Decision Making

A kirana store owner must decide by Friday whether to stock a new brand of atta. Waiting is also a choice, and it has a cost. Every manager's week is made of choices like this.

Peter Drucker put its weight plainly: whatever a manager does, he does through making decisions.

Decisions rest on three foundations:

• Logic: reasoning from what is known to what follows.
• Analysis: breaking the problem down and examining the facts and figures.
• Intuition: a judgement that comes from experience, faster than reasoning it out.

Back to the kirana store. Logic says that if customers keep asking for the brand, some will buy it. Analysis checks the numbers: how many asked last month, the margin per bag, and the shelf space it needs. Intuition is the owner's sense, from years at the counter, of whether her regulars will really switch.

Done well, decisions improve efficiency, effectiveness, adaptability, competitiveness and organisational performance.

Features. Decision making:

• is a mental, rational process;
• is selective, so there must be alternatives to choose from;
• is purposive, aimed at a goal;
• implies commitment until the decision is carried out;
• involves evaluating both the alternatives and the results;
• happens at every level;
• can be negative: a choice not to act is still a decision.

Don't confuse decision making with problem solving. Problem solving is the whole effort to find and close a gap. Decision making is the point inside it where one option is chosen. And with only one option there is no decision, since nothing is selected.

Where it breaks. Each foundation has a weakness. Logic is only as sound as its starting assumptions. Analysis needs data and time, which a manager may not have. Intuition is fast, but it can carry habits and biases from situations that no longer apply.

For exams. Know the definition, Drucker's line, the three foundations and the features, especially that a decision needs alternatives and that not acting is a decision.

Logic, analysis, intuition.

A bank branch manager approves a routine travel claim in a minute, then spends a month on whether to open a branch in a new district. Both are decisions, but they need very different handling.

Decisions can be sorted in four ways.

One public-sector bank, many types.

• Approving a travel claim within the rules: programmed, operational, routine.
• Opening branches in a new district: non-programmed, strategic, major.
• Turning that plan into a target and staff list for each new branch: tactical.
• Which teller covers the counter at lunch: operational, minor.
• A credit committee approving a large loan: a group, organisational decision.
• The branch manager choosing to take an evening course: personal.

The four ways of sorting cut across each other, so one decision carries several labels at once.

Don't confuse programmed with minor. Programmability is about whether a rule exists, not about the stakes. A bank's standard home-loan approval is programmed, yet each loan involves lakhs of rupees. A one-off choice of venue for a staff picnic is non-programmed, yet minor.

Where it breaks. The categories have blurred edges. A programmed decision becomes non-programmed when conditions change and the old rule no longer fits.

In practice. Turn recurring decisions into rules, so that they can be delegated. That frees senior managers' time for the non-programmed ones, which usually rise to the top of the organisation.

For exams. Expect a case to classify. Programmed against non-programmed is the pair asked most often.

Programmed or not. Strategic, tactical or operational.

A manager who picks the first option that comes to mind may be lucky, but cannot explain or repeat the choice. The rational process makes a decision traceable, step by step.

1. Define the problem and its urgency.
2. Gather information.
3. Identify the criteria.
4. Weight the criteria.
5. Develop alternatives.
6. Evaluate them against the criteria on feasibility, cost, risk and outcome.
7. Choose the best balance of benefit and risk.
8. Implement, by assigning responsibility and resources.
9. Monitor results and correct where needed.

Short form: identify, inform, alternatives, evaluate, select, implement, review.

Worked example. A Pune auto-parts firm must choose a site for a new warehouse. Its criteria and weights are rent (40%), distance to its main customers (35%) and labour available (25%). Each site is scored out of 10, where higher is better: a high rent score means low rent. The scores are illustrative.

Site B wins, although its rent is the highest. Site C has the lowest rent but comes last, because distance carries heavy weight. Then the firm implements: it names a project head and a budget. Six months later it reviews whether delivery times actually fell.

Don't confuse criteria with alternatives. Criteria are what you judge by: rent, distance, labour. Alternatives are what you choose between: sites A, B and C. Set the criteria before scoring, or the favourite option tends to shape them.

Where it breaks. The model assumes full information, clear goals and time to compare every option. Herbert Simon argued that real managers have bounded rationality: limited information, time and mental capacity. So they satisfice: they choose the first option that is good enough, rather than search for the best.

For exams. Know the order of the steps and the short form, and that the process ends with monitoring, not with the choice.

Go deeper for practitioners

Simon's three phases. Simon also described decision making as intelligence (searching for problems that call for a decision), design (developing possible courses of action) and choice (selecting one). The nine steps above fit inside these three.

Sensitivity. Before trusting a weighted score, change the weights a little and see whether the winner changes. In the example, weight rent 50%, distance 25% and labour 25%. Sites A and C then score 6.75 each and site B 6.50, so B drops to last. A decision that flips on a small change in weights needs more thought, or more data.

A decision. Your team favours site C before any scoring is done. Ask each member to set and weight the criteria privately first, then compare. It is harder for the favourite to bend the weights once they are written down.

Links. Planning, in Functions of Management, uses the same core steps: alternatives, evaluation, selection. Groupthink, the main risk when a group decides, is covered in Group Behaviour, Teams and Leadership (Unit 2).

A store owner choosing how much stock to buy, and a plant manager choosing which defects to fix first, face different kinds of decision. Each kind has a tool built for it.

The table pairs twelve common techniques with what each is best for.

Decision tree, worked. A kirana store plans a line of Diwali gift hampers. It can stock 200 or 100. The owner judges the chance of high demand at 0.6 and of low demand at 0.4. The payoffs are illustrative.

• Stock 200: EV = 0.6 × ₹60,000 + 0.4 × (−₹20,000) = ₹36,000 − ₹8,000 = ₹28,000.
• Stock 100: EV = 0.6 × ₹30,000 + 0.4 × ₹5,000 = ₹18,000 + ₹2,000 = ₹20,000.

Stocking 200 has the higher expected value. But it can lose ₹20,000, while stocking 100 never loses. An owner who cannot afford that loss may still choose 100.

Payback, worked. A chai stall buys a coffee machine for ₹1.2 lakh. It adds ₹4,000 a month in net cash. Payback = ₹1,20,000 ÷ ₹4,000 = 30 months, or two and a half years.

Pareto, applied. A bank branch sorts a year's customer complaints by cause. It may find that two causes, such as long queues and failed ATM withdrawals, account for most of them. Fixing those two first does the most good.

Don't confuse PEST with SWOT. PEST looks only outside the firm. SWOT looks inside (strengths and weaknesses) and outside (opportunities and threats). Nor the Delphi method with a meeting: Delphi experts never meet and answer anonymously, so no one voice dominates.

Where it breaks. A tool is only as good as its inputs: probabilities and scores are often guesses. Payback ignores everything earned after the investment is recovered, and treats a rupee next year as equal to a rupee today. The 80/20 split is a rule of thumb, not an exact law.

For exams. Match each technique to its use, and expect a short expected-value or payback calculation.

Match the tool to the decision.