Marketing case studies: how to read one properly, and how to write one that holds up

A marketing case study is the record of one decision and what followed it: the situation, the constraint, the choice that was made, what was done, what changed, and how the change was measured. The useful ones always name three things, the baseline before the work, the exact time period, and the method used to attribute the result. The useless ones give a percentage with no baseline and no dates. Read every case study with the five questions below, and write yours so that a sceptical reader can answer all five without asking you.

This matters in two directions. You will read case studies to decide what to try and who to hire, and you will publish them to win work. Both go better when the standard is the same.

The five questions

1. What was the baseline? A rise of sixty percent is meaningless without the starting number. Sixty percent of a small number is a small number.
2. Over what period, and compared with what? Against the previous month, the same month last year, or a control group that did not get the campaign? Seasonality explains more marketing results than marketing does.
3. How was the result attributed? Last click in analytics, a holdout region, a brand survey, a coupon code? The method decides how much of the result belongs to the work described.
4. What else changed at the same time? A new product, a price change, a competitor leaving, a television campaign running in parallel. A case study that mentions none of these is describing a laboratory that does not exist.
5. Who wrote it and what do they sell? Not a disqualification, just a weighting. An agency case study about the agency's own service deserves the same reading as any other marketing material.

Why the famous campaigns are poor teachers

Everybody has read about the same handful of campaigns, and they are famous precisely because they were unusual. Dove's Real Beauty work, Old Spice's response videos, the ALS Ice Bucket Challenge and Spotify Wrapped are genuinely interesting, and they are also the surviving members of a very large sample. Thousands of campaigns with similar logic went nowhere and were never written up. Studying only the winners teaches you what winning looked like, not what caused it.

The practical correction is to prefer case studies from businesses that look like yours, in your category, at your size, with your budget. A regional distributor learns more from another regional distributor's channel decision than from a global brand's campaign film.

Where to find case studies that behave like evidence

Source	Strongest use	Read it for
Awards archives and effectiveness papers	Strategy and creative reasoning	The written method and the measurement section
Platform help centres and product documentation	Mechanics of a channel	Setup detail, not the headline result
Public company reports and investor decks	Category level spend and outcome	Audited figures and segment commentary
Academic marketing journals	Repeated findings across many firms	Sample size and controls
Agency and vendor case studies	What a provider actually does week to week	Scope, team, timeline and the five questions
Your own past campaigns	Decisions for your exact business	Everything, if you wrote it down at the time

That last row is the one most teams neglect. The best marketing case studies available to you are the ones about your own business, and they only exist if somebody records the baseline before the campaign starts.

How to write one that stands up

1. Get permission first, in writing, naming exactly which figures may be published. Doing this after the draft wastes the draft.
2. Choose a representative client, not your best one. A typical result that a reader can picture themselves getting is more persuasive than an outlier they will discount.
3. Write the baseline before anything else. If you cannot find the baseline, you do not have a case study, you have a testimonial, which is a fine thing to publish under its own name.
4. State the period in dates, start and end, and say what the comparison period was.
5. Name the constraint. Budget, headcount, a legacy platform, a regulated category. The constraint is the part readers find credible and relate to.
6. Describe the work in enough detail to be copied. Vagueness reads as having nothing to describe.
7. Show the measurement method, including what you could not separate out. Saying that a price change ran in the same quarter raises your credibility rather than lowering it.
8. Include one thing you would do differently. It is the fastest way to be believed on everything else.
9. Date the page and keep the date current when you revise it.

The measurement section, done properly

Most case studies compare the campaign period with the period before it and stop. Three cheap improvements make the result far more defensible:

• Hold something back. Run the campaign in some regions, stores or segments and not others, then compare. This is the closest a normal business gets to a controlled test.
• Use a longer pre period. Twelve months of history shows the seasonal shape, so you can tell a campaign effect from the usual spring.
• Name the one metric that decided it before the campaign launches, and report that metric afterwards whichever way it went. Choosing the metric after seeing the data is how case studies lose their value.

A template you can fill in

1. Client and category          who they are, what they sell, where
2. Situation                    the business problem, with the baseline number
3. Constraint                   budget, team, platform, regulation
4. Decision                     what we chose to do and what we chose not to do
5. Execution                    channels, assets, timeline with dates
6. Measurement method           comparison period, control group, attribution
7. Result                       the agreed metric, before and after
8. What else was happening      price, product, competitors, season
9. What we would change         one honest paragraph
10. Approvals and date          who signed off, when published

This page covers reading and writing case studies as evidence. For planning the campaign itself, a marketing plan template is the right companion document, and the measurement section above is the part to agree before anything launches.

Questions buyers ask about marketing case studies

How long should a marketing case study be?

Long enough to answer the five questions, which is usually eight hundred to fifteen hundred words. Add a short summary box at the top with the situation, the action and the result, because many readers will only read that.

Can I publish one if the client will not be named?

Yes. Describe them by category, size and region, say plainly that the name is withheld at their request, and keep every other detail specific. An anonymous case study with real method beats a named one with none.

Should percentages or absolute numbers be used?

Both. A percentage alone hides the baseline, and an absolute number alone hides the scale of the change. Giving the pair removes the suspicion that the figure was chosen to flatter.

How many case studies does a website need?

Enough that a prospect finds one resembling their own situation, which usually means coverage across your main categories and sizes rather than a large total. Three precise ones outperform fifteen generic ones.

Do case studies help with search?

They do when they are written as pages about a problem rather than pages about you. Title them with the situation and the category, because that is what somebody searches for, and the client name is what nobody searches for.

What makes a reader distrust one instantly?

A result with no date, a chart with no axis labels, and language that claims certainty about something the method could not have isolated. Removing those three things is most of the job.

A case study read by something that skips the template

Your five question structure suits a human reader. An assistant lifts the one sentence it needs: the client, the number, the method. Make those sentences stand alone, each with its context attached, because the rest may not travel with them. Then ask an assistant for examples in your field and see whether yours appears. If a rival's case study is the one named, read it as a buyer would. AI Knows Us records which firms and which pages come up for such questions.