Reading Brand Advertising Case Studies

Updated September 2026 · Written and maintained by the Progression Agency strategy team

Brand advertising case studies are the most-read and least-useful genre in marketing writing. They describe what a company did, attribute its success to that, and omit everything that cannot be seen from outside — the budget, the distribution deal, the timing, the failures that preceded it. This page does something narrower and more honest: it separates what is genuinely observable about a few widely-studied brands from what is inferred, explains which mechanisms transfer to a business without their resources, and gives you a method for reading any case study without drawing the wrong conclusion from it.

Read the constraint, not the result

An award case study states what was achieved and rarely what was forbidden. The transferable part is the limitation the team worked inside, because that is what makes the solution a solution.

Check who commissioned the evidence

Results reported by the agency that produced the work carry a different weight from results verified by an awarding body that required a control or a baseline.

Reading advertising case studies

The most-published format in marketing, and the small number that teach anything.

Best advertising case studies, as a category, are dominated by award submissions, which are themselves marketing documents written by the agency that made the work. That does not make them useless; it means reading them for the constraint rather than for the result. An ad case study worth its length states what the business problem was, what was tried, and what the alternative would have been.

Marketing strategy case studies from the effectiveness literature — the awards and journals that require submitted results to be evidenced — are the more reliable source, because the entry requirements force a control or a baseline. Marketing cases used in teaching are better still on that count, since the failures are included.

Marketing success stories published by platforms and vendors are the weakest tier, because the publisher selects which stories to tell and the counterfactual is never present. They are worth reading for the mechanics of how something was executed and worth discounting entirely on whether it worked.

Field studies and usability questions

Two research methods that answer different questions, and are frequently substituted for each other.

What is field study research: observing people in the setting where they actually do the thing, rather than in a lab or on a call. A field study surfaces the context that participants never think to mention — the interruptions, the workarounds, the second screen, the colleague they ask. It is the most expensive research method per participant and routinely the only one that finds the real constraint.

Field study examples in a commercial setting: watching a warehouse team use a stock system during a shift, sitting with a clinician through a clinic session, riding along with a service technician for a day. Types of field studies range from pure observation, where the researcher does not intervene, through contextual enquiry, where they observe and ask as work happens, to diary studies, where participants record their own experience over weeks.

Usability study questions are a different instrument entirely, used when something already exists and the question is whether people can use it. The rule that produces usable answers is to ask about the task rather than the interface: what would you do next, what do you expect to happen, what were you looking for. A usability question phrased as do you like this produces an opinion, which is not what the session is for.

UserTesting questions and the equivalent on any remote platform have one additional constraint: nobody is in the room to probe, so each prompt has to stand alone and each task has to be unambiguous. Writing them well means piloting with two people before spending the panel budget, because a prompt that can be misread will be misread by everyone.

Louis Vuitton marketing is worth setting beside the other examples here because it demonstrates a strategy most brands cannot use and can still learn from. Marketing Louis Vuitton, or any comparable maison, depends on scarcity maintained deliberately: distribution is controlled, the product is rarely discounted, and the brand’s presence in culture is bought through patronage, collaboration and spectacle rather than through performance advertising. The observable pattern is that the advertising sells the world rather than the object, and the object is then sold in an environment the brand fully controls. The transferable lesson is narrow but real: where a brand can control its distribution, discounting becomes optional, and a brand that never discounts is making a promise about value that customers can verify over years.

Two more worked examples: Gucci customers and Warby Parker

Gucci’s target market is instructive because the visible advertising and the revenue base are aimed at different people. Gucci target customers, in the brand’s own reported terms, skew substantially younger than luxury historically did, while much of the volume sits in accessible entry products rather than ready-to-wear. Gucci technology investment has followed the same logic — digital flagship experiences, in-app try-on and gaming partnerships aimed at an audience that will not buy a runway piece this year and may in ten. Warby Parker branding is the counter-example worth setting beside it: a category where the incumbent’s pricing was the entire story, addressed by a brand built on a single legible claim about price and access, with the home try-on program functioning as both a distribution mechanism and the proof of the claim. One brand sells aspiration at the top and monetizes at the bottom; the other collapses the ladder entirely. Both work, and the reason each works is structural rather than aesthetic.

Reading a luxury brand target market: the Gucci example

Luxury advertising is frequently misread because the visible campaign is not aimed at the person who buys most of the volume. The Gucci target market, as the brand’s own reporting and industry analysis describe it, skews substantially younger than luxury historically did — the house’s returns since the mid-2010s have been attributed publicly to millennial and Gen Z demand — while the revenue mix still leans heavily on accessible entry products rather than ready-to-wear. That combination explains the advertising: the imagery signals to an aspirational audience that will buy a small leather good or a fragrance, while the runway work maintains the credibility that makes the entry product worth its price. The transferable lesson is that in categories with a wide price ladder, brand advertising is usually addressed to the top of the ladder and monetized at the bottom, and judging such a campaign against direct-response metrics misses what it is built to do.

Why brand case studies mislead

Because they present observable behavior and inferred causation together, while omitting everything that would let you test whether the one caused the other. Budget, distribution agreements, timing, and the campaigns that failed are all invisible from outside, and all four frequently matter more than the strategy being described.

The scorecard splits cleanly: everything you can observe is about behavior, and everything you cannot observe is about causation. A case study that draws a causal conclusion is therefore reasoning from the half of the evidence it does not have.

Survivorship is the worst of the failures

Only successes get written up. For every documented company that grew using a particular tactic, there are undocumented companies that used the same tactic and did not, and nobody writes those up. This makes any tactic look more reliable than it is, and it is invisible in the individual article.

Authorship matters more than quality

A large share of published case studies are written by the agency that ran the work. That does not make them false and it does make them sales documents, selected for outcome and framed for credit. Check the byline before the argument.

Tactics expire, structures do not

A channel arbitrage that worked in 2016 does not work now, because the arbitrage closed. A structural choice — removing steps from a purchase, owning your customer data — is as true now as it was then. When reading anything, ask which of the two you are looking at.

A method for reading any case study

Six questions: what is observable, what is inferred, what is invisible, who wrote it, what is the counterfactual, and what is the mechanism stripped of the brand name. The last question is the only one whose answer you can use.

Gucci advertising: what is observable

Gucci is a luxury house whose advertising is publicly visible, heavily art-directed and distinctive, and whose brand assets have been maintained over a very long period. Those three things are observable from outside. The budget, the media agreements and the internal results are not, and this page does not state them.

What is genuinely instructive about a luxury house is not the creative work, which most businesses could not use and could not afford. It is the discipline: recognizable assets applied consistently over decades, scarcity maintained deliberately rather than discounted away, and distribution controlled tightly. All three are decisions, and two of the three are available at any size.

What transfers from luxury advertising

Consistency and restraint. A small business that uses the same typeface, the same colors and the same tone for five years accumulates recognition that a business rebranding every eighteen months never will. That is the cheapest advantage in marketing and almost nobody takes it, because consistency is boring to the people producing it long before it is boring to the audience.

What does not transfer

The scale, the heritage, the celebrity association and the media budget. A campaign built on fame reaching more fame does not scale down; it simply becomes expensive photography. Reading a luxury case study for tactics is the clearest example of the error this page is about.

Warby Parker advertising: the mechanism worth copying

The observable mechanism is friction removal. Buying prescription eyewear traditionally required a shop visit, and the company built a purchase process that did not. Whatever else is true, that structural choice is visible from outside and it is the part that transfers.

It is worth being precise about what is being claimed. We are describing a publicly observable business model, not asserting a causal explanation of the company’s results, which would require data nobody outside it has. The transferable observation is narrow and useful: the company identified the single step that stopped people buying and removed it.

Ask what your equivalent step is

Every business has one — a form that is too long, a quote that takes three days, a delivery date that is not shown, a returns policy that is hidden, a phone number nobody answers after five. Finding and removing it costs nothing and it is the closest thing to a universal lesson in this entire genre.

Owning the customer relationship

Selling directly rather than exclusively through intermediaries means holding the customer data, the email relationship and the ability to talk to buyers again. That is a structural advantage available to almost any business, and it compounds in a way advertising does not.

Apostrophe and the direct-to-consumer health model

Searches for apostrophe skincare are mostly people researching the service rather than studying its marketing, and it is worth answering that intent honestly: it is a telehealth-style prescription skincare service, and questions about suitability, prescriptions and pricing belong to the company and to a clinician rather than to a marketing page.

The marketing-relevant observation is structural again. Prescription categories carry regulatory constraints on claims, which forces the marketing toward education rather than assertion — explaining what a condition is, what treatment involves and what to expect. That constraint turns out to be an advantage, because educational content is exactly what people search for.

Regulated categories: the constraint that helps

When you cannot make strong claims, you have to be useful instead. Businesses in unregulated categories can assert things and frequently do, which produces marketing nobody trusts. The discipline of being accurate is worth adopting whether or not a regulator requires it.

CPG brands: why the category is studied so often

Consumer packaged goods are studied because the category is enormous, the products are familiar, and the marketing is visible to everybody. It is also the category whose lessons transfer worst, because CPG economics depend on shelf distribution, trade spend and scale that almost no reader of a case study has.

People searching cpg brands are usually doing one of three things: looking for a list of companies, researching the industry as a job market, or studying its marketing. All three are legitimate and only the third is served by a case study, which is worth knowing before you write one.

What CPG actually teaches

Distribution beats advertising. A product on the shelf at eye level in ten thousand stores outsells a better product nobody stocks, and much of what reads as brilliant CPG marketing is the visible surface of a distribution position. For a small business the equivalent question is where your product or service is encountered, not what your advertisement says.

Why CPG tactics invert at small scale

Mass awareness advertising works when a small percentage shift across a huge audience pays for it. At small scale the same spend reaches too few people to produce that shift, which is why brand-awareness campaigns copied from CPG reliably disappoint smaller businesses. The mechanism requires the scale; it is not scale-independent.

The right-hand column is the one worth reading twice. Each of these is genuinely instructive about one narrow thing and routinely cited as evidence for something much broader.

Which mechanisms actually transfer?

Removing purchase friction, owning the customer relationship, staying consistent for years, and answering buying questions publicly. All four are structural, all four are cheap, and all four are copied far less often than the tactics that do not transfer.

The quadrant chart is the practical summary of this page. Everything on the left is a structural choice available at any size and moves results substantially. Everything on the right belongs to that company’s circumstances. Most case-study writing dwells on the right-hand side because it is more interesting to read.

How to verify a claim in a case study

Look for a primary source, check whether figures are attributed, check the date, ask who benefits from the framing, and check whether the result has been repeated. Unattributed numbers travel fastest and are the least reliable.

The right-hand column is not cynicism, it is the base rate. Applying these tests to a well-known case study is a useful exercise precisely because the tests usually fail, which recalibrates how much weight the genre deserves.

Direct-to-consumer: the model behind several of these

Answer first: selling directly rather than through intermediaries changes who owns the customer relationship, who sets the price, and who carries the acquisition cost. It is a distribution decision that gets written up as a marketing story.

Why the acquisition cost question decides everything

Answer first: a direct model shifts the cost of finding customers from the retailer to you. That works when repeat purchase or lifetime value is high and fails when it is not, which is why the same model produces both celebrated and failed companies in the same category.

Retail versus direct: the trade-off in plain terms

Answer first: retail gives you distribution and takes your margin and your customer data. Direct gives you margin and data and hands you the acquisition problem. Neither is superior; they suit different products and different repeat rates.

The last row is the honest summary. Every celebrated direct-to-consumer story is a bet that lifetime value will outrun acquisition cost, and the stories that got written up are the ones where it did.

What a good case study looks like

Answer first: it states what was done, what changed at the same time, what cannot be attributed, and what the author’s relationship to the work is. Very few do all four, and the ones that do are worth more than a dozen that do not.

Writing honest case studies about your own work

Answer first: describe the mechanism, name what else changed, and decline to claim causation you cannot demonstrate. A case study that acknowledges its limits reads as more credible, not less, to the people capable of buying from you.

The last row is counterintuitive and reliably true: stating what you cannot claim makes everything you do claim more believable. It is also the single easiest way to distinguish your case studies from the genre they sit in.

Benchmarks: why most of them are unusable

Answer first: most published marketing benchmarks come from vendor surveys of self-selected customers, which makes them a description of that vendor’s client base rather than of your industry. Check the methodology and the sample before using any figure as a target.

Your own data beats any benchmark

Answer first: your conversion rate last quarter is a better target than an industry average, because it controls for your product, your price, your audience and your market. Benchmarks are useful for orientation and useless as goals.

When copying a competitor is reasonable

Answer first: when you can see the mechanism rather than infer it, when your economics resemble theirs, and when you would still do it if they had not. Copying an execution you cannot explain is how companies end up with expensive marketing nobody internally understands.

The three questions that make any case study useful

Answer first: what is the mechanism, does it survive at my scale, and what would I have to be true for it to work here. If you can answer all three, the case study has done its job. If you cannot answer the first, close the tab.

Where to find better evidence than case studies

Answer first: company filings for public companies, platform documentation for how a channel actually works, regulator guidance for what may be claimed, and your own tests for everything else. All four are primary sources; a case study almost never is.

The ordering is by how much the source’s interests are aligned with getting the answer right rather than with a particular answer. Your own test sits top because nobody is selling you its conclusion.

What to do instead of copying a case study

Run the four structural checks on your own business, in order: what step stops people buying, whether you own the customer relationship, whether your identity has been consistent long enough to be recognized, and whether you answer the questions buyers actually ask before purchase.

1. Buy from yourself. Note every step, every field, every wait. Delete what you can.
2. Check whether you hold customer contact details you may lawfully use, or whether a platform does.
3. Look at your last three years of marketing side by side and see whether it looks like one company.
4. List the ten questions buyers ask before purchase and check whether each is answered publicly.
5. Find where your product or service is actually encountered, and whether you control that surface.
6. Photograph your own work rather than licensing stock imagery of somebody else’s.
7. Pick one of the above and do it properly rather than all six badly.

None of those require a case study, a budget or an agency. They are the mechanisms the good case studies are describing underneath the narrative, and they are available now.

Our branding services page covers the consistency question in practice, and our advertising page covers where paid media genuinely fits for a business without CPG-scale budgets.

How should you read a brand advertising case study written by the agency that made the work?

Almost every case study in circulation was written by a party with an interest in the outcome. That does not make them worthless, but it does mean reading them for structure rather than conclusion. Three questions separate evidence from a showreel.

What was the comparison?

A result reported against the period before the campaign is not a result — it is a before-and-after in a business that was also doing other things, in a market that was also moving. The useful comparisons are geographic holdouts, matched-market tests, or a genuine control group who did not see the work. When a case study reports a lift with no comparison named, the lift is the sum of the campaign and everything else that happened.

Was the metric chosen before or after the results came in?

Brand campaigns generate a large number of measurable quantities, and any campaign will move some of them. A case study reporting an unusual composite metric, or one that appears in no other case study by the same agency, was very likely selected once the data was in. The tell is specificity without precedent: awareness among a narrow subgroup, consideration on a proprietary index, sentiment measured a particular way.

What is missing from the account?

Spend is the most common omission, and without it no efficiency claim can be evaluated. Time period is second. Then whether anything else launched alongside — a price change, a distribution expansion, a product improvement — any of which would move the same numbers. A case study that names its own confounders is unusual and worth taking seriously for exactly that reason.

Survivorship is the structural problem

Nobody publishes the campaign that did nothing. Every case study you can read is drawn from the successful tail, which means the genre systematically overstates how reliably any technique works. This is not dishonesty by any individual agency; it is an artefact of what gets written up. Treat the body of published case studies as a catalog of what is possible, never as a base rate for what is likely.

What to do with one anyway

Read it for the mechanism rather than the number. Why did the agency believe this idea would change behavior, what did they think the barrier was, and does that reasoning transfer to your situation? A case study with a clearly articulated hypothesis is useful even where the results are unverifiable, and one with a spectacular number and no reasoning is not useful even if the number is true.

Advertising, brand and measurement talks from the platform publishers

Publicly available sessions on advertising effectiveness, brand building and measurement. None of these are ours; each is credited to its channel and upload date, every identifier was checked live before publication, and each tile loads its player only when clicked.

Frequently asked questions