A teaching case, often just called a case study, is a written account of a real or realistic organisation at a moment of choice. It gives you a situation, the people involved, some numbers and usually more detail than you need, and then it stops before anyone decides. The protagonist is the person through whose eyes the case is told; the decision-maker is the person who must choose, and they are not always the same: in Kettle & Crumb's case Dan analyses, but Maria decides. Business schools use cases because they let you practise judgement safely: you weigh incomplete information, commit to a view and defend it, as a manager would, without the cost of getting it wrong in a real business.

Ellet (2018) groups cases by the task they set. A problem case asks you to diagnose why something is happening, such as falling sales. A decision case asks you to choose between options. An evaluation case asks you to judge how well something worked, such as a campaign or a strategy. Spotting the type early tells you what your answer has to deliver.

The most common mistake new students make is to describe the case rather than analyse it. Description retells what the case says, usually in the order the case says it. Analysis selects the facts that matter, connects them, explains what they mean for the decision and judges which matter most. At postgraduate level, markers give few marks for retelling: they assume you have read the case and want to see what you made of it.

A useful habit is to read a case twice. The first time, read it straight through to get the story. The second time, sort what you read into three piles: facts you could check, opinions held by people in the case, and gaps where information you need is missing. Opinions are evidence of what someone believes, not of what is true, and gaps become assumptions you must label as assumptions. Finally, remember that most cases have no single right answer. You are marked on the quality of your reasoning, so a well-argued recommendation for the less popular option can outscore a weak argument for the obvious one.

A process stops you jumping straight to a favourite answer. This module uses six steps. Situation: summarise what is going on, briefly. Problem or decision: say what actually has to be solved or decided. Analysis: use evidence and simple lenses to understand the causes and the forces at work. Options: set out the realistic choices. Criteria: agree what a good choice must achieve. Recommendation: choose, say why, and say what would need to be true for it to work. In practice you loop back, because analysis often reveals that the first problem statement was wrong, but the order keeps you honest.

The second step matters most. Cases are full of symptoms: sales are flat, staff are stretched, a competitor has opened nearby. A symptom is what you can see; a cause is why it is happening. Keep asking "why?" until you reach something the organisation can act on. Then write a decision question: one sentence that says who must decide what, by when, and to achieve what. A good decision question rules things out. "How can we grow?" invites a list of ideas; "Which of two investments should we commit to, to grow over the next three years?" invites a choice.

Options should be genuinely different, realistic and few, usually two or three. Always include a baseline of carrying on as now, because every option has to beat it. If two options could be combined, say so, and treat the combination as a further option rather than pretending they exclude each other.

Criteria are the tests a good option must pass, such as up-front cost, risk if things go wrong, fit with what the organisation already does well, or time until it pays back. Agree them before you judge the options. Otherwise it is tempting to pick the option you like and then choose criteria that flatter it. For each criterion, note how you would know whether an option meets it; that note tells you what evidence you will need, which is where Lesson 3.4 picks up.

Once you know the decision, you need to understand the situation behind it. Analysis lenses help: they are sets of prompts that make sure you look in the right places. Used well, they surface what matters. Used badly, they produce long lists that look like analysis but say nothing about the decision.

Start with the business model you met in Module 1. Ask which building blocks each option would change. An option that adds a new customer segment or revenue stream changes the business more than one that adds another outlet of the same kind, and the blocks that change are where the risks and the evidence needs sit.

SWOT sorts what you find into four boxes. Strengths and weaknesses are internal: things the organisation has or does, such as skills, capacity, reputation or finances. Opportunities and threats are external: things happening around it, such as customer trends, competitors or a new partner. A common error is to put an internal fact in an external box, so check each entry with the question "is this about us, or about the world around us?" Note too that the same fact can be a strength for one option and a weakness for another.

PESTLE is a prompt for the external environment: political, economic, social, technological, legal and environmental factors. For a large organisation in many markets, all six may matter. For a single decision in a small business, usually only two or three do, such as the local economy, a shift in what customers value, or a change in the rules on food labelling. Pick those and leave the rest out. Deeper use of these tools belongs to a strategy course; here they are first-pass prompts.

Every lens needs the "so what?" test. After each point, ask "so what does this mean for the decision?" and write the answer in one sentence: which option it favours, what risk it raises, or what you now need to check. A SWOT with a "so what?" after every entry is analysis. A SWOT without one is a list, and a list is where analysis starts, not where it ends.

Evidence is any information you use to support a claim, and it is only as good as its source. Primary evidence is collected first-hand for the question at hand: a survey you run, interviews, your own sales records, a count you make yourself. Secondary evidence was collected by someone else for another purpose: published reports, articles, official statistics. Primary evidence fits your question closely but is often small-scale; secondary evidence is often larger and more expert but was built for someone else's question.

Secondary sources come in broad families. Academic sources, such as peer-reviewed journal articles, have been checked by other experts before publication. Official sources, such as government statistics, follow published methods. Trade sources, such as industry associations and trade magazines, know a sector well but may speak for it. Grey sources, such as consultancy reports, company white papers and blogs, vary widely and need the most care.

The CRAAP test, developed by Sarah Blakeslee at the Meriam Library, California State University, Chico (Blakeslee, 2004), gives you five questions to ask of any source. Currency: when was it produced, and is that recent enough for your question? Relevance: does it address your question, market and customers? Authority: who produced it, and are they qualified to? Accuracy: is it supported by evidence, with a method you can see and figures you could check? Purpose: why does it exist: to inform, to persuade or to sell? This module adds a sixth: fit to my question, because a good source on the wrong question is still the wrong source.

For surveys, also ask who answered, how many, how they were chosen and what share of those asked replied. A survey of your most loyal customers can be accurate about them and still mislead about everyone else; this is sample bias. Watch for claims without evidence too. A figure with no source, no period and no base ("up 30%" from what, since when, and where?) cannot be checked, so it cannot carry weight, however confident it sounds.

The same habits apply to answers from generative AI tools. They can write fluently, summarise wrongly and give references that do not exist or do not say what is claimed. Treat an AI answer as a lead, not a source: trace each fact you want to use to the original publication, read it there, and cite that. Rules on using AI in assessed work differ between institutions and change often, so always follow your own institution's current policy. Scoring sources is a judgement, not a calculation, so record a reason beside every score.

An argument is not an opinion stated firmly. It is a claim supported by evidence, with reasoning that shows why the evidence supports the claim. Toulmin (1958) set out a fuller model of how practical arguments work, with six parts: the claim, the data offered for it, the warrant that licenses the step from data to claim, backing for the warrant, a qualifier showing how strong the claim is, and rebuttals stating when it would not hold. This module uses a simplification of our own: claim, evidence and reasoning, plus an honest statement of limits.

The claim is the point you want your reader to accept. The evidence is what you offer in support: figures, findings, quotations, observations. The reasoning is the link, and it is the part beginners most often leave out. Placing a statistic next to a claim does not show that it supports the claim; you have to say why it does. A useful check is to read your reasoning sentence on its own and ask whether a sceptical reader would agree that, if the evidence is true, the claim follows.

Real cases contain conflicting evidence. Weigh it by quality, not by quantity: one well-collected, relevant source can outweigh three weak ones, which is why you scored your sources in Lesson 3.4 before arguing. Then face the strongest objection to your claim, not the weakest. Say what the counter-evidence is, and explain why your claim still stands or how you have narrowed it to survive. An argument that ignores obvious counter-evidence loses the reader's trust, even if its conclusion is right.

Finally, state your limits. Which evidence is thin, which figures are assumptions, and what would change your mind? Use proportionate language: "the survey suggests", "on this evidence", "if the footfall pattern holds". Words such as "proves" and "clearly" promise more than a business case can usually deliver. Stating limits is not weakness. It shows a marker, or a managing director, that you know exactly how far your evidence goes, and it tells them what to check next.