Websites

The web design guide: sites that convert

Most business websites fail quietly. They load, they look acceptable, and almost nobody contacts anyone. This guide is about the specific reasons that happens and what to do instead.

A wide grid of identical pale plaster tiles rising over one smooth shallow wave, raked by warm light across the crest.
In short

What makes a business website actually produce enquiries?

A business website produces enquiries when a visitor can tell within seconds who you help and what to do next, when the page loads fast enough that they are still there to read it, and when contacting you takes one obvious action rather than a search through a menu.

Everything else serves those three: design quality, content depth, technical build and accessibility all matter because they remove reasons for a visitor to leave, not because a visitor admires them.

The most common failure is not ugliness, it is a site that describes the business rather than addressing the visitor's problem and buries the phone number in a contact page two clicks away.

The first screen

A visitor decides before they read

Someone arriving from a search has a question in their head and very little patience. The first screen either answers it or it does not, and almost nothing further down the page can recover a visitor who has already left.

The answer they need is narrower than most businesses assume. It is not your history, your values or your team photo. It is whether you do the specific thing they searched for, whether you do it where they are, and how they get hold of you. A page that leads with those three, plainly, outperforms a more beautiful page that leads with a mission statement.

Read the full breakdown: A visitor decides before they read3 more paragraphsHide the full breakdown: A visitor decides before they read

This is why home pages are usually the wrong destination for a specific search. A visitor searching for a specific service and landing on a general home page has to navigate before they can act, and navigation is friction. A service page that continues the promise of the search removes that step entirely.

It is also why one call to action beats four. When a page offers to call, book, download, subscribe and chat all at once, it has moved the decision of what to do next onto the visitor, and a visitor who has to decide often decides to leave. Pick the action you most want, make it the most prominent thing, and let the others be secondary.

None of this is a design opinion. It is what the analytics of a failing site usually show: traffic arriving, a short time on page, and a contact form that almost nobody reaches. The fix is structural, not decorative.

The route to an enquiry

What a visitor has to get through

Between a click and a booked job there are perhaps five moments where someone can quietly give up. Each is fixable, and knowing which one is costing you is most of the work.

From a search to a booked jobA path running left to right: a search, then your page, then a branch into either a phone call or a form and chat, then a booked job. A faint branch drops away from the page to show the people who leave instead.FROM A SEARCH TO A BOOKED JOBSEARCHA QUERY WITH INTENTYOUR PAGEPROOF AND A NEXT STEPCALLFORM OR CHATBOOKEDTRACKED TO ITS SOURCELEAVESNOT EVERY CLICK CONVERTS.THE PAGE’S JOB IS TO LOSE FEWER OF THEM.EVERY STEP IS A PLACE TO LOSE SOMEONE, OR A PLACE TO MAKE IT EASIER.
Arrival, comprehension, trust, action, and follow-up. A site that fails at any one of them produces traffic and no enquiries.

Arrival is speed and stability. If the page takes too long, or if it jumps around while loading and the visitor taps the wrong thing, the rest of the design never gets a chance.

Comprehension is the first screen doing its job. Who you help, what you do, where you do it, stated in plain language rather than in the industry's internal vocabulary.

Read the full breakdown: What a visitor has to get through3 more paragraphsHide the full breakdown: What a visitor has to get through

Trust is the part most sites underinvest in. Real photographs of the actual premises and people, genuine reviews, credentials and licences, a real address, clear pricing where you can publish it. Stock imagery and generic claims do the opposite of what they are meant to.

Action is friction. A tappable phone number on mobile, a form that asks for the minimum you actually need, and a booking option for businesses where booking is the natural next step. Every extra required field costs you some percentage of the people who started.

Follow-up is what happens after. An enquiry that reaches a mailbox nobody checks until Monday is, from the customer's point of view, identical to a website that did not work.

Speed, measured

What a tenth of a second was worth

In 2020 Google commissioned a study conducted by Deloitte and 55, monitoring mobile load times across 37 brand sites and more than 30 million user sessions. It measured what happened when mobile load time improved by 0.1 seconds. The study is now several years old, so read it as evidence that speed has a real commercial effect rather than as a current benchmark.

8.4%Retail conversion lift from 0.1s faster load (2020 study)
10.1%Travel conversion lift from the same change
21.6%Lift in progression to lead generation form submission
9.2%Retail average order value lift

SourceGoogle, Deloitte and 55, Milliseconds Make Millions, 2020 (37 sites, 30m+ sessions)

The lead generation figure is the one that matters most for a service business, because it measures people getting further through a form rather than buying something. It is also the clearest argument for treating speed as a conversion investment rather than a technical nicety.

The current standard

Core Web Vitals, and what the numbers mean

Google publishes three metrics with specific thresholds, and they are the most useful shared vocabulary available for talking about site performance without hand-waving.

Core Web Vitals as three gaugesThree half-circle gauges, one each for Largest Contentful Paint, Interaction to Next Paint and Cumulative Layout Shift. Each gauge carries a poor, needs work and good zone, and a needle resting somewhere along it.THREE THINGS GOOGLE MEASURES ABOUT THE PAGE EXPERIENCELCPLARGEST CONTENTFUL PAINTHOW FAST THE MAIN THING LOADSINPINTERACTION TO NEXT PAINTHOW FAST IT REACTS TO A TAPCLSCUMULATIVE LAYOUT SHIFTHOW MUCH THE PAGE JUMPS ABOUTPOORNEEDS WORKGOODSCHEMATIC GAUGES. REAL SCORES COME FROM FIELD DATA FOR YOUR OWN PAGES.
Largest Contentful Paint, Interaction to Next Paint and Cumulative Layout Shift, each with a published threshold for good.

Largest Contentful Paint measures how long until the main content of the page has rendered. Google's threshold for good is 2.5 seconds or less. In practice this is usually decided by your hero image, your web font loading strategy and your server response time, in that order.

Interaction to Next Paint measures responsiveness: how long the page takes to visually respond after someone taps or clicks. The threshold for good is 200 milliseconds or less. INP replaced First Input Delay as a Core Web Vital, so any advice referring to FID as current is out of date.

Read the full breakdown: Core Web Vitals, and what the numbers mean3 more paragraphsHide the full breakdown: Core Web Vitals, and what the numbers mean

Cumulative Layout Shift measures unexpected movement while the page loads, the effect where you go to tap a link and an advert or an image pushes it out from under your finger. The threshold for good is 0.1 or less, and the usual causes are images without dimensions set and content injected after the initial render.

Two details matter for judging any report you are shown. These thresholds are assessed at the 75th percentile of page loads, which means three quarters of visits need to hit them, not the average visit. And they are segmented by mobile and desktop, which is why a site can pass comfortably on a desktop test and fail for most of its actual visitors.

That last point is worth dwelling on. A score from a single test run on a fast connection is a laboratory measurement. Field data, gathered from real visits, is what Google uses and what your customers experience. When those two disagree, believe the field data.

The build, in order

How to build or rebuild a site that works

This is the sequence we follow. The reason design is not first is that designing before you know what the pages have to say produces a site that looks finished and says nothing.

  1. Decide what the site has to achieve, in numbers

    Before anything visual, agree what a successful site does: the enquiry types that count, roughly how many of them would represent a good month, and which services you most want more of. Everything downstream becomes a decision you can make rather than an opinion you can argue about.

    • The specific actions that count as an enquiry
    • The services you want more of, ranked
    • The areas you serve, and which ones you want to grow
  2. Plan the pages around what people search for

    Map the pages to real questions and real services rather than to your internal structure. Each service you sell separately deserves its own page, because a combined page cannot rank for or answer any of them properly.

    • One page per service you would quote separately
    • Location pages only where you can write something genuinely local
    • Supporting answer pages for the questions customers ask before buying
  3. Write the content before the design

    Design built around real copy fits real copy. Design built around placeholder text produces a layout the actual content then has to be squeezed into, which is how sites end up with three-word headings that say nothing and service descriptions truncated to fit a box.

    • Real headings, real service descriptions, real proof
    • Plain language, checked against how customers describe the problem
    • Nothing claimed that cannot be substantiated
See the remaining steps: How to build or rebuild a site that works3 more stepsHide the remaining steps: How to build or rebuild a site that works
  1. Design mobile first, and check on a real phone

    Google predominantly uses the mobile version of a page for indexing and ranking, and for most local businesses the majority of visits are on a phone. Designing the desktop layout first and adapting it downward reliably produces a mobile experience that feels like an afterthought, because it is one.

    • The primary action reachable with one thumb, without scrolling
    • Tap targets large enough to hit accurately
    • Tested on an actual device, not only in a browser's device emulator
  2. Build to WCAG 2.1 AA, and test it

    Conformance is a build standard, not a plugin. It means sufficient colour contrast, every image carrying meaningful alternative text, every form field labelled, a logical heading order and everything reachable and operable by keyboard alone.

    • Contrast checked against the 4.5 to 1 ratio for normal text
    • Keyboard navigation tested end to end, including every form
    • Headings in order, with no levels skipped
  3. Instrument it before launch, then keep working on it

    Analytics, call tracking and form tracking should be in place on day one, so the first month produces data rather than impressions. Then treat the site as something that gets better, because the pages that matter are rarely the ones anyone predicted.

    • Calls, forms and chats measured separately from the start
    • Core Web Vitals monitored with field data, not just lab scores
    • A regular look at which pages produce enquiries and which do not

Timelines vary with scope, content readiness and how many people have to approve things. The honest driver of a schedule is usually whether the content and photography exist, not how long the build takes.

Accessibility

How common accessibility failures actually are

The WebAIM Million runs automated accessibility testing across the top one million home pages. Its 2026 analysis found detectable failures on almost all of them. These are automatically detectable barriers only, and WebAIM is explicit that true conformance is worse than these numbers suggest, because many failures cannot be caught by a machine.

95.9%Home pages with detected WCAG failures
56.1Average detected errors per home page
83.9%Home pages with low contrast text
53.1%Home pages missing image alternative text

SourceWebAIM Million 2026 (automated analysis of 1,000,000 home pages)

Six error types account for the large majority of everything detected, which is genuinely good news: most of the problem is a small number of repeated, fixable mistakes rather than a thousand different ones. Low contrast, missing alternative text, unlabelled form fields, empty links and empty buttons are most of it.

The minimum

What a small business site genuinely needs

Not a wish list. This is the set that, in our experience, separates sites that produce enquiries from sites that produce compliments.

  • A home page that states who you help, what you do and where, above the fold
  • One page per service you would quote separately, written properly
  • A tappable phone number visible on every page on mobile
  • A form that asks only for what you genuinely need to respond
  • Real photographs of your premises, your team and your actual work
See the full checklist: What a small business site genuinely needs7 more itemsHide the full checklist: What a small business site genuinely needs
  • Genuine reviews, displayed without being edited into shape
  • Your full name, address and phone, matching your Google Business Profile exactly
  • Location or service area pages with real local substance, not a city name swapped
  • Answers to the questions customers ask before they are ready to buy
  • A privacy notice that reflects what your site actually collects
  • Analytics, call tracking and form tracking, configured and checked
  • A maintenance arrangement covering updates, backups and monitoring

The penultimate item is the one most often missed after launch, and the last is the one most often skipped entirely. A site nobody updates becomes a security problem long before it becomes a design problem.

The recurring question

Template or custom

This gets framed as a quality question and it is really a fit question. Both produce good sites and bad sites. We have deliberately left figures out, because every published cost range for this comparison traces back to agency blogs rather than to any survey.

ItemTemplate or page builderCustom build
Best forA clear, conventional site with standard needsSpecific functionality, scale, or performance requirements
Speed to launchFaster, because the structure already existsSlower, because the structure is being made
Performance ceilingLimited by what the template loadsDetermined by decisions you control
AccessibilityDepends entirely on the template chosenCan be built to the standard deliberately
Editing it yourselfUsually straightforwardDepends on what was built for you
Main riskOutgrowing it, or inheriting its bloatPaying for complexity you did not need

The honest answer for most local service businesses is that a well-chosen, well-built template site with real content will outperform a custom site with weak content, every time. Content and structure decide whether a site works. The platform decides how pleasant it is to live with.

Before you sign anything

Own it, and be able to prove it works

Ownership first, because it is the thing people discover too late. Your domain should be registered to your business, in an account you control. Your site files and database should be yours, and you should be able to take them elsewhere. Your hosting can be arranged by whoever builds the site, and it should not be a hostage. If a proposal cannot answer what you keep if the relationship ends, that is the answer.

This is not a hypothetical concern. The most common version we encounter is a business that cannot update its own site because the only administrative login belongs to a developer who stopped replying two years ago. Nothing about that situation is dramatic, and it is expensive to unwind.

Read the full breakdown: Own it, and be able to prove it works3 more paragraphsHide the full breakdown: Own it, and be able to prove it works

Then proof. A site should be judged on enquiries, not on compliments, and enquiries have to be measured to be judged. That means tracking calls separately from forms, knowing which pages produce them, and being able to see whether a redesign helped or simply looked different.

One honest caveat on before-and-after comparisons. A new site usually changes the URL structure, the tracking setup and sometimes the analytics property all at once, which makes a clean comparison harder than anyone admits. Decide before launch what you will compare, make sure the old numbers are recorded, and keep the definition of an enquiry identical on both sides. Otherwise the first report after launch is an argument rather than a measurement.

Finally, accessibility deserves one plain statement. We build to WCAG 2.1 AA because it is the recognised standard and because the WebAIM data shows how unusual that actually is. What we can say is what we build to and how we test it, which is with a keyboard, a screen reader and a contrast check rather than only an automated scan.

Want to know why your current site is not producing enquiries?

We will look at speed, structure and the path to contact, and tell you what we would change first.

Questions

Straight answers.

What decides how long a website build takes?

It depends far more on content readiness than on build time. A site with copy written, photography taken and decisions made moves quickly. A site waiting on service descriptions, staff bios and approval from four people takes as long as those take.

We will not quote a timeline in a guide, because a number offered before anyone has seen your scope is not a plan. What we will say is that the honest question to ask any supplier is which stage they expect to be waiting on you, and how they handle it when that stalls.

Does my website need to be mobile friendly?

Yes, and the reason is structural rather than stylistic. Google predominantly uses the mobile version of a page for indexing and ranking, so content that is missing or hidden on mobile may effectively not exist for search purposes.

Separately, for most local businesses a large share of visits and almost all phone calls come from a phone. A site that is awkward on mobile is awkward for the majority of the people who see it.

What load time should I be aiming for?

Google's published threshold for good Largest Contentful Paint is 2.5 seconds or less, with Interaction to Next Paint at 200 milliseconds or less and Cumulative Layout Shift at 0.1 or less, all assessed at the 75th percentile of page loads.

Those are the thresholds worth designing against. They are more useful than a single performance score because they describe three different experiences of slowness, and a site can be fine on one and poor on another.

Does my website need to be accessible?

It should be built to WCAG 2.1 AA as a matter of practice. That standard covers contrast, alternative text, form labels, heading order and keyboard operability, and the WebAIM Million's 2026 analysis found detected failures on 95.9% of the top one million home pages, so it is a genuine differentiator rather than a given.

We are describing a build standard, not giving legal advice. If you have specific legal obligations, take advice on those from someone qualified to give it.

Why is my website not getting any leads?

Work through it in order rather than guessing. Is anyone arriving at all, and from what. If people are arriving, do they stay long enough to have read anything. If they stay, do they reach a page where contacting you is possible. If they reach it, does the form work and does the enquiry arrive somewhere a person looks.

That last one is more common than it should be. Broken forms, forms sending to an inbox nobody monitors, and phone numbers that are images rather than tappable links account for a surprising share of sites that appear to have no demand.

Should I use WordPress, Shopify, or something else?

Choose based on what the site has to do. Content-led sites for service businesses are well served by WordPress, which is why so much of the web runs on it. Selling physical products is what Shopify is built for and it handles the awkward parts you would otherwise build twice.

The platform matters less than people expect. A well-built site on either will outperform a poorly built site on the other, and the decisions that determine whether a site works are mostly about structure, content and speed rather than about the underlying software.

How often does a website need redesigning?

Less often than agencies suggest and more often than most businesses do it. The genuine triggers are a change in what you sell, a structure that no longer fits the business, performance or accessibility problems that cannot be fixed incrementally, or an underlying platform that is no longer maintained.

Wanting it to look more current is a real reason too, and it is usually better served by a refresh of the visual layer than by rebuilding a structure that works. A full rebuild resets your URLs and your measurement, which is a cost worth incurring deliberately rather than for appearance alone.

Next step

Talk to the team

A short call, a look at how the business currently shows up, and a straight answer on what we would do first.