Joana wants to book a pottery class for her mother's sixtieth birthday. She types “pottery class Porto” into her phone, opens the first two results in new tabs and flicks between them.
The first tab is crowded. A black strip shouts about a discount, the logo floats in the middle in curly italics, and small boxes of text compete with tiles of photos. The second tab has a name in the corner, a sentence that says “Pottery classes in Porto”, a photo of a bowl and one button.
She closes the first tab before it has finished loading. She could not tell you why. It just felt off.
The studio she closed has the better teacher, a kiln twice the size and cheaper classes. None of that reached her.
Joana and both studios are imagined. The research and laws below are real.
That flick between tabs is where most small-business websites win or lose. This chapter is about what happens in it: what the research measured, what it did not, and what you can change so that a fast, unfair judgment lands in your favor.
What 50 ms really measured
You have probably seen the claim: “users form an opinion of your website in 50 milliseconds”. It comes from a real paper with a cheeky title, “Attention web designers: You have 50 milliseconds to make a good first impression!”, by Gitte Lindgaard and colleagues at Carleton University, published in Behaviour & Information Technology in 2006.
The details matter, so here they are.
In the first study, 22 students saw screenshots of 100 homepages for 500 milliseconds each and rated their visual appeal on an unmarked slider. The pages had been chosen by lab members because they “looked really good or looked really bad”, and they were deliberately not famous sites. Each person rated every page twice. Their ratings were remarkably consistent.
The researchers then took the 25 highest-rated and 25 lowest-rated pages. In the third study, 40 new participants saw those 50 pages for either 50 ms or 500 ms (20 people in each group) and rated appeal on a 1 to 9 scale. The average rating each page received from the 50 ms group correlated with the average from the 500 ms group at r = .95.
That is a striking number. In aggregate, a glance shorter than a blink produced almost the same ranking of pages as a glance ten times longer.
Now the part that rarely makes it into the slide decks. The right side of Figure 1 shows that individual participants agreed with each other far less, and less again at 50 ms. In the 50 ms group, 40% of participants' own round-one-to-round-two correlations fell below .60; in the 500 ms group, none did. The glance gives a reliable crowd verdict built from noisy individual ones.
And the study removed the middle. Studies 2 and 3 used only pages that had been rated clearly beautiful or clearly ugly. Your page is probably neither. Telling a stunning page from a terrible one in 50 ms is easier than telling a decent page from a slightly better one.
So here is what the 50 ms finding supports: people form a visual appeal reaction very quickly, and in aggregate that reaction is stable. Here is what it does not show: that people decide to buy, understand your offer, judge your prices or settle on trust in 50 ms. Nobody read a word in that study. Nobody was choosing a pottery class.
Lindgaard's team did go one step further in a 2011 follow-up. With the same pages and the same 50 ms exposure, people also rated perceived usability and trustworthiness. All three judgments turned out to be largely driven by visual appeal, though the more demanding judgments seemed to be processed differently and leaned on somewhat different visual attributes. In plain words: at a glance, “trustworthy” mostly means “looks good”.
What to do with this: treat the first glance as a gate, not a verdict. Its job is to stop people from closing the tab. Understanding, trust and the decision happen afterwards, on the evidence. If you want to check whether a glance communicates your offer, that is a different test, covered in the three-second test.
Try a 50 ms glance
Below are two invented homepages for the same pottery studio. Flash them, then say which felt more trustworthy. Your screen refreshes in steps of roughly 8 to 17 ms, so the browser can only approximate 50 ms; the demo reports what it actually achieved.
Both pages describe the same fictional studio with the same offer. Only the presentation differs. Nothing is recorded or sent anywhere.
Which page felt more trustworthy?
Pottery classes in Porto
Small groups, all materials and firing included. Beginners welcome.
See class datesOriginal, fictional mock-ups. This is a demonstration of brief exposure, not a measurement of you or a validated test of these designs. Flashes are spaced more than a second apart and use a mid-grey mask.
If you picked the calm page, you are in line with the research below, and you may not be able to say why. The next three sections explain the why, because once you know the mechanism you can design for it on purpose.
Simple and familiar wins
In 2012 Alexandre Tuch and colleagues at the University of Basel, working with Google's user experience research team in Zurich, asked what in a page drives that instant verdict. Their paper in the International Journal of Human-Computer Studies focused on two properties: visual complexity (how busy a page looks) and prototypicality (how much it looks like a typical website of its kind).
They used screenshots of real company websites that participants did not know. In the first study, 59 people saw them for 50, 500 or 1,000 ms. In the second, 82 people saw them for just 17, 33 or 50 ms. Both properties affected beauty ratings at every exposure time, even at 17 ms. Pages that were low in complexity and high in prototypicality were rated the most beautiful; busy, unusual pages did worst.
Two details make this more useful. First, at the very shortest exposures, complexity mattered more than typicality; typicality caught up as viewing time grew. Busy-versus-calm is processed first, “does this look like a proper website?” a moment later. Second, the authors list the limits themselves: passive viewing of screenshots, company websites only, a fairly young sample, and no breakdown of which features create complexity.
Does “typical wins” mean every website should look the same? Not quite. In a 2003 study of industrial design, Paul Hekkert and colleagues found that people preferred products that were “most advanced, yet acceptable”: typicality and novelty both raised preference, but each suppressed the other. People liked novelty as long as it did not cost typicality. Your personality belongs in the details, the photography, the words. The skeleton can be conventional.
What to do with this: count the things on your first screen. Every extra box, badge, banner and font size adds complexity. Keep the layout predictable (name top left, menu top right, headline, one action), and spend your originality where it does not cost recognition. For more on how restraint signals quality, see the psychology of premium-looking sites.
Easy to see feels true
Why would a simpler, more familiar page feel better? The best-supported explanation is processing fluency: the felt ease of taking something in.
In a review in Personality and Social Psychology Review, the authors proposed that aesthetic pleasure depends on the perceiver's processing dynamics: the more fluently we can process an object, the more positive our response. Symmetry, figure–ground contrast, repetition and prototypicality all make things easier to process, and all tend to raise liking. It is an integrative theory built on many experiments, not a single study.
The effect reaches beyond beauty. In a short 1999 experiment, Reber and Schwarz showed people trivia statements such as “Osorno is in Chile” in colors that were easy or hard to read against white. Highly visible statements were judged true more often than moderately visible ones, which were judged true at chance level. Same words, different contrast, different belief.
Be careful with that finding. It concerns unknown trivia with no other evidence available. Low contrast will not make customers disbelieve your opening hours. But it shows the direction of the effect: when nothing else is known, ease is read as a signal.
And on a first visit, almost nothing else is known. That is the whole point of a first impression.
What to do with this: remove friction for the eye. Strong text contrast, one clear typeface, generous spacing, a single focal point, images that are quick to recognize. These are not decoration choices; they lower the effort of the first second. The chapters on typography and whitespace go into the details.
Pretty looks usable
In 1995, Masaaki Kurosu and Kaori Kashimura at Hitachi's Design Center ran a modest experiment that started a whole field. They asked 26 people (designers, engineers and secretaries) to arrange the same buttons for a cash machine screen, producing 26 layouts. Then 252 students rated each printed layout on two 10-point scales: how beautiful it looked, and how easy to use it looked.
The two ratings correlated at .589. The researchers also scored each layout on design principles that are supposed to make interfaces genuinely easier, such as sensible grouping and a logical operation sequence. Most of those barely related to apparent ease of use at all.
The headline everyone quotes is “beautiful interfaces look easier to use”. The detail almost nobody quotes is the top bar of Figure 3. The single strongest predictor of “looks easy” was familiarity. Hold that thought for the section on Jakob's law.
The obvious objection: nobody actually used those layouts. Noam Tractinsky, Adi Shoval-Katz and Dror Ikar addressed exactly that in 2000 with a paper titled “What is beautiful is usable”. Participants used a computerized cash machine that varied in both aesthetics and actual usability, and rated it before and after use. The strong link between perceived beauty and perceived usability held even after using it, and the system's aesthetic level affected post-use usability ratings while its actual usability level did not.
Thorndike noticed that people who rate others on separate qualities cannot keep them separate. In one example in his paper, a flight commander rated 137 aviation cadets on physique, intelligence, leadership and character, with strict instructions to judge each independently. His intelligence ratings still correlated .51 with physique, .58 with leadership and .64 with character, which Thorndike judged too high and too similar to reflect reality. He called it “a constant error”: a general impression colors every specific judgment.
The aesthetic–usability effect is a halo effect for interfaces. A page that looks cared for gets the benefit of the doubt on things the visitor cannot yet check: whether booking will work, whether emails get answered, whether the classes are any good.
The halo also cuts the other way. A cramped, dated page passes its doubt on to the business behind it, which is what happened to Joana's first tab.
What to do with this: polish the parts people see before they can test anything: alignment, spacing, image quality, consistent buttons. Then make sure the page deserves the halo, because a halo that is followed by a broken form turns into a feeling of being tricked. For how trust is earned once the visitor starts checking, see the trust chapter.
Looks and credibility
The most quoted statistic about websites and trust comes from a large Stanford study led by B.J. Fogg, run with Consumer WebWatch and published as a full report in 2002 and a conference paper in 2003.
In the study, 2,684 people each evaluated two live websites from one of ten categories, such as health, finance, travel or news, ranked them for credibility and wrote comments explaining why. The researchers coded the 2,440 comments. The most common topic, present in 46.1% of them, was the “design look” of the site: layout, typography, color, overall visual appeal.
This is a real and useful finding. It is also routinely misquoted as “46% of people judge credibility by design” or “design is 94% of first impressions”. What it measured is narrower: how often design came up in written comments from volunteers comparing two sites.
The authors themselves flag the context. Participants took part to earn a donation for a charity of their choice, not because they needed health advice or a mortgage. The sample was not representative. And the report argues that people with low motivation lean on peripheral cues like appearance; with more at stake, they would likely scrutinize content more. A parallel study by Sliced Bread Design found that health and finance experts cared far more about the quality and sources of information than about looks.
So the honest summary is: when people are not deeply invested, visual design is the most noticeable cue they report using to judge credibility. That describes a lot of first visits.
What to do with this: assume your first-time visitor is in low-motivation mode, skimming between tabs like the study's volunteers. Design for that visitor first, then give the motivated visitor something solid to check: real names, real addresses, real work.
Familiar beats clever
Go back to the top bar of Figure 3, and to the typicality effect in Figure 2. Both point to the same idea, which Jakob Nielsen stated as a law of the web.
In his July 2000 column “End of Web Design”, Nielsen wrote that this means users prefer your site to work the same way as all the other sites they already know. It is a principle drawn from observation and usability testing, not a single experiment. The research above gives it experimental support: familiar layouts look easier to use and are judged more beautiful at a glance.
The pages your visitors use every day have taught them where things live. The logo in the top left goes home. The menu sits at the top or behind a three-line icon on a phone. The main action looks like a button. Prices are on a page called “Prices” or “Pricing”.
When you break those expectations, the visitor has to learn your site before they can learn about your business. On a first visit, that is a tax they rarely agree to pay.
There is a fair objection: if everyone follows convention, how do you stand out? Hekkert's “most advanced, yet acceptable” answer applies. Stand out in what you say, show and promise. Keep the controls where people expect them.
What to do with this: open the three websites your customers most likely visited before yours (competitors, the booking platform they use, a big brand in your field). Note where each puts the name, menu, main action, prices and contact. Wherever they agree, follow them. Wherever they differ, you have room to choose.
Impressions the law sees
A first impression can make a page look more trustworthy than the business is. That is where design psychology meets consumer law, because regulators judge the overall impression, not just whether each sentence is technically true.
Article 6(1) treats a commercial practice as misleading if it contains false information or “in any way, including overall presentation, deceives or is likely to deceive the average consumer, even if the information is factually correct”, about things like the main characteristics of the product, the price or the trader's qualifications, awards and approvals, and if it is likely to change a decision the consumer would otherwise have taken. Each EU country applies this through national law. The UK has the same idea in section 226 of the Digital Markets, Competition and Consumers Act 2024, in force since 6 April 2025: an overall presentation “may be deceiving even if the information it contains is true”.
Sources: Directive 2005/29/EC (consolidated text, EUR-Lex); DMCC Act 2024 s. 226 (legislation.gov.uk). General information, not legal advice; check the rules in your own jurisdiction.
In the US, the Federal Trade Commission's 1983 Policy Statement on Deception makes the same point from the visitor's side. It notes that accurate information in the body text may not fix a false headline, because reasonable consumers may glance only at the headline. The law assumes people skim, exactly like the research does.
For a small business this rarely means a lawsuit. It means a simple rule: the fast impression and the slow facts should tell the same story. Fake “as seen in” logos, award badges you did not win, stock photos posing as your team or a discount strip that is permanently “this week only” all borrow the halo without earning it. Several of them, such as awards, approvals and price advantages, are named in Article 6 itself.
What to do with this: list every badge, logo, number and photo on your first screen. For each, ask: could I prove this today? If not, remove it. A calm, honest first screen beats a decorated one, both for the glance and for the regulator.
What to change tomorrow
You cannot stop people from judging in a blink. You can make sure the blink has very little to object to, and that the second look confirms it.
Run your own glance test
Show your homepage to three people for one second on a phone, then ask for one word that describes it. Words like “busy”, “old” or “confusing” point to complexity or atypicality; fix those before touching copy. This tests appeal, not understanding; use the three-second test for that.
Delete one thing from the first screen
Remove the weakest element above the fold: a second button, a badge row, a news ticker, a decorative graphic. Less complexity is the most robust finding in this chapter.
Put things where people expect them
Name top left, linking home. Menu top right or behind a standard icon. One primary action that looks like a button. Contact and prices findable from the menu.
Raise the contrast and lower the effort
Dark text on a light background, one typeface, a clear size difference between headline and body, images that are recognizable at thumbnail size.
Make the halo honest
Check every claim, badge and photo on the first screen against what you can prove. Replace stock “team” photos with real ones, or with none.
Give the second look something to check
Under the first screen, add the evidence a motivated visitor wants: real work, a real address, named people, clear prices. The glance gets them to stay; this gets them to act. The next chapter, anatomy of a hero, covers how to build that first screen.
What the research supports
Strong: people form consistent visual-appeal judgments within 50 ms in aggregate (Lindgaard 2006), and design properties shape ratings at exposures as short as 17 ms (Tuch 2012); simpler, more typical pages are rated more appealing; perceived beauty and perceived usability are correlated, even after use. Moderate: brief-exposure trust and usability judgments are largely driven by appeal (Lindgaard 2011); fluency raises liking and belief when other information is lacking. Weak or misquoted: that people “decide” or “buy” in 50 ms; that “46% judge credibility by design” (it was 46.1% of written comments from low-motivation volunteers). None of these lab studies measured conversions.
Sources
- Lindgaard, Fernandes, Dudek & Brown (2006). Attention web designers: You have 50 milliseconds to make a good first impression! Behaviour & Information Technology, 25(2), 115–126. Full paper consulted for sample sizes, stimulus selection, the r = .947 result and inter-rater correlations.
- Lindgaard, Dudek, Sen, Sumegi & Noonan (2011). An exploration of relations between visual appeal, trustworthiness and perceived usability of homepages. ACM Transactions on Computer-Human Interaction, 18(1). Abstract consulted.
- Tuch, Presslaber, Stöcklin, Opwis & Bargas-Avila (2012). The role of visual complexity and prototypicality regarding first impression of websites. International Journal of Human-Computer Studies, 70(11), 794–811. Author preprint consulted for methods, samples, results and stated limitations.
- Hekkert, Snelders & van Wieringen (2003). “Most advanced, yet acceptable”: Typicality and novelty as joint predictors of aesthetic preference in industrial design. British Journal of Psychology, 94(1), 111–124. Abstract consulted.
- Reber, Schwarz & Winkielman (2004). Processing fluency and aesthetic pleasure: Is beauty in the perceiver's processing experience? Personality and Social Psychology Review, 8(4), 364–382. Abstract consulted.
- Reber & Schwarz (1999). Effects of perceptual fluency on judgments of truth. Consciousness and Cognition, 8(3), 338–342. Abstract consulted.
- Kurosu & Kashimura (1995). Apparent usability vs. inherent usability: experimental analysis on the determinants of the apparent usability. CHI '95 Conference Companion, 292–293. Full short paper consulted for the method and all correlations in Figure 3.
- Tractinsky, Katz & Ikar (2000). What is beautiful is usable. Interacting with Computers, 13(2), 127–145. Publisher's abstract consulted.
- Thorndike (1920). A constant error in psychological ratings. Journal of Applied Psychology, 4(1), 25–29. Full text consulted for the aviation-cadet correlations.
- Fogg, Soohoo, Danielson, Marable, Stanford & Tauber (2003). How do users evaluate the credibility of Web sites? A study with over 2,500 participants. Proceedings of DUX 2003. The full 2002 Consumer WebWatch report was consulted for the method, the 2,440 comments, the category percentages and the authors' discussion of motivation.
- Nielsen (2000). End of Web Design. Nielsen Norman Group, 22 July 2000. Source of Jakob's law.
- Directive 2005/29/EC (Unfair Commercial Practices Directive). Consolidated text, Article 6. EUR-Lex.
- Digital Markets, Competition and Consumers Act 2024. Section 226, Misleading actions, and commencement note (in force 6 April 2025). legislation.gov.uk.
- Federal Trade Commission (1983). FTC Policy Statement on Deception, 14 October 1983. Full text consulted for the passage on headlines and fine print.
Sources checked 18 October 2026. Laboratory studies used screenshots or simulated machines and measured ratings, not sales. Figures 1, 3 and 4 redraw published numbers; Figure 2 and the demo pages are original, fictional illustrations. The pottery studios and their visitor are invented.