest net gain, given how uncertain the inputs are for a new product.

Decision trees give Pinnacle Group a structured, quantitative way to approach risky strategic decisions — major capital-allocation choices such as market entry, acquisitions or large-scale investment — but as a technique built on estimated probabilities and expected value, their usefulness depends on the quality of those estimates and how Pinnacle uses the results.

Why decision trees are useful. Trees turn complex, risky decisions into a clear, logical structure. They force Pinnacle to identify its options, map the possible outcomes and their probabilities, and attach financial payoffs — making assumptions explicit and ensuring the downside is considered, not just the hoped-for result. Calculating expected values and net gains lets Pinnacle compare competing investments objectively and allocate capital on a rational basis rather than instinct. The discipline of building the tree can itself improve decision-making by prompting thorough analysis. For a group making major strategic choices, this structure and quantification are valuable.

Their limitations. However, decision trees have serious weaknesses. Their results depend entirely on the estimated probabilities and payoffs, which are often subjective guesses — and a change in a probability can flip the recommended option, so a tree can give false confidence. Basing decisions on expected value assumes Pinnacle is risk-neutral: EV ignores the spread of outcomes, so it might recommend a high-EV option that carries a significant chance of catastrophic loss — a strategic bet that could impair the whole group — that a firm should avoid. Trees also ignore qualitative factors (strategic fit, brand, staff, ethics, competitor response) central to strategic decisions, and the data can be manipulated to justify a preferred choice.

What it depends on. How useful decision trees are for Pinnacle depends on several factors. It depends on the reliability of the probability and payoff estimates — good data (from research/experience) makes trees far more useful than guesswork. It depends on Pinnacle's attitude to risk — EV suits a risk-neutral firm, but a risk-averse one must look beyond the average at the downside exposure. It depends on the type of decision — trees suit decisions with clear, quantifiable outcomes, less so highly qualitative strategic ones. And it depends on whether Pinnacle treats the tree as the decision or as one input alongside judgement.

Conclusion. On balance, decision trees are a useful structuring and analytical tool but not a reliable stand-alone decision-maker for Pinnacle Group. Their value lies in imposing structure, forcing consideration of probabilities and outcomes, and quantifying options for comparison — genuinely helpful for major, risky capital-allocation decisions, so Pinnacle is right to use them. But because they rely on uncertain estimates, assume risk-neutrality, and ignore qualitative factors, Pinnacle should treat the net-gain result as one input, not the answer: it should ground the probabilities in the best available data, consider the downside and its risk appetite (not just EV), and weigh the numbers against strategic and qualitative factors. Used that way — as a disciplined framework to inform judgement — decision trees are highly useful; used as a mechanical rule that automatically picks the highest net gain, they can mislead. Their usefulness ultimately depends less on the technique than on the quality of the estimates and the wisdom with which Pinnacle combines them with judgement.