# Don't Design for Designers: Why Beautiful Websites Fail to Convert > Award-winning web design can quietly destroy clarity, speed and sales. The data-backed case for building websites around customers, not other designers. URL: https://www.bravebrand.com/learn/brave-branding/dont-design-for-designers Author: Luke Carter · Published: Aug 11, 2026 You can spend $10,000 on a website that looks award-worthy, then watch qualified customers leave before they understand what you sell. Modern web tools have made visual spectacle almost effortless. A designer can now build scroll-driven video, WebGL scenes, custom shaders, dynamic canvases and elaborate micro-interactions without a large engineering team. That is an extraordinary creative achievement. It has also created a dangerous illusion: that the website which looks the most advanced must be the one doing the most valuable work. It often is not. Digital galleries celebrate aesthetic innovation. Peer-judged platforms reward technical ambition. Social feeds turn dramatic transitions and experimental navigation into thirty-second showreels. Yet the customer arriving on the actual website is usually trying to answer something painfully ordinary: What do you do? Is this for me? Can I trust you? What should I do next? When design makes those questions harder to answer, visual sophistication stops being an asset. It becomes friction. This is the case for a simple principle: **do not design for designers. Design for the person whose decision pays for the website.** ## The emergence of aesthetic narcissism The democratisation of advanced front-end tools has dramatically lowered the barrier to producing visually dense web experiences. Individual creators can now configure sticky animation stacks, particle systems, canvas interactions and responsive motion without writing an entire rendering engine from scratch. Consequently, design galleries are full of websites characterised by dark-mode glassmorphism, variable typography, ambient video, custom cursors, horizontal scroll, experimental menus and objects that rotate because your pointer happened to pass nearby. The problem is not that these techniques exist. The problem is the incentive system that decides when to use them. When mechanical constraints disappear, design can quietly pivot from solving a business problem to seeking validation from professional peers. Call it **aesthetic narcissism**, or peer-driven design: work that increasingly performs for the people who understand how difficult it was to make, rather than for the people who need to use it. There is a useful comparison in competitive bodybuilding. Someone may begin training to become healthier or more attractive. As the subculture takes over, extreme muscularity earns more admiration from other bodybuilders while becoming less appealing to the broader public the person originally hoped to impress. Web design can follow the same curve. Technical excess wins compliments from other designers who recognise the craft. The end customer, who wants immediate clarity and low-friction utility, is left staring at a loading sequence or trying to discover which abstract shape is the menu. This creates a direct conflict between agency incentives and client outcomes. An agency can use a client budget to produce a portfolio piece, attract industry praise and win the next project even when the current client receives fewer leads, registrations or sales. The work succeeds inside the designer's world while failing inside the customer's. ## The two loops: peer validation versus customer reality The divergence becomes clearer when you separate two feedback loops. ### The peer-validation loop - A designer creates a WebGL-heavy site with experimental navigation. - The site is clipped into a showreel and submitted to design galleries. - Other designers praise its novelty and technical execution. - The next project pushes the visual complexity even further. ### The customer-reality loop - A prospective customer lands on the same site, often on a phone. - Large assets and JavaScript delay the message they came to see. - Scroll hijacking and ambiguous copy make the page harder to scan. - The visitor leaves without understanding the offer or taking action. Only one of these loops produces applause. Only the other produces revenue. ## The financial cost of visual bloat The first casualty of aesthetic over-engineering is technical performance. Multiple variable-font files, uncompressed video, 3D render pipelines, canvas instances and heavy scroll listeners all compete for bandwidth and processing time before a visitor can do anything useful. [Portent analysed more than 100 million page views](https://portent.com/blog/analytics/research-site-speed-hurting-everyones-revenue.htm) across 20 B2B and B2C websites, covering more than 27,000 landing pages and a page-speed sample of 5.6 million sessions. Six sites were B2C e-commerce businesses and 14 were B2B lead-generation businesses. The results make the commercial trade-off hard to ignore. ### B2B lead generation - At a one-second load time, the average conversion rate was almost **40%**. - At two seconds, it fell to **34%**. - At three seconds, it fell again to **29%**. - A one-second site converted **three times better** than a five-second site. - A one-second site converted **five times better** than a ten-second site. ### B2C e-commerce - One second: **3.05%** conversion. - Two seconds: **1.68%** conversion. - Three seconds: **1.12%** conversion. - Four seconds: **0.67%** conversion. Portent modelled what those figures mean for 1,000 visitors buying a $50 product. At a one-second load time, the page produces **$1,525**. At four seconds, it produces only **$335**. That is more than **$1,190** in lost potential sales, or a roughly **78% reduction**, with the same traffic and the same offer. The important lesson is not that every company will reproduce those exact percentages. Conversion studies differ by audience, traffic source and offer. The lesson is that speed is not a technical decoration applied after the design is approved. It is part of the commercial design. ## A tenth of a second can change the economics The Google-commissioned Deloitte study [Milliseconds Make Millions](https://www.deloitte.com/ie/en/services/consulting/research/milliseconds-make-millions.html) analysed mobile-site data across retail, travel, luxury and lead-generation brands in Europe and the United States over four weeks. Following a natural **0.1-second improvement** in mobile site speed, Deloitte observed: - **8.4%** higher retail conversions and **9.2%** higher average order value. - **10.1%** higher travel conversions and **1.9%** higher average order value. - **8.6%** more page views per session for luxury brands. - An **8.3%** improvement in bounce rate on lead-generation information pages. An older Amazon benchmark, attributed to former Amazon engineer Greg Linden, claimed that every additional 100 milliseconds of latency reduced revenue by approximately **1%**. It is best treated as an influential historical benchmark rather than a universal law, but its conclusion has been repeatedly echoed by newer field data: customers respond to speed. Google's mobile research found that as load time increases from one to three seconds, bounce probability rises by **32%**. At five seconds, it rises by **90%**. Meanwhile, mobile phones accounted for approximately [64.35% of global web traffic in July 2025](https://soax.com/research/mobile-website-traffic). A design that is impressive on a founder's new laptop but exhausting on a mid-range phone is not a premium experience. It is a broken priority. ## Scrolljacking and the loss of user control Performance is only one layer of the problem. Over-designed websites also fail by changing the basic interaction rules people already understand. Scrolljacking, or scroll hijacking, overrides the browser's native scroll behaviour. A physical gesture may trigger a fixed animation sequence, change the speed of movement, pin a scene in place or send the visitor sideways instead of down. The technique can produce a dramatic presentation. It can also strip away control and predictability. Under normal conditions, a gesture maps directly to document movement. When that relationship is replaced by an animation timeline, visitors can become disoriented: they do not know how far they have travelled, when the page will release them or how to return to the information they just passed. The accessibility cost can be worse. Custom interactions often ignore arrow keys, Spacebar and Page Down, or fail to communicate state changes to assistive technology. The [W3C Web Accessibility Initiative](https://www.w3.org/WAI/fundamentals/accessibility-principles/) is explicit that functionality available with a mouse must also be available from a keyboard, focus must not become trapped and moving or scrolling content must be controllable. Animation is not inherently inaccessible, and scroll-driven storytelling is not automatically bad. But when a decorative interaction breaks native navigation, the burden of proof belongs to the decoration. ## People scan websites; they do not attend them A website is not a cinema. The visitor did not buy a ticket, silence their phone and agree to experience the designer's sequence from beginning to end. They scan. [Nielsen Norman Group eye-tracking research](https://www.nngroup.com/articles/f-shaped-pattern-reading-web-content/) shows that one common behaviour is an F-shaped scan: people read more across the top, move down and scan shorter horizontal sections, then continue along the left side. Other patterns include scanning headings, skipping large blocks and searching specifically for links, numbers or key phrases. The exact pattern changes with the task and the layout, but the principle is stable: most visitors are trying to find an answer with the minimum effort required. Artificial scroll pauses and hover-to-reveal layouts interfere with that information extraction. They force a visitor through the designer's choreography when the visitor wants a price, a service description, proof or a button. Hick's Law helps explain the resulting friction. In controlled choice tasks, decision time rises approximately logarithmically with the number of available alternatives. A landing page full of competing motion cues, floating controls, ambient particles and non-standard icons creates more things to interpret before the core decision can even begin. When every element moves, nothing has hierarchy. The visitor's cognitive effort is spent navigating the interface instead of evaluating the value proposition. ## What actually converts The alternative is not ugly design. It is disciplined design: strong visual hierarchy, fast performance, explicit value, human proof and an obvious next step. Several well-known conversion experiments show what happens when those priorities move to the front. ### Highrise: 102.5% more sign-ups 37signals, now Basecamp, tested a radical redesign of the Highrise CRM homepage. The original information-heavy page was replaced by a much simpler, person-focused page with a large customer photograph, a short testimonial and a clear action. The result was a [**102.5% lift in account sign-ups**](https://www.abtestcases.com/testcases/testimonial-redesign/). Paid sign-ups increased by **47%**. The same company also tested the headline on its sign-up page. Replacing “Start a Highrise Account” with a headline centred on a “30-day Free Trial on All Accounts” produced a [**30% improvement**](https://signalvnoise.com/posts/1525-writing-decisions-headline-tests-on-the-highrise-signup-page). The visual design did not need to shout louder. The offer needed to become clearer. ### Nature Air: 591% more conversions Nature Air tested the placement of contextual calls to action across 17 destination pages. Making the action prominent inside the content increased conversion from approximately **2.78% to 19%**, a [**591% relative increase**](https://www.blastanalytics.com/nature-air-case-study). The broader optimisation programme contributed to a **70% increase in web sales** compared with the same month in the prior year. The offer had not suddenly become 591% more valuable. The route to it had become visible. ### EA's Sims 3: 128% more registrations Electronic Arts tested the value proposition in the Sims game launcher. The winning approach made the registration benefit specific and tangible instead of relying on an unfocused portal and abstract rewards. Community registrations increased by [**128%**](https://conversion.com/blog/how-to-create-an-awesome-value-proposition/). ### Underwater Audio: 35.6% more sales Underwater Audio adjusted the visual hierarchy of a landing page so visitors encountered the relevant testimonial before the call to action. That straightforward change produced a [**35.6% increase in online sales**](https://vwo.com/landing-page-optimization/). None of these examples proves that one layout, photograph or CTA position will work for every company. That is what testing is for. Together, however, they reinforce a consistent point: people convert when the design helps them understand the value and act on it. ## A practical “don't design for designers” framework ### 1. Apply the five-second clarity rule Show the page to someone outside the company for five seconds, then hide it. Ask them four questions: - What does this company offer? - Who is it for? - What outcome does it create? - What should you do next? If they cannot answer, do not add another animation. Fix the message and hierarchy. ### 2. Establish a mobile performance budget Google recommends a Largest Contentful Paint of [**2.5 seconds or less**](https://web.dev/articles/lcp) at the 75th percentile, measured separately across mobile and desktop. Treat that as the ceiling, not the ambition. For conversion pages, move as close to the one-to-two-second range as the content permits. Set budgets for image weight, JavaScript, font files and third-party scripts before visual production. Every new effect spends from that budget and must earn its cost. ### 3. Preserve native browser behaviour Keep vertical scrolling predictable. Keep links recognisable. Make every function available by keyboard. Respect reduced-motion preferences. Do not replace the pointer unless the replacement improves the task. Do not hide navigation behind an unexplained symbol simply because the symbol looks cleaner. ### 4. Evaluate every visual element by outcome For each animation, video, canvas and interaction, ask: - Does this make the offer easier to understand? - Does it increase trust? - Does it demonstrate something words cannot? - Does it guide attention towards the next useful action? - Is the gain worth the performance and accessibility cost? If the honest answer is “it will look great in our portfolio”, the client should not be paying for it. ### 5. Test against business metrics, not taste Measure qualified leads, registrations, revenue per visitor, task completion, bounce behaviour and mobile performance. A design review can tell you whether the team likes a page. Only real behaviour can tell you whether the page works. ## Beautiful is not the enemy This argument is not a defence of bland websites. Craft matters. Emotion matters. Distinctive art direction can signal quality, create memory and make an unfamiliar offer easier to trust. The distinction is purpose. Good design uses beauty to clarify. Bad design uses the customer as an audience for beauty. Good motion directs attention. Bad motion demands attention. Good technology removes effort. Bad technology asks the visitor to admire the effort. Modern web tools are force multipliers. Applied to a clear commercial strategy, they can create remarkable experiences that are fast, usable and persuasive. Applied to an unclear strategy, they simply make the confusion more expensive. The best website is not the one other designers wish they had made. It is the one customers understand, trust and act on. **Design for them.** ## Research note This article was adapted from a longer BraveBrand research brief. Conversion figures are reported from the linked studies and case reports; they should be interpreted in the context of each experiment rather than treated as guaranteed universal lifts. An unsupported claim in the raw brief, that more than 94% of B2B landing pages fail a five-second test, was deliberately excluded because no auditable dataset could be located. Portent's contradictory five- and ten-second e-commerce rows were also excluded in favour of the figures explicitly supported in its current published analysis.