How to Write a Case Study? 9 Simple Steps, 22 Essential Questions, and Expert Tips for Better Results

Must read

Cynthia Cooper
Cynthia Cooper
Cynthia Cooper is an educational consultant based in Boston, Massachusetts. She holds a degree in Education from Harvard University and has extensive experience in curriculum development, teacher training, and educational leadership. Cynthia is known for her innovative approaches to education, her commitment to improving teaching practices, and her dedication to enhancing student learning outcomes. She works with schools and educators to implement effective educational strategies and foster a culture of continuous improvement.

A case study is a short, evidence-based account of one real situation: what happened, why it happened, and what it means for anyone facing the same thing. The nine steps below work for a university assignment and for a customer story on a company website, because both live or die on evidence you can point at. This guide covers how to write a case study from picking the subject to the interview questions that get you a quote worth printing. 

Which kind of case study are you writing? 

Two very different jobs share one name. That is why so much of the advice you find contradicts itself. Sort out which one you are doing before you touch the structure. It changes your evidence, your tone and your ending. 

 Academic assignment Business or customer story 
Who reads it A marker working through set criteria A buyer weighing you up against two rivals 
What it proves You can apply theory to a messy real situation A result you produced once can happen again 
Evidence it runs on Module reading, published research, the case material itself Interviews, dashboards, invoices, before and after figures 
Usual length 1,500 to 3,000 words 600 to 1,200 words 
How it ends Recommendations, then a reference list A headline result, a quote, and one next step 
What sinks it Description with no analysis Big claims with no numbers 

Key takeaways 

  • Settle the type first. Marking criteria and sales copy pull in opposite directions. 
  • One subject, one question. A piece that asks three things answers none of them properly. 
  • Gather the evidence before you draft, or you will start inventing it. 
  • The interview carries the whole piece. Weak questions get you “it was great”. 
  • Numbers beat adjectives. One before figure and one after figure do more than a page of praise. 

How to write a case study in nine steps? 

How to write a case study in nine steps 

Work in this order. Each step feeds the next one, and skipping the early ones is exactly how a draft ends up summarising a situation instead of arguing about it. 

  1. Pick one subject and one question. Not “our work with Halifax Joinery”, but “why did their order backlog halve after they changed one process?” The question keeps you honest for the next eight steps. 
  1. Read what you are being judged on. For an assignment, that means the learning outcomes and the marking rubric, printed and sitting next to you. For a business piece, it means the one objection this story has to answer. 
  1. Collect evidence before you write a word: emails, meeting notes, spreadsheets, screenshots, module readings. Put every source in one folder and name the files properly. 
  1. Interview the people who were actually there. Twenty minutes with the person who did the work beats a fortnight of guessing at their reasoning. 
  1. Build a timeline—before state, trigger, decision, action, outcome, each with a date. Gaps in the timeline are usually gaps in your evidence. 
  1. Choose your frame. An assignment needs a named model or theory and a sentence saying why that one. A business piece needs a single headline metric instead. Where the story turns on how an offering performed over its life, the vocabulary of product service management gives you a frame the reader already knows. 
  1. Draft the middle first. Write the analysis or the results while the evidence is fresh. Openings written early get rewritten anyway. 
  1. Write the opening and the ending last. Only now do you know what the piece actually says, so only now can you promise it in the first line. 
  1. Get it checked and signed off.—atutor, a colleague, or the client whose name is on it. Fix the flabby sentences on the same pass. 

A reusable case study structure 

This skeleton fits both versions. 

Watch the share of the word count, because that is where most drafts come apart. People spend a third of the piece setting the scene, then have four hundred words left for the part that earns the marks. 

Section What it has to do Share of the words 
Title and one-line summary Name the subject and the outcome, so a skim-reader gets both 5% 
Context Who, where, when, and the state of play before anything moved 15% 
The problem The specific issue, sized with a figure rather than an adjective 15% 
Approach What was tried, by whom, and why that rather than the obvious alternative 20% 
Analysis or results What changed, measured, with the awkward parts left in 30% 
Conclusion What it means and what you would do again 10% 
References or next step Your sources, or the single action you want from the reader 5% 

Academic versions fatten the analysis section and close with a reference list in whichever style your department asks for. The business version fattens the results and closes with a next step. 

Same bones, different muscle. 

The interview questions that produce usable quotes 

The interview questions that produce usable quotes 

Almost every guide tells you to interview someone and then leaves you to it. The questions are the whole game. Closed questions get you a yes. An open question about one specific moment gets you a sentence you can print without editing it into meaninglessness. 

Record it, with permission. Ask for a number early and ask again at the end, because people remember better once they have been talking for a while. Interviewing colleagues rather than a client brings its own etiquette, and agreed team norms about who speaks for the group save an awkward conversation later. 

Before: what the situation really looked like 

  • Walk me through a typical week before any of this changed. 
  • What was the moment you decided the old way had to go? 
  • How much time or money was the problem costing you a month? 
  • Who else in the organisation was feeling it, and how? 
  • What had you already tried that did not work? 

During: what changed, and who did it 

  • What did the first week of the change look like day to day? 
  • Who had to be persuaded, and what convinced them? 
  • What nearly went wrong? 
  • What would you have done differently starting again? 
  • How long before anyone noticed a difference? 

After: the results, in numbers 

  • What is the figure now, against the figure we started with? 
  • Which measure improved that you were not even tracking? 
  • What can your team do now that it could not do in January? 
  • What did this free up, in hours or in budget? 
  • If the change vanished tomorrow, what breaks first? 

Quote-getters 

  • What would you say to someone sitting where you sat a year ago? 
  • What surprised you most? 
  • Finish this sentence: the thing nobody tells you about this is… 
  • What is the one detail you would want a sceptic to hear? 

Closing 

  • Is there anything I have not asked that matters? 
  • Who else should I speak to? 
  • Are you happy for me to quote you by name and job title? 

When the numbers are awkward 

When the numbers are awkward 

Mixed results are not a problem. They are a credibility asset, because a piece where everything improved reads like an advert. Report the measure that moved, name the one that did not, and say what you think explains the difference. A marker rewards that sort of honesty, and so does a buyer who has read forty glossy write-ups this quarter. 

Sometimes the client will not release a figure at all. Ask for a percentage instead of an absolute, or a ratio, or a proxy: tickets closed per person, days to first response, repeat order rate. One joinery firm would not share turnover, yet was perfectly happy to say its returns handling cost fell from £4.10 to £2.60 an order. 

Then get sign-off in writing before anything goes live. Send the draft, the exact quotes and the figures you plan to publish, and ask for a yes by email. Skipping that is the mistake that ends a client relationship rather than a paragraph. 

Mistakes that cost marks or credibility 

  • Describing instead of analysing. The marker knows what happened. They want to know why it happened, and what it shows. 
  • Naming a theory once and abandoning it. If you open with a model, the analysis has to keep using it. 
  • Quoting praise instead of detail. “They were brilliant” proves nothing. “We cut the handover from four days to one” proves something. 
  • Writing the introduction first and never revisiting it. It will promise the wrong article. 
  • Hiding the number. The headline figure belongs in the title and the first paragraph, not buried on page three. 

Your next step 

Open a blank document and write one sentence describing the subject and the outcome. Then book the interview. Everything on this page follows from those two moves. Finished pieces double as portfolio material, and a folder of them says far more about your career trajectory than a list of duties on a CV ever will. 

Case study FAQs 

How long should it be? 

An assignment usually runs 1,500 to 3,000 words unless the brief says otherwise, and the brief wins every time. A customer story on a website works best at 600 to 1,200 words. Your reader is scanning for a result and a name they recognise, not settling in for an essay. 

Can I write about a project that went badly? 

Yes, and those often score highest. An honest account of a failure, with a clear read on the cause, shows more analytical skill than a success story where nothing was ever in doubt. For a client-facing piece, get their agreement first. 

Do I need permission to name the organisation? 

Yes, for anything you publish. On an assignment, follow your department’s ethics guidance and anonymise where it asks you to, which usually means swapping real names for a letter or a role title. 

What if I cannot interview anyone? 

Use documents. Meeting minutes, published reports, support tickets and internal messages all carry evidence, and a well-sourced desk study beats a thin interview. Nobody masters how to write a case study from a template alone, so treat the first two as practice and expect the third to come out sharper.

More articles

Latest article