Tomasz renovates bathrooms. His website says what most of them say: “Quality bathrooms, fair prices, over 200 happy customers.” Below that, a gallery of thirty gleaming photos. White tiles, chrome taps, a towel folded into a swan.
Elena is choosing a fitter for her parents' flat. The bathroom there is narrow, the pipes are old, and her father uses a walking frame. She looks at the gallery for about four seconds. Every room is beautiful. None of them looks like her parents' bathroom.
The next site she opens shows one job. A 1970s bathroom, 1.6 metres wide. A wall that could not be moved. A shower tray swapped for a level floor, so a walker could roll in. Nine working days. A short quote from the owner, a retired teacher, about being able to shower without help again.
Elena books a visit with that company. She never counted their customers.
Tomasz and Elena are imagined. The research, laws and rulings below are real.
Specific beats impressive
“Quality bathrooms” is a claim anyone can make. “A level-access shower in a 1.6-metre room with the original wall kept” is a claim only someone who did it would think to write. That difference is most of what this chapter is about.
Psychologists have known for decades that concrete words, the ones you can picture, are easier to process and remember than abstract ones. Allan Paivio's dual-coding account, summarized in a 1991 review, explains it this way: a concrete word gets stored both as language and as a mental image, so it has two routes back into memory.
More recently, Grant Packard and Jonah Berger tested what concreteness does in a commercial setting. In their 2021 paper in the Journal of Consumer Research they analyzed real customer-service conversations and ran controlled experiments. In 941 email exchanges at a retailer, employees who used more concrete language were associated with higher customer spending over the next 90 days. In the experiments, tiny wording changes mattered: replies that said “that top” or “the t-shirt” instead of “that” left customers more satisfied.
The mechanism they found is the interesting part. Concrete language made customers feel listened to. A specific word is evidence that someone paid attention to your particular situation.
That is exactly what a visitor is trying to judge on a services page. Not “Is this company good?” but “Will they understand my job?” Vague praise cannot answer that question. A specific story can.
Abstract
HeadlineWe transform spaces.
Proof200+ satisfied customers across the region.
Quote“Fantastic service, highly recommend!”
True, perhaps. But it could describe any company, so it proves nothing about this one.
Concrete
HeadlineA walk-in shower in a 1.6 m bathroom, without moving a wall.
Proof9 working days. Level floor, fold-down seat, grab rails at 85 cm.
Quote“I can shower on my own again.” Owner, aged 78
Every line can be pictured, and every line could be checked.
Do this: take the vaguest sentence on your page, usually the one with “quality”, “solutions” or “passionate” in it, and rewrite it so a stranger could draw a picture of it.
One client, not two hundred
Counting customers feels like strong proof. Large numbers are reassuring. But there is a well-known line of research suggesting that one named case can move people more than a statistic, even a bigger one.
People tend to give more to help one specific, named person than to help a larger, statistical group. The 2007 study ran field experiments in which people who had just earned $5 for a survey were asked to donate to Save the Children. Some read about Rokia, a seven-year-old girl from Mali, with her photo. Others read statistics about food shortages affecting millions.
In the paper's third study, with 159 participants, people who saw only Rokia gave an average of $2.38. People who saw only the statistics gave $1.14. And people who saw Rokia plus the statistics gave $1.43. Adding the big numbers to the personal story reduced giving.
Now the honest part. In 2024, Maximilian Maier, Yik Chun Wong and Gilad Feldman ran a pre-registered replication of the first and third studies and found no support for the identifiable victim effect. Their reanalysis of an earlier meta-analysis also pointed to publication bias in the literature. They note one important difference: their participants donated hypothetical money, while the 2007 participants gave away real dollars.
So treat this as a strong analogy, not a law of conversion. Charitable giving is not the same decision as hiring a bathroom fitter. What transfers is the practical lesson most marketers and journalists already use: a single, well-chosen case gives the reader something to picture and compare with their own situation. A count of happy customers gives them nothing to hold.
Do this: keep your numbers, but stop leading with them. Put one detailed case first and let the totals support it. For more on counts, ratings and reviews, see Social proof patterns.
Stories lower the guard
Why would a story work differently from a list of facts? One influential answer comes from narrative persuasion research.
When people become absorbed in a story, with vivid images, emotion and focused attention, they are “transported” into it. Transported readers tend to adopt beliefs that fit the story and to evaluate its characters more warmly. They also look less hard for flaws.
In Green and Brock's experiments, 97 participants read a true account adapted from Sherwin Nuland's book How We Die, about a child killed at a shopping mall. The more transported readers were, the more their beliefs matched the story. A second experiment with 69 participants asked readers to circle anything that rang false. Highly transported readers found fewer false notes. Strikingly, labeling the story as fact or as fiction made little difference to either transportation or beliefs.
That finding cuts both ways for a business.
The good news: a well-told case study can carry a visitor past the defensive “prove it” mood that marketing copy usually triggers. A real sequence of problem, decision and outcome reads as a story, not a pitch.
The responsible news: if stories lower people's guard, then a misleading story is more harmful than a misleading bullet point. Which is one reason the law, as you will see below, treats testimonials and results claims as seriously as any other advertising.
Do this: write your case study in time order. Where they were, what was in the way, what you did, what happened. Resist starting with the result. The result means more when the reader has felt the problem.
The anatomy of a case study
Most “case studies” on small-business sites are a photo, a client logo and a one-line quote. That is a credential, not a case. A case shows judgment.
Guidance written for a different audience turns out to be useful here. In a 2019 Nielsen Norman Group article on portfolios, Rachel Krause lists what hiring managers want from a case study: the problem, your role, how you reached a solution, the challenges and abandoned ideas, the effect on users and the business, and what you learned. She also makes the point that dealing with constraints is a key part of any design. Clients choosing a supplier are asking the same questions as a hiring manager.
- The situation, in their wordsWho the client is and what was wrong. “The bathroom was 1.6 m wide and Mr K. could no longer step into the bath.”
- The constraintsBudget, deadline, the wall you could not move, the regulation you had to meet. Constraints are what make your decisions interesting.
- The decisions, including the ones you rejected“We considered a shower tray with a low lip, but a level wet-room floor was safer with a walking frame.” This is where expertise becomes visible.
- The result, with numbers and a baselineWhat changed, measured against how things were before, over a stated period. Say what you do not know.
- The client's voiceA short, real quote, ideally one that admits a doubt before the job. Include their name and role only with permission.
Two details make this structure work.
First, the rejected options. They show the reader that you made choices rather than following a template, and they quietly answer “Why didn't you just…?” before it is asked.
Second, the shape of the quote. In NN/g's research on business-to-business sites, summarized by Therese Fessenden in 2019, participants responded best to testimonials where the writer began with some reservation and ended confident. People also cared who wrote it, including the author's job title and company. A quote that starts with “I wasn't sure we needed a full redesign” is more believable than one that starts with “Amazing!”
Fessenden adds another useful point: case studies help most when your customers come from different industries, because a varied set lets more visitors find someone like themselves. If you serve landlords, restaurants and dentists, one case from each beats three from the same kind of client.
Do this: pick your best job of the last year and write it in the five parts above, in under 400 words. That is one case study. Do it for two more clients who differ from the first.
Numbers that mean something
A result without context is just a big number. “Enquiries up 300%” could mean three enquiries became twelve, or that a seasonal business compared December with June.
NN/g's Evan Sunwall, writing about keeping a portfolio over time in 2023, puts it bluntly: before-and-after pictures are not enough, you need to show whether the change achieved a goal. His practical advice is to capture the starting state, as screenshots or data, before any project begins. You cannot report a baseline you never recorded.
| Element | Bare | With context |
|---|---|---|
| Baseline | “Bookings doubled.” | “From 14 to 29 online bookings a month.” |
| Timeframe | “Fast results.” | “Measured over the 3 months after launch, compared with the same 3 months a year earlier.” |
| Other causes | (not mentioned) | “They also started a newsletter in the same period, so not all of this is the website.” |
| Typicality | “Our clients see 2× growth.” | “This was our strongest result last year. Most clients see a smaller change.” |
| Source | (not mentioned) | “Figures from the client's booking system, shared with permission.” |
The last two rows are not just good manners. Saying whether a result is typical is, in the United States, close to a legal requirement, as the next sections explain. The “other causes” row is what makes a careful reader trust you: admitting a limitation signals that the rest is not inflated.
If a result is not a number at all, say so plainly. “The owner can now update the menu herself in two minutes instead of emailing us” is a concrete, checkable outcome.
Do this: before your next project starts, take screenshots and write down three numbers you could measure again in three months. Agree with the client that you may publish them.
Before and after, honestly
A before/after comparison is the most compact case study there is. It is also the easiest to fake by accident. Change the lighting, the angle and the season between two photos and almost anything looks like an improvement.
On the web, before/after images are often shown with a draggable divider. It is a satisfying pattern, and it is frequently built in a way that excludes people. A divider that only responds to dragging fails anyone using a keyboard, a switch device or a shaky hand.
WCAG 2.2 added a specific success criterion for this. Success Criterion 2.5.7, Dragging Movements (Level AA) says anything that works by dragging must also work with a single pointer without dragging. The W3C's own explanation gives a slider as the example: clicking or tapping the track should move the thumb there.
The simplest way to meet that is to not reinvent the control. A native <input type="range"> already supports arrow keys, Page Up and Down, Home and End, tapping the track, and screen-reader announcements. Try the one below.
Fictional page mock-ups. Without JavaScript the split stays at the halfway point. The control is a native range input with a visible label, a spoken value and button alternatives.
Three implementation details matter more than the visual polish. The range input needs a visible label, not just an icon. It should announce something meaningful through aria-valuetext, such as “mostly after”, rather than a bare number. And both images need text alternatives that describe the difference, because a person who cannot see the slider still deserves the comparison.
Do this: if you show before/after images, shoot the “after” from the same spot as the “before”, and make sure the comparison works with the Tab and arrow keys alone.
Results claims and the law
Case studies are advertising. When you publish a client's result to win new clients, consumer protection law treats it like any other claim about what your product or service does.
The FTC's Endorsement Guides say that an endorsement describing a consumer's experience on a central attribute will likely be read as representative of what consumers generally achieve. If the advertiser cannot substantiate that, it should clearly and conspicuously disclose the generally expected performance in the circumstances shown. The FTC's staff guidance adds that statements like “Results not typical” or “Individual results may vary” do not change how the ad is understood. That safe harbor was removed in the 2009 revision; the July 2023 revision kept the approach.
The Guides also say endorsements must reflect the endorser's honest opinion and experience (§255.1(a)), and quotes may not be reworded or taken out of context to distort them (§255.1(b)).
Sources: 88 FR 48092, 26 July 2023; FTC, Endorsement Guides: What People Are Asking. General information, not legal advice.
In plain words: if you put your best client's result in the spotlight, either be able to show that most clients get something similar, or say clearly what most clients actually get. A grey footnote saying results vary does not do that job.
Since 21 October 2024, the FTC also has a binding Rule on the Use of Consumer Reviews and Testimonials (16 CFR Part 465). It makes it unlawful to write or create a testimonial that misrepresents that the person exists, that they used the service, or what their experience was, and to publish one you knew or should have known was fake. Unlike the Guides, violations of the Rule can bring civil penalties. In December 2025 the FTC sent warning letters under the Rule to ten companies.
The Unfair Commercial Practices Directive treats a practice as misleading if it deceives, or is likely to deceive, the average consumer about the main characteristics of a product, expressly including “the results to be expected from its use” (Art. 6(1)(b)), even if each statement is factually correct but the overall presentation misleads. Since the Omnibus Directive 2019/2161 (applied from 28 May 2022), Annex I point 23c bans outright submitting or commissioning false consumer reviews or endorsements, or misrepresenting reviews or social endorsements, to promote products.
The UCPD protects consumers. If your clients are businesses, the Misleading and Comparative Advertising Directive 2006/114/EC protects traders and uses the same test of “the results to be expected” (Art. 3(a)). In the UK, the Digital Markets, Competition and Consumers Act 2024 (Schedule 20, para. 13) bans fake and concealed-incentive consumer reviews, and publishing reviews in a misleading way, from 6 April 2025.
Sources: UCPD consolidated text; Directive 2006/114/EC; DMCC Act 2024, Sch. 20 para. 13. General information, not legal advice; check the rules where you and your clients are.
None of this makes case studies risky. It makes exaggerated case studies risky. The structure from earlier, with a baseline, a timeframe, the other causes and a word on typicality, is also the structure that keeps you on the right side of these rules.
Do this: next to any headline result, add one plain sentence on how typical it is. If you cannot write that sentence honestly, choose a more typical case to feature.
Permission, privacy and ownership
A case study is about someone else's business. Before you publish one, three questions need a yes.
1. Did the client agree, in writing?
Many service contracts include confidentiality clauses or separate NDAs. Even without one, publishing a client's numbers, problems or internal screens without asking can end a relationship. Ask at the start of the project, not after launch, and put it in the contract: what you may show, whether you may name them, and whether they review the text first.
When you cannot name a client, you can still show the work. Krause's NN/g article suggests showing process material, blurring or redacting identifying details, or recreating the design in a neutral style. An anonymous case (“a 40-room hotel in the Algarve”) is weaker than a named one, but far stronger than nothing.
2. Is there personal data in it?
Under the GDPR, a client contact's name, photo, job title and quote are personal data (Regulation (EU) 2016/679, Art. 4(1)). You need a lawful basis to publish them (Art. 6). For a named testimonial with a photo, consent is the usual and cleanest basis, and consent can be withdrawn at any time (Art. 7(3)), so be ready to remove it.
Screenshots are the sneaky part. A dashboard, a booking calendar or an email inbox in a portfolio shot can expose the client's own customers. Use dummy data or blur it. Sunwall's NN/g article makes the same point from the hiring side: careless handling of sensitive data in a portfolio reads as a warning sign.
3. Do you have the right to show the design?
Who owns work made for a client depends on the country and the contract, and the answer surprises people in both directions.
United States. Work by an employee within their job is “made for hire” and belongs to the employer. Work by an independent contractor counts as made for hire only in nine listed categories and only with a signed written agreement (17 U.S.C. §101). Otherwise the creator owns it until a signed written transfer (§204(a)). If you have assigned copyright to the client, you need their permission, ideally a portfolio clause, to keep showing the work.
Germany. Copyright itself cannot be transferred during the author's life; the client receives rights of use (Nutzungsrechte) instead (§29 UrhG). Authors also keep moral rights, including recognition of authorship, which the Berne Convention protects in Article 6bis.
Sources: US Copyright Office, Circular 30: Works Made for Hire; 17 U.S.C. §204; §29 UrhG. Rules differ across EU member states and the UK. General information, not legal advice.
The practical fix is the same everywhere: one sentence in your contract granting you the right to show the finished work and a short description in your portfolio and marketing, with or without the client's name, plus a list of anything that stays confidential. Clients rarely object when asked up front.
Two smaller traps. A client's logo is their trademark, so showing it in a “trusted by” strip implies an endorsement they may not have given. And photos of the client's staff need those people's agreement, not only the business owner's.
Do this: add a portfolio and case-study clause to your standard contract today, and ask current clients for a written yes before you publish.
What to change tomorrow
Elena did not need two hundred customers. She needed one job that looked like hers, told well enough that she could see herself in it. Here is how to give your visitors that.
Replace one gallery with one case
Keep the photos, but lead with a single project written in five parts: situation, constraints, decisions, result, client's voice. Put it high on your main page or one click away.
Make every claim picturable
Rewrite abstract lines (“quality solutions”) as concrete facts (“level-access shower, 1.6 m room, 9 days”). If a stranger could not draw it, it is not specific enough.
Give every number a baseline and a timeframe
“From 14 to 29 bookings a month over three months” beats “bookings doubled”. Mention other causes that you know about.
Say how typical the result is
One sentence near the headline result. In the US this is how the FTC expects results testimonials to work; in the EU and UK it keeps the overall impression honest.
Use quotes that admit a doubt
Ask clients what they were unsure about before working with you. Publish their exact words, with name and role only if they agreed in writing.
Shoot before/after photos the same way
Same viewpoint, light and crop. If you build a comparison slider, use a native range input and test it with the keyboard alone.
Fix the paperwork before the next project
Add a portfolio clause, a consent line for named testimonials, and a rule to blur third-party data in screenshots.
What the research supports
Concrete language has good evidence behind it, including field data and experiments in customer service (Packard & Berger). Narrative transportation is a well-established finding, though the original studies used dramatic stories rather than business case studies. The identifiable victim effect is famous but contested: a 2024 pre-registered replication found no effect with hypothetical donations, so use it as an analogy, not a guarantee. NN/g's guidance on case studies comes from usability research and hiring-manager interviews, not conversion experiments. No study here shows that one case study will lift your enquiries by a specific amount.
Sources
- Packard, G. & Berger, J. (2021). How Concrete Language Shapes Customer Satisfaction. Journal of Consumer Research, 47(5), 787–806. Publisher's full text consulted for the email field study, the wording experiments and the perceived-listening mechanism.
- Paivio, A. (1991). Dual coding theory: Retrospect and current status. Canadian Journal of Psychology, 45(3), 255–287. Bibliographic record and summaries consulted for the concreteness account.
- Small, D. A., Loewenstein, G. & Slovic, P. (2007). Sympathy and callousness: The impact of deliberative thought on donations to identifiable and statistical victims. Organizational Behavior and Human Decision Processes, 102(2), 143–153. Author-hosted full text consulted for the Study 3 design, sample and means.
- Maier, M., Wong, Y. C. & Feldman, G. (2024). Revisiting and Rethinking the Identifiable Victim Effect: Replication and Extension of Small, Loewenstein, and Slovic (2007). Collabra: Psychology, 9(1). Abstract consulted for the null result, the publication-bias reanalysis and the hypothetical-donation limitation.
- Green, M. C. & Brock, T. C. (2000). The role of transportation in the persuasiveness of public narratives. Journal of Personality and Social Psychology, 79(5), 701–721. Full text consulted for the experiments, samples and fact-versus-fiction result.
- Krause, R. (2019), Nielsen Norman Group. 5 Steps to Creating a UX-Design Portfolio. Case-study elements, constraints and showing work under NDA.
- Sunwall, E. (2023), Nielsen Norman Group. How to Maintain a UX Portfolio Over Time. Capturing a baseline, showing outcomes, handling sensitive data.
- Fessenden, T. (2019), Nielsen Norman Group. What B2B Designers Can Learn from B2C About Building Trust. Testimonial shape, author details and varied case studies.
- Federal Trade Commission (2023). Guides Concerning the Use of Endorsements and Testimonials in Advertising, 88 FR 48092, 16 CFR Part 255. Regulatory text of §§255.1 and 255.2 consulted.
- Federal Trade Commission. FTC's Endorsement Guides: What People Are Asking. Staff answer on “Results not typical” disclaimers.
- Federal Trade Commission (2024). Trade Regulation Rule on the Use of Consumer Reviews and Testimonials, 16 CFR Part 465. Definitions and §465.2 consulted; December 2025 warning letters as reported by Venable (29 December 2025).
- Directive 2005/29/EC (Unfair Commercial Practices Directive), consolidated 28 May 2022. Articles 6–7 and Annex I points 23b–23c consulted.
- Directive 2006/114/EC on misleading and comparative advertising. Articles 1–3 consulted.
- Digital Markets, Competition and Consumers Act 2024, Schedule 20, paragraph 13. Text consulted; CMA guidance CMA208 (4 April 2025) noted.
- Regulation (EU) 2016/679 (GDPR). Articles 4(1), 6 and 7(3).
- US Copyright Office. Circular 30: Works Made for Hire, and 17 U.S.C. §204. Definition, nine categories and the written-transfer rule.
- §29 Urheberrechtsgesetz (Germany). Non-transferability of copyright and grant of usage rights.
- W3C (2023). Understanding Success Criterion 2.5.7: Dragging Movements. WCAG 2.2, Level AA; slider example.
Sources checked 18 October 2026. Legal summaries are general information, not legal advice; rules differ by country and change over time. Tomasz, Elena, the bathroom project and all mock-ups are fictional; the figures are original illustrations. Figure 2 reproduces published means from Small et al. (2007), Study 3.