Web design that earns attention and holds it

Design is a commercial decision before it is an aesthetic one

There is a version of web design that is entirely about taste, and it produces sites that win awards and lose enquiries. There is another version that treats the page as a piece of commercial machinery, and it produces sites that convert well and look like a template. The useful work sits between the two: a site that is genuinely well made and does its job, where the restraint is deliberate rather than a shortage of ideas.

That is the standard we work to. It starts by being specific about what the site is supposed to change. “Modernise our brand” is not a brief anyone can design against or measure. “Get more qualified enquiries from mid-market manufacturers who currently call three competitors before deciding” is — and it immediately tells you what the homepage has to prove, what a service page has to answer, and why a case study matters more than an office photograph.

What actually loses you the visitor

After enough audits, the same handful of failures recur, and almost none of them are about how the site looks.

The value proposition is implied rather than stated. The homepage says “Innovative solutions for a connected world” where it could say what the company does and for whom. Visitors will not work it out; they will leave and check the next result.

The page is slow to show anything. Largest Contentful Paint is the moment the main content appears, and it is heavily influenced by design decisions — a full-bleed hero image, a webfont that blocks rendering, a carousel that loads four slides to show one. Google’s threshold for a good LCP is 2.5 seconds at the 75th percentile of real visits, and hero-image-led designs routinely miss it on mobile connections.

The layout moves while it loads. Cumulative Layout Shift is what happens when images have no declared dimensions or a webfont swaps to a differently-proportioned face. It reads to a visitor as cheapness even when they cannot name what went wrong, and it makes people tap the wrong thing.

Navigation reflects the org chart. Menus built around internal departments rather than what visitors are trying to do. Card sorting with actual users resolves this quickly and cheaply, and it is one of the few research methods that pays for itself in a single afternoon.

The form asks for too much too early. Every additional required field is a reason to abandon. Ask for what you need to have a first conversation, and nothing more.

Accessible design is not a separate workstream

Roughly one in six people worldwide lives with a significant disability, and the accessibility measures that serve them serve everyone else too — larger touch targets, sufficient contrast in bright sunlight, captions on a video played on a silent commute, a form that can be completed without a mouse.

Treating accessibility as a design constraint rather than a remediation phase is also simply cheaper. Fixing a contrast failure in a design token takes a minute. Fixing it after two hundred pages have been built takes a sprint. So we check contrast when the palette is chosen, define focus states when components are designed, and specify heading structure when wireframes are drawn — not in a pre-launch audit that discovers the brand’s primary colour was never legible as body text.

The floor is WCAG 2.2 AA: contrast ratios of at least 4.5:1 for body text and 3:1 for large text and interface components, every function operable by keyboard, a visible focus indicator that meets the newer focus-appearance criteria, and no content that depends on colour alone to be understood.

The system matters more than the pages

A website is never finished. Someone will add a landing page for a campaign, a new service page, a job posting. Whether the site still looks coherent in two years depends almost entirely on whether it was designed as a system or as a set of pages.

A design system for a site this size is not a heavyweight artefact. It is: tokens for colour, type, spacing, radii and elevation; a documented set of components with every state drawn; and a short set of rules about when to use which. That is enough for a marketer to build a new page that looks like it belongs, without a designer in the loop.

The alternative is what most audits find — four button styles, three card treatments, two type scales, and a navigation that behaves differently in one section because it was built by a different agency in 2022.

Designing for the devices people actually use

“Responsive” has come to mean “does not break on a phone”, which is a low bar. Designing mobile-first means deciding what matters most when you only have a small viewport and an impatient user on a train — and that decision usually improves the desktop layout too, because it forces a hierarchy.

Practically: touch targets at least 24 by 24 CSS pixels with adequate spacing, thumb-reachable primary actions, no hover-dependent functionality, forms with the right input types so the correct keyboard appears, and text that is readable without zooming. We design at three breakpoints and test on real devices, because a browser window narrowed to 390 pixels is not a phone.

Redesign or new build: how to tell which you need

Most enquiries arrive asking for a redesign. A meaningful share of them do not need one.

A redesign is the right answer when the structure is wrong — when the navigation reflects your org chart rather than what visitors are trying to do, when there is no coherent template system, when the site cannot be edited without a developer, or when the underlying build is slow in ways no amount of tuning will fix. Those are structural problems, and structural problems need structural solutions.

A redesign is the wrong answer when the real problem is content, measurement or a single broken journey. We have seen sites where the entire “we need a new website” conversation was actually “our pricing page does not explain what is included” — a two-day fix that a six-figure rebuild would have obscured rather than solved. If your analytics show people reaching the right pages and leaving, look hard at what those pages say before you replace them.

The diagnostic is cheap. Look at which pages currently earn traffic and enquiries, and ask whether the problem is that people cannot find them, or that they find them and are not persuaded. The first is a structure problem. The second is a content problem wearing a design costume.

What a redesign must not break

If the site already ranks, a redesign carries an obligation a new build does not: preserving what works. That means auditing which URLs earn traffic before touching them, mapping every existing URL to its destination with a single-hop 301, keeping or improving the content that currently performs, and benchmarking speed before and after so a “modern” rebuild does not quietly ship slower than what it replaced.

We have seen redesigns lose half a site’s organic traffic because nobody mapped the URLs, and the agency involved had genuinely not considered it their job. Ask whoever quotes you whether migration planning is in scope, and get the answer in writing.

What goes above the fold, and how we decide

“Above the fold” is a slightly outdated term — there is no single fold across the device landscape — but the underlying question is real: what does a visitor see in the first second, and does it answer the question they arrived with?

The answer is almost never a carousel. Rotating hero banners are one of the most reliably tested-against patterns in interface design: they push content down, they cost a second of load time, most people never see slide two, and their existence usually signals that nobody could agree on a message. When a client asks for one, the honest reading is that there is an unresolved internal disagreement, and the useful work is resolving it rather than building the carousel.

What earns the space instead: a headline stating what you do and for whom in language a customer would use, one supporting sentence, and one primary action. Everything else can be below it. This is unglamorous and it consistently outperforms the alternative.

Designing the content model, not just the page

Somewhere between wireframes and visual design, a decision gets made about what fields a page type has — and that decision determines whether your team can maintain the site afterwards.

A service page that is one large rich-text field will drift, because every editor makes a different structural choice. The same page modelled as discrete fields — a heading, a standfirst, a list of deliverables, an FAQ block — stays consistent no matter who edits it, and it makes the content reusable elsewhere on the site.

This is design work, not just development work, and it is the difference between a site your marketing team can run and one they have to raise a ticket against. We define the model during design and hand it over as part of the specification.

What actually drives the cost

We do not publish prices, because a figure produced before the scope is known is a guess that somebody eventually has to absorb. But the variables are not a secret, and knowing them lets you brief better and compare quotes honestly.

The number of genuinely unique templates. This is the single biggest driver, and it is not the same as the number of pages. A forty-page site with five templates is considerably cheaper than a nine-page site where every page is bespoke. Before asking anyone to quote, count your distinct layouts rather than your pages.

Whether research is needed. If you have analytics, Search Console access and a sales team who can describe what buyers ask, that may be enough. If you are entering a new market or serving an audience nobody internally has spoken to, it is not, and skipping the research just moves the cost to being wrong later.

Whether a design system exists. Extending a documented system is a fraction of the cost of building one. If you have brand guidelines that only cover print, expect to pay for the digital extension — type scale at small sizes, accessible colour pairings, focus states, motion.

How much content exists. Designing against real content is non-negotiable for us, so if the copy does not exist, someone is writing it. That is either a cost or a dependency on your calendar, and pretending otherwise is how projects slip.

Integrations and edge cases. A booking flow, a member area, a product configurator, multi-language. Each adds states that have to be designed — empty, loading, error, partial, permission-denied — and states are where design budgets actually go.

How this connects to everything else

Design rarely arrives alone. A new site usually needs someone to build it properly, and if it sells directly then e-commerce development brings its own set of design problems around checkout and product discovery. If your visitors sign in and use the thing repeatedly, what you actually need is UI/UX design — different discipline, different definition of success. And if the site currently ranks, the redesign needs technical SEO running alongside it so the migration does not cost you the traffic you already have.

We would rather tell you which of those you need than sell you all of them.