Hanna runs a refill shop for cleaning products in Vienna. Customers bring their own bottles; she fills them with washing-up liquid, laundry gel and surface spray, priced by weight.
Her homepage headline reads: “Eco-friendly cleaning solutions for a greener tomorrow.” Underneath: “Austria's best refill experience. Join the movement!”
A visitor reads it and still does not know what the shop sells, where it is, or whether she can just walk in with an old bottle. The words sound good and say almost nothing.
Since 27 September 2026, two of them may also be against the law.
Hanna and her shop are imagined. The research, laws and dates below are real.
The previous chapter built the structure of a first screen. This one is about the words that go into it, and into the rest of the page: how people actually read them, what makes them easy and believable, and where marketing language crosses into a legal problem.
People scan, then read
In 1997 Jakob Nielsen published a short column with a blunt opening: people rarely read on the web. In NN/g's usability studies, 79% of test users always scanned any new page; only 16% read word by word.
The same column summarized a study he ran with John Morkes. They built a small fictional website, “Travel Nebraska”, and wrote it five ways: a promotional “marketese” control, a concise version with about half the words, a scannable version with bullet lists, bold keywords and more headings, an objective version without boasting, and a version combining all three. Fifty-one users each tried one version, doing search and judgment tasks. The researchers combined task time, errors, memory, site-map accuracy and satisfaction into one usability score.
The combined version scored 124% higher than the promotional control. That number gets quoted a lot, so here are the caveats the full report makes visible. Each version was tested by about ten people. Several individual differences were only marginal or not significant. The site was tiny and fictional, the tasks were fact-finding, and the control was written to be over the top.
The direction has held up, though. Kate Moran's 2020 NN/g update, “How People Read Online”, concludes that people are still far more likely to scan than read word for word, and that how much they read depends on motivation and task.
Note the objective bar. Removing boasting helped on its own. “Austria's best refill experience” is exactly the kind of sentence the study cut.
What to do with this: write for the scanner first. Put the meaning in the first words of every heading and sentence. Cut your draft by half, then cut the adjectives. Use lists for anything people will compare.
Plain language is a standard
“Write plainly” sounds like taste. It is increasingly a formal standard, and in some places a legal duty.
In June 2023, the International Organization for Standardization published ISO 24495-1, Plain language, Part 1: Governing principles and guidelines. Its definition is a useful test for any web page: a communication is in plain language if its wording, structure and design are so clear that the intended readers can easily find what they need, understand what they find, and use that information.
Notice that only one of the four principles is about words. The others are about choosing the right content, structuring it so people can find things, and testing it. That is a good description of a landing page.
In the United States, the Plain Writing Act of 2010 (Public Law 111-274, signed on 13 October 2010) requires federal agencies to use clear communication “that the public can understand and use” in documents about benefits, services and how to comply with requirements. It does not apply to private businesses, but it reflects the same idea.
The UK government's publishing guidance is the most practical version. It states that plain English is mandatory for all of GOV.UK, notes that 1 in 6 adults in England have very poor literacy skills, recommends short familiar words (“buy”, not “purchase”), the active voice, and splitting sentences longer than 25 words.
Plain language is not dumbing down. Specialists prefer it too, because it is faster. Your customers are reading on a phone between other things; every unfamiliar word is a small toll.
What to do with this: read your homepage aloud to someone outside your industry. Every time you have to explain a word, replace it with the explanation.
Readability, measured
Readability formulas turn some of this into a number. The best known is Rudolf Flesch's reading ease score, from “A new readability yardstick”, published in the Journal of Applied Psychology in 1948.
The score goes up when sentences are shorter and words have fewer syllables. Higher means easier. It measures only those two surface features: it cannot tell whether a sentence makes sense, is true, or answers the reader's question. A string of short nonsense words scores beautifully.
Use readability scores like a bathroom scale: useful for spotting a trend, useless for judging health on their own. If your score drops sharply after an edit, you probably added long words or merged sentences. If it is already high, the problem is elsewhere: usually vagueness.
Try it on Hanna's headline below, or paste your own copy. The tool also flags phrases that tend to be vague, need proof, or, in the EU, may now be banned outright.
With JavaScript off, the text box still works as a scratchpad. Count words per sentence by hand and look for the phrases listed in the legal section below.
Runs entirely in your browser; nothing is sent or stored. Syllables are estimated with a simple English heuristic, so scores are approximate. Flags are prompts to check, not legal findings.
Specific beats impressive
Hanna's specific version does not sound clever. It says: refill your cleaning products in Vienna; bring any clean bottle to the shop in Neubau; washing-up liquid, laundry gel and surface spray by weight, from €2.40 a litre; open Tuesday to Saturday.
It wins because a visitor can picture it. There is research on why that matters.
In four experiments published in 2010, Manuel Hansen and Michaela Wänke found that statements with the same content were judged more probably true when written in concrete rather than abstract language. The effect appeared to work through vividness: concrete statements were easier to imagine.
A second line of evidence comes from customer service. Grant Packard and Jonah Berger analyzed more than 1,000 real conversations between customers and employees, plus experiments, in a 2021 Journal of Consumer Research paper. When employees used more concrete language, customers were more satisfied, more willing to buy and bought more, because they inferred the employee was listening. That is spoken service, not web copy, so treat it as supporting, not proof for headlines.
Specific copy has a second advantage: it is honest by construction. You cannot write “from €2.40 a litre” unless it is true. Vague copy lets a business hide from its own offer; specific copy forces it to have one.
What to do with this: for every adjective on your page, ask “compared with what, measured how?” Replace it with the number, the name, the place or the example that would answer. If nothing answers, delete the adjective.
You, we and I
Most copywriting advice says: talk about “you”, not “we”. The UK government's guidance agrees for its own pages: address the user as “you” where possible, and write as if talking to them one to one.
That is sound guidance for clarity. Instructions written to “you” are easier to follow than instructions about “the customer”. But the research on persuasion is more surprising than the slogan.
In a 2018 Journal of Marketing Research paper, Grant Packard, Sarah Moore and Brent McFerran studied customer service replies. Managers, staff and consumers all believed agents should emphasize “we” (the firm) serving “you” (the customer). Across five studies with lab experiments and field data, agents who said “I” instead of “we” increased perceived empathy and agency, which raised satisfaction and purchases. Customer-referencing “you” had little impact and sometimes backfired.
This was service conversations, not landing pages. But it suggests the real mechanism is not the pronoun count. It is whether a real, accountable person seems to be on the other side. For a small business, that is an advantage big brands cannot copy.
What to do with this: use “you” for anything the visitor has to do (“Bring any clean bottle”). Use “I” or named people where it is true (“I refill every bottle myself”). Replace a faceless corporate “we are passionate about” with a fact about what you do.
Borrow your customers' words
The second of Jakob Nielsen's ten usability heuristics, developed with Rolf Molich in 1990 and refined in 1994, is “match between the system and the real world”: speak the users' language, with words, phrases and concepts familiar to the user, rather than internal jargon.
For copy, the users' language is rarely the language in your pitch deck. Hanna calls her product “concentrated refill formulations”. Her customers call it “washing-up liquid” and ask “can I bring my own bottle?”
This practice is often called voice-of-customer research, and it needs no budget. Read your reviews, the questions people email, what they say on the phone, what they type into your site search, and how they describe you to friends. Note the exact phrases, especially the ones that repeat.
Those phrases do three jobs. They match what people search for. They make the visitor feel understood (the same listening effect Packard and Berger found). And they surface objections you should answer on the page. If three customers asked “do I need to clean the bottle first?”, that sentence belongs near the top.
What to do with this: collect twenty real customer phrases this week. Highlight the five that repeat. Make sure at least two appear, verbatim or nearly, in your hero or first section.
Buttons that say what happens
Button text is the shortest copy on the page and often the most important. The best evidence here is qualitative, from usability testing, and it points one way: say what will happen.
In a 2017 NN/g article, Aurora Harley and Kim Flaherty described how “Get Started” stops users. On a meal-delivery site, eight users were asked to find out what the service offered; six clicked “Get Started” and landed in sign-up steps without learning what it was. On another site, eye-tracking showed users who saw a “Get Started Now” button stop looking for details and assume they had to sign up to learn more. The authors' advice: name the specific action, because a link is a promise.
Vague labels
- Submitwhat happens?
- Learn moreabout what?
- Get startedwith what?
- Click herewhere to?
The visitor has to guess the next step.
Specific labels
- Send my enquiry✓
- See refill prices✓
- Book a 15-minute call✓
- Download the menu (PDF)✓
The label is a preview of the next screen.
Original example labels for illustration, not test results.
You may have read that changing “your” to “my” in a button (“Start my free trial”) raised conversions by 90%. That figure circulates widely in marketing articles and is usually traced to a single test on a practitioner's blog. We could not find a primary write-up with sample sizes or a replication, so we cannot assess it. Treat micro-wording lifts like that as hypotheses to test on your own traffic, not as laws.
What does generalize is the information-scent logic: a specific label lets people predict the next screen, so the ones who click are the ones who wanted it. The chapter on calls to action covers placement, size and contrast.
What to do with this: read each button on your site and complete the sentence “When I click this, I will…”. If the button text does not already finish that sentence, rewrite it so it does.
Claims the law will check
Here is where vague copy stops being a style problem. Regulators on both sides of the Atlantic treat many marketing phrases as factual claims that need evidence, and the EU has just banned a whole category of them.
From 27 September 2026, the directive adds new items to the UCPD's blacklist of practices that are always unfair. They include: making a generic environmental claim when the trader cannot demonstrate recognised excellent environmental performance relevant to the claim (point 4a); making an environmental claim about the entire product or business when it concerns only one aspect (4b); claiming, based on offsetting greenhouse gas emissions, that a product has a neutral, reduced or positive impact on the environment (4c); and displaying a sustainability label that is not based on a certification scheme or established by public authorities (2a).
The recitals give examples of generic claims: “environmentally friendly”, “eco-friendly”, “green”, “nature's friend”, “ecological”, “climate friendly”, “gentle on the environment”, “carbon friendly”, “energy efficient”, “biodegradable”, “biobased”. Offset-based examples include “climate neutral”, “CO₂ neutral certified”, “carbon positive” and “climate compensated”. A claim stops being generic when its specification is given in clear and prominent terms on the same medium: the recital contrasts “climate-friendly packaging” (generic) with “100% of energy used to produce this packaging comes from renewable sources” (specific, still subject to the general misleading rules). Colors and images combined with words can also make a generic claim.
The directive also amends Article 6: claims about future environmental performance (“net zero by 2030”) need a detailed, realistic, independently verified implementation plan, and advertising benefits that are irrelevant or do not result from the product is misleading.
Source: Directive (EU) 2024/825, Arts. 1 and 4, recitals 9 and 12 (EUR-Lex). Member States had to transpose it by 27 March 2026; check your country's implementing law. A separate proposed Green Claims Directive was paused by the Commission in June 2025 and, as far as we could verify, has not been adopted. General information, not legal advice.
For Hanna, “eco-friendly cleaning solutions” is a textbook generic claim. She can replace it with the specific, true thing her shop does: customers reuse their own bottles. That is both legal and better copy.
Article 6 covers misleading actions, including false or deceptive claims about a product's main characteristics, benefits, results to be expected, price advantages or the trader's awards and approvals. Article 7 covers misleading omissions: hiding material information, or giving it in an unclear, ambiguous or untimely way. Article 12 lets courts and authorities require a trader to furnish evidence of the accuracy of factual claims, and to treat claims as inaccurate if the evidence is missing or insufficient. Article 5(3) keeps room for “the common and legitimate advertising practice of making exaggerated statements or statements which are not meant to be taken literally”. Since the Omnibus Directive, fines in coordinated cross-border cases must be able to reach at least 4% of annual turnover in the Member States concerned (Art. 13).
Source: Directive 2005/29/EC, consolidated text (EUR-Lex). Not legal advice; national rules and courts decide individual cases.
Under the FTC's 1984 Policy Statement Regarding Advertising Substantiation, advertisers must have a reasonable basis for objective claims before disseminating them; if an ad says “tests prove” or “studies show”, it needs at least that level of support. The 1983 Policy Statement on Deception says the Commission generally will not pursue obviously exaggerated or puffing representations that ordinary consumers do not take seriously, but notes that some exaggerated claims are taken seriously and are actionable. It also warns that accurate text may not fix a false headline, because reasonable consumers may glance only at the headline. For environmental claims, the FTC's Green Guides (16 CFR § 260.4) say marketers should not make unqualified general environmental benefit claims, and give “Eco-friendly” as an example.
Sources: FTC Policy Statement Regarding Advertising Substantiation (1984); FTC Policy Statement on Deception (1983); 16 CFR Part 260 (eCFR). Not legal advice.
CAP Code rule 3.7 requires advertisers to hold documentary evidence for claims consumers are likely to regard as objective. ASA guidance treats “best-selling”, “leading” and “No. 1” claims as comparisons with identifiable competitors, usually needing unit-sales or market-share data; absolute claims are read as whole-market comparisons. In 2022 the ASA ruled against the AA's “The UK's number 1 breakdown service provider” because it lacked competitors' membership figures. The DMCC Act 2024, in force for unfair practices since 6 April 2025, adds the misleading-actions rules described in chapter 1.
Source: ASA, “Types of claims: Best-selling” (CAP advice, 11 April 2025). Not legal advice.
Put together, the rules suggest a practical sorting for superlatives:
Probably fine: obvious opinion or exaggeration nobody takes literally (“the friendliest refill shop we know”, “cleaning that smells like a holiday”).
Needs evidence: anything a reasonable person could read as a measurable fact: “#1”, “best-selling”, “leading”, “fastest”, “cheapest in Vienna”, “50% less plastic”, “recommended by experts”. Keep the evidence on file before you publish.
In the EU, banned unless you meet the high bar: generic environmental claims and offset-based neutrality claims.
The safest copy is also the most persuasive: specific, verifiable statements. “Bring your own bottle” needs no lawyer. “Austria's best” needs a market survey.
What to do with this: search your site for the flagged words in the tool above. For each, either attach the evidence, make it specific, or delete it. Do the same for product pages, ads and packaging, since the rules apply to all commercial communication, not just the homepage.
What to change tomorrow
Cut your hero copy in half
Then remove every adjective that cannot answer “compared with what?”. Concise and objective both improved usability in Morkes and Nielsen's study.
Rewrite the headline one rung down
Move from abstract (“solutions”) to category (“refillable cleaning products”) to concrete (what, where, how much). The tip on hero headlines has more examples.
Collect twenty customer phrases
From reviews, emails, calls and site search. Use the repeating ones in your headings and answer the repeating questions near the top.
Put a person behind the words
Use “you” for instructions and “I” or named people where true. Replace “we are passionate about” with a fact.
Make every button finish a sentence
“When I click this, I will…” Replace Submit, Learn more and Get started with the specific next step.
Audit your claims this week
Remove generic green words (eco-friendly, green, climate neutral) unless you can meet the EU's bar, and keep evidence on file for any “best”, “#1” or number you publish. If in doubt, ask a lawyer in your country.
Check the typography next
Clear words still fail in tiny gray text. The next chapter covers typography that reads as quality.
What the research supports
Strong: people scan web pages more than they read them (NN/g, 1997 to 2020). Moderate: concise, scannable, objective writing improved a composite usability score in one small 1997 study; concrete statements are judged more likely true in lab experiments; concrete language and first-person “I” improved satisfaction and purchases in customer-service settings. Weak: claims that a single word swap (“your” to “my”) reliably lifts conversions. Readability formulas measure sentence and word length only. The legal sections describe rules in force on 18 October 2026, not predictions of how a given claim will be judged.
Sources
- Nielsen (1997). How Users Read on the Web. Nielsen Norman Group, 30 September 1997. Source of the 79% / 16% finding and the summary of usability gains.
- Morkes & Nielsen (1997). Concise, SCANNABLE, and Objective: How to Write for the Web. Nielsen Norman Group. Full report consulted for the five conditions, the 51-user measurement study, the composite score and its limitations.
- Moran (2020). How People Read Online: New and Old Findings. Nielsen Norman Group, 5 April 2020.
- ISO (2023). ISO 24495-1:2023 Plain language, Part 1: Governing principles and guidelines. Catalog record (published June 2023) consulted; definition and principles as quoted in the CLEO house style guide.
- Plain Writing Act of 2010. Public Law 111-274, 13 October 2010. GovInfo.
- GOV.UK. Use clear language and Use the right tone. Writing to GOV.UK standards.
- Flesch (1948). A new readability yardstick. Journal of Applied Psychology, 32(3), 221–233. Bibliographic record consulted; formula as standardly given.
- Hansen & Wänke (2010). Truth from language and truth from fit: The impact of linguistic concreteness and level of construal on subjective truth. Personality and Social Psychology Bulletin, 36(11), 1576–1588. Abstract consulted.
- Packard & Berger (2021). How concrete language shapes customer satisfaction. Journal of Consumer Research, 47(5), 787–806. Abstract consulted.
- Packard, Moore & McFerran (2018). (I'm) Happy to Help (You): The impact of personal pronoun use in customer–firm interactions. Journal of Marketing Research, 55(4), 541–555. Abstract consulted.
- Kaley (2018). Match Between the System and the Real World. Nielsen Norman Group. With Nielsen, 10 Usability Heuristics for User Interface Design (1994, reviewed 2024) for the heuristics' origin.
- Harley & Flaherty (2017). “Get Started” Stops Users. Nielsen Norman Group, 20 August 2017.
- Directive (EU) 2024/825. Official text, OJ L, 6 March 2024. Annex I points 2a, 4a–4c, amended Article 6, recitals 9 and 12, Article 4 (dates).
- Directive 2005/29/EC (UCPD). Consolidated text, Articles 5, 6, 7, 12 and 13. EUR-Lex.
- CMS (2025). The Green Claims Directive: a legal update. Law-firm summary of the proposal's June 2025 pause.
- Federal Trade Commission (1984). FTC Policy Statement Regarding Advertising Substantiation.
- Federal Trade Commission (1983). FTC Policy Statement on Deception. Full text consulted for puffery and headline passages.
- 16 CFR § 260.4. General environmental benefit claims (FTC Green Guides). eCFR.
- Advertising Standards Authority / CAP (2025). Types of claims: “Best-selling”. 11 April 2025, including the AA (2022) ruling summary.
Sources checked 18 October 2026. Figure 1 redraws published numbers; Figures 2 to 4, the button boards and the demo copy are original illustrations. Hanna, Klar and every price in this chapter are invented. Legal summaries are general information, not legal advice; check your own jurisdiction and, for anything important, ask a qualified lawyer.