SEO Web Design: How to Build a UK Site That Ranks and Converts

SEO web design is the practice of building a website so search engines can crawl, understand and rank it while visitors can use it and buy from it. It combines technical foundations, content architecture and conversion design from the first sitemap onwards, rather than adding search considerations after the site is built.

Your website collects thousands of search impressions every month and almost nobody clicks. You can see it in Search Console: the pages are being shown, the position sits somewhere in the thirties or forties, and the clicks trickle in single figures. Meanwhile the paid budget climbs to cover the gap.

That gap is expensive and it compounds. Every month a page sits on the fourth page of results is a month of demand handed to a competitor whose site was built to be found. The traffic you are buying through ads is traffic you could be earning, and the moment you pause the spend it disappears.

This guide covers how SEO web design closes that gap: the architecture decisions that determine whether a site can rank at all, the technical standards Google actually measures, the content structure that wins competitive commercial terms, and a ninety-day programme for proving it works on your own site before committing to a full rebuild.

What SEO Web Design Actually Means

SEO web design is the practice of building a website so that search engines can crawl, understand and rank it, while real people can still use it and buy from it. The two goals are built at the same time, by the same team, from the same plan.

That sounds obvious. In practice it almost never happens. Most websites are built in one of two ways. A design studio produces something beautiful that no one can find, because the structure, the page speed and the content were never part of the brief. Or an SEO specialist arrives eighteen months later and tries to retrofit rankings onto a site whose foundations will not take them.

SEO web design removes the handover. The sitemap is a keyword plan. The page templates are built around the content that has to rank. The performance budget is agreed before a single component is designed.

The Three Layers That Have to Agree

A site that ranks and converts is working on three layers at once, and a weakness in any one of them caps the other two. The technical layer covers crawlability, indexation, page speed, structured data and mobile rendering. This is what decides whether Google can read the site at all.

The content layer covers what each page is about, how pages relate to one another, and whether the writing answers the question someone actually typed. This is what decides whether Google thinks the page deserves to rank. The experience layer covers whether a visitor can find what they need, trust what they read, and complete the action you want. This is what decides whether the ranking is worth anything.

You can hold a first-page position with a poor experience layer for a while. You will not hold it for long, and you will not make money from it in the meantime.

What It Is Not

SEO web design is not a plugin. Installing an SEO plugin on a badly structured site gives you a tidy interface for filling in metadata on pages that were never going to rank.

It is also not a one-off project with a launch date and an invoice. Search results move, competitors publish, and Google changes what it rewards. A site built for search in 2023 and left alone is, by 2026, a site built for a search engine that no longer exists in that form.

Our SEO web design work treats the build as the start of the programme rather than the end of it.

Why Most UK Business Websites Fail at Both Jobs

The failure is rarely dramatic. Sites do not usually collapse. They just quietly underperform, for years, while everyone assumes the problem is the ad budget.

Here is what that looks like in the data. We recently reviewed a service page that had accumulated 624,515 search impressions over sixteen months and earned 23 clicks from them. An average position of 43. Google was showing the page to hundreds of thousands of people and almost none of them clicked, because the page sat on the fourth page of results where clicks effectively do not exist.

Impressions without clicks is the single most common pattern we see on UK business sites. It means the site is relevant enough to be considered and not good enough to be chosen.

The Five Failures Behind That Pattern

The first is a sitemap that mirrors the org chart instead of the customer’s search behaviour. Businesses structure sites around how they are internally organised, which is almost never how anyone searches. The second is thin service pages. A page of 300 words about a competitive commercial term will be shown, considered and skipped over. Google has better options.

The third is speed. A site that takes four seconds to render its main content on a mid-range Android phone on 4G is losing both rankings and revenue, and most businesses test on a fast laptop on office broadband and conclude the site is fine.

The fourth is the absence of any supporting content. A service page with no cluster of related articles beneath it has nothing to establish that the business knows the subject. The fifth is a conversion path that assumes the visitor is ready to buy. Most are not, and a site offering only “contact us” captures only the small fraction who are.

Why the Problem Persists

Nobody owns it. The web developer’s job finished at launch. The marketing team inherited a site they did not specify. The agency running ads is judged on cost per lead and has no mandate to rebuild page templates. So the site stays as it is, and the paid budget goes up to compensate. That is an expensive way to solve a structural problem, and it stops working the moment you pause the spend.

A Worked Diagnostic

Take a page you suspect is underperforming and run this sequence. It takes about twenty minutes and it identifies which of the five failures you are dealing with. Open Search Console, filter to that page, and set the range to the last sixteen months. Note impressions, clicks and average position.

If impressions are high and average position is beyond 20, the page is relevant but not competitive. That is a content depth and authority problem. Check the word count against the three pages currently ranking in the top five and check how many internal links point at your page.

If impressions are high and average position is inside 10 but clicks are low, the page ranks and nobody chooses it. That is a title and meta description problem, and it is the cheapest of all these to fix.

If impressions are low despite the page targeting a term with real volume, Google may not understand what the page is about, or may not have indexed it properly. Check the URL Inspection tool for indexation status and check whether the primary keyword appears in the title, the H1 and the first paragraph.

If impressions are healthy, position is good and clicks arrive but nothing converts, the ranking is fine and the page is failing at the experience layer. That is a conversion design problem and no amount of SEO will fix it.

The Honest Version of the Diagnosis

Most sites have more than one of these at once, and the temptation is to fix everything simultaneously. Resist it. Fix indexation first because nothing else matters without it, then titles because they are quick, then content depth because it takes longest, then conversion because it needs traffic to test against.

Working in that order means you see movement early, which is what keeps the programme funded long enough to reach the parts that take two quarters.

The Business Case: What a Rankable Site Is Worth

Before the technical detail, it helps to put a number on this, because SEO web design is usually sold on vague promises and bought on hope. The arithmetic is simple enough to do on the back of an envelope, and doing it changes how seriously the work gets taken.

Working the Numbers

Take a real example. A commercial term generates 50,000 impressions a year for your site. At position 43 you earn a click-through rate of roughly 0.04%, which is 20 clicks a year. Move that page to position 5 and the click-through rate for that position sits somewhere around 5% to 6% depending on the search feature layout. Call it 5%. That is 2,500 clicks a year from the same impressions.

Now apply a realistic conversion rate. A well-built service page with a proper enquiry path converts somewhere between 2% and 5% of visitors into enquiries. At 3%, those 2,500 clicks produce 75 enquiries. If you close one in four, that is roughly 19 new customers. At an average first-year value of £6,000, the same page has gone from producing almost nothing to producing around £112,000.

Why the Investment Case Usually Fails Anyway

The numbers above are not controversial, and businesses still do not act on them. The reason is that the cost is immediate and certain while the return is delayed and probabilistic. A rebuild costs real money this quarter. The rankings arrive over two to three quarters. Whoever signs the invoice carries the risk and may not be there for the payoff.

The honest answer to that is to stage the work. You do not need a full rebuild to test the thesis. Take the two or three pages with the highest impressions and worst positions, fix those properly, and measure what happens over ninety days. If the model holds on three pages, it will hold on thirty.

A Caution on the Arithmetic

Treat any click-through rate by position table you find online with suspicion, including the figures above. They vary enormously by query type, by whether an AI Overview or a local pack sits above the results, and by industry.

We could not trace the commonly quoted position-one CTR figures to a single original study that holds up across sectors. Most cite each other. Use the model to size the opportunity and to decide whether the work is worth doing, then measure your own results rather than trusting the benchmark.

Where This Advice Does Not Apply

Most guides of this length imply their advice suits everyone. It does not, and knowing the exceptions saves money.

When Organic Search Is the Wrong Channel

If your product is genuinely new and nobody is searching for it, SEO cannot help until the category exists. You are creating demand rather than capturing it, and paid social, outbound and PR do that better. If your sales cycle runs through procurement frameworks, tenders or a small number of named accounts, your buyers are not finding you through search. Account-based marketing will outperform any amount of content.

If you need pipeline within eight weeks, SEO web design is the wrong instrument. It is a two to three quarter discipline and telling you otherwise would be dishonest. Run paid to buy time, and start the organic work in parallel so it is producing by the time the paid budget is under review.

When the Site Is Not the Problem

Sometimes traffic and rankings are fine and revenue is not, and the cause sits downstream. If enquiries arrive and nobody follows them up within a day, fixing the site adds volume to a leaking process. Lead response time is usually worth more than another twenty rankings.

If the enquiries are the wrong sort, the problem is targeting and qualification rather than volume. More traffic of the same kind produces more wasted time. Check the pipeline before commissioning the rebuild. It is a cheaper conversation and occasionally it makes the rebuild unnecessary.

When a Smaller Intervention Is Enough

A site with sound architecture, decent speed and thin content does not need a rebuild. It needs writing, and hiring a rebuild to solve a content problem is an expensive category error. A site with good content and a technical fault needs a developer for a fortnight, not a programme.

The full discipline earns its cost when the structure itself is the constraint: wrong URL architecture, templates that cannot support the content depth required, a platform capping performance, or no content model at all. Those are the conditions where nothing smaller will work.

Information Architecture: Structure, URLs and Internal Links

Architecture is where the whole discipline either works or does not, and it is decided in the first week of the project when the sitemap is drawn. Get it right and everything afterwards is easier. Get it wrong and you spend three years patching.

Build the Sitemap From Search Demand

Start with keyword research, not with a list of what the business does. Those two things overlap but they are not the same, and the gaps between them are where the opportunities sit. Pull the terms people actually search, group them by intent, and let the groups become your page structure. If there are 20,000 monthly searches split across four distinct variations of a service, that is four pages, not one page with four paragraphs.

One page per search intent is the rule. Two pages competing for the same intent creates keyword cannibalisation, where your own pages split the signal and neither ranks.

URL Structure

Keep URLs short, readable and stable. A URL should describe the page to a human who sees it in a search result with no other context. Use hyphens between words, lower case throughout, and no dates, IDs or session parameters in the path. Nest URLs to reflect genuine hierarchy, so a service that sits beneath a category lives at /category/service/ rather than floating at the root.

The most important rule is that URLs do not change. Every change costs you link equity through the redirect and risks losing the page entirely if the redirect is missed. Decide the structure once, properly, at the start.

Internal Linking Is a Ranking System

Internal links tell Google which of your pages matter and what they are about. Most sites treat them as navigation furniture and waste the signal entirely.

Three rules cover most of the value. Link from high-authority pages to the pages you want to rank, because authority flows along links. Use descriptive anchor text that tells Google what the target page is about, never “click here” or “read more”. Make sure every important page is reachable within three clicks of the homepage.

A useful diagnostic is to list your twenty most important commercial pages and count the internal links pointing at each. If your highest-value service page has four internal links and your privacy policy has sixty, your architecture is telling Google the wrong story. Our wider SEO work almost always starts with this audit, because it is the cheapest meaningful improvement available on most sites.

The Most Common Architecture Mistake

The mistake we see most often is one page trying to serve several distinct intents because someone wanted to avoid duplication. A business offering web design, SEO and hosting builds a single “services” page covering all three. It ranks for none of them, because it is not specifically about any of them, and the three terms each had enough demand to justify their own page.

The opposite mistake is almost as common. A business builds fourteen near-identical pages targeting minor keyword variations of the same intent, which splits the signal and triggers cannibalisation. The test is whether two search terms would be satisfied by the same page. If someone searching each term would be happy landing on the same content, it is one page. If they would want different things, it is two.

Core Web Vitals in 2026: What to Measure and How to Fix It

Core Web Vitals are Google’s attempt to put numbers on whether a page feels good to use. They are a confirmed ranking signal, and they matter more as a tie-breaker between comparable pages than as a way to beat better content. There are three, and the thresholds have been stable for longer than most commentary suggests.

The Three Metrics and Their Real Thresholds

Largest Contentful Paint measures how long it takes for the biggest visible element to render. Google’s threshold for good is 2.5 seconds or less. Interaction to Next Paint measures how quickly the page responds when someone taps or clicks. Good is 200 milliseconds or less. It replaced First Input Delay in March 2024 and carries the same weight as the other two.

Cumulative Layout Shift measures how much the page jumps around while loading. Good is 0.1 or less. All three are assessed at the 75th percentile of real Chrome user data over a rolling 28-day window, split between mobile and desktop, as Google’s Search Central documentation sets out. That last detail matters: you are judged on your slowest quarter of real visitors, not on your average, and not on the lab score your developer sees.

One correction worth making, because it is circulating widely. Several SEO blogs claim Google tightened the LCP threshold to 2.0 seconds in March 2026. That is not accurate. The official threshold published by Google remains 2.5 seconds. Build to 2.5, and treat anything faster as margin rather than a requirement you have failed to meet.

Why Lab Scores Mislead

Open any page in Lighthouse and you get a score out of 100. Businesses fixate on that number and it is the wrong one to chase. Lighthouse runs a simulated load on a throttled connection from wherever you happen to be. It is a diagnostic tool, useful for finding what is slow and why. It is not what Google uses for ranking.

Google uses field data from the Chrome User Experience Report, gathered from real people on real devices and real networks. A site can score 95 in Lighthouse and fail Core Web Vitals in the field, usually because real users are on older Android phones with slower processors than the simulation assumes.

Check the field data first, then use the lab tools to work out why it looks the way it does. Search Console’s Core Web Vitals report and PageSpeed Insights both surface field data when there is enough traffic to populate it.

Fixing Largest Contentful Paint

LCP is usually one of four things, and they are worth checking in order. Slow server response is the first. If the server takes 800 milliseconds to return the HTML, you have burned nearly a third of your budget before the browser has started. Better hosting, caching and a content delivery network usually fix this.

Render-blocking resources are the second. CSS and JavaScript loaded in the head stop the page from painting. Inline the critical CSS needed for the visible area, defer the rest, and load JavaScript asynchronously wherever it is not needed immediately.

The LCP element itself is the third, and on most marketing sites it is the hero image. Serve it in a modern format, size it correctly for the viewport, and never lazy-load it. Lazy-loading the hero image is one of the most common self-inflicted LCP failures we find.

Client-side rendering is the fourth. If the main content is drawn by JavaScript after the page loads, LCP includes all of that work. Server-side rendering or static generation moves it back into the initial response.

Fixing Interaction to Next Paint

INP problems are almost always JavaScript blocking the main thread. The browser cannot respond to a tap because it is busy executing something else. Audit what is actually running. On a typical WordPress marketing site the culprits are a tag manager firing a dozen scripts, a chat widget, a cookie consent tool and three tracking pixels, all competing for the same thread during the first few seconds.

Break long tasks into smaller ones so the browser can respond between them. Defer anything not needed for the initial interaction. Load chat widgets and marketing tags after the page is interactive rather than during load. Be willing to remove things. Every tag on the site was added by someone who wanted data. Few are ever removed when the need passes, and the accumulated cost lands on your slowest users.

Fixing Cumulative Layout Shift

CLS is the easiest of the three to fix and the most annoying to experience. Set explicit width and height attributes on every image and video so the browser reserves the space before the file arrives. Reserve space for ad slots and embeds. Avoid injecting banners or cookie notices above existing content after load.

Web fonts cause a large share of remaining CLS. Use font-display: swap with a fallback font metrically matched to the web font, so the swap does not reflow the text. Google’s own Web Vitals documentation is the authoritative reference for all three metrics and worth reading directly rather than through secondary summaries.

A Troubleshooting Sequence for a Failing Page

When a page fails Core Web Vitals and nobody knows why, work through this in order rather than guessing. Start with the field data in Search Console and identify which of the three metrics is failing and on which device type. Mobile and desktop frequently differ and fixing the wrong one wastes a sprint.

Run PageSpeed Insights on the specific URL and read the diagnostics section rather than the score. It names the largest contentful paint element explicitly, which tells you what to attack. If LCP is the failure, check the time to first byte. Above 800 milliseconds and your problem is server-side, so caching and hosting come before anything in the browser. Below that, the problem is in the page.

Check whether the LCP element is lazy-loaded. This single mistake accounts for a large share of the LCP failures we see, and it is a one-line fix. If INP is the failure, open the Performance panel in Chrome DevTools, record an interaction, and look for long tasks over 50 milliseconds. Whatever is at the top of that list is your answer, and it is usually a third-party script.

If CLS is the failure, load the page with a slow connection simulated and watch what moves. Images without dimensions, late-loading fonts and injected banners account for nearly all of it.

A Worked Example

A client site was failing LCP on mobile at 4.1 seconds against the 2.5 second threshold. The Lighthouse score was 78, which had convinced everyone the site was broadly fine. Time to first byte was 210 milliseconds, so the server was not the issue. The LCP element was the hero image, served as a 2.4MB JPEG at 3,000 pixels wide, scaled down by CSS to display at 800 pixels.

Converting it to WebP, serving it at 1,600 pixels for retina displays, and removing the lazy-load attribute took the file from 2.4MB to 180KB. Field data moved to 2.2 seconds over the following four weeks. One image, one afternoon, and the page moved from failing to passing. The rankings did not transform overnight, and the page stopped being held back by something that had nothing to do with its content.

Mobile-First Design and Indexing

Google indexes the mobile version of your site. Not the desktop version, not a blend. What is not on the mobile version effectively does not exist for ranking purposes. This has been the case for years and a surprising number of UK business sites still fail it, usually by hiding content on mobile to keep the design tidy. Google’s mobile-first indexing documentation is explicit about the requirement.

What Mobile-First Indexing Actually Requires

The mobile page must contain the same primary content as the desktop page. The same headings, the same body copy, the same structured data, the same meta tags. Collapsing content into accordions is fine. Google indexes content inside accordions. Removing it entirely from the mobile DOM is not fine, and it is what many responsive themes do by default with sidebars, comparison tables and secondary sections.

Images and video need to be present on mobile too, with the same alt text. If your desktop page has six images explaining the service and your mobile page has one, you have thinned your own content for the version that counts.

Designing for the Thumb

Mobile design for SEO is not only about what Google reads. Conversion on mobile is where most UK business sites lose the visitors they worked hardest to attract. Tap targets need to be at least 48 by 48 pixels with adequate spacing. Forms need appropriate input types so the right keyboard appears. Phone numbers need to be tappable links.

The single most valuable mobile change we make on service sites is putting a sticky call or enquiry bar at the bottom of the viewport. On a phone, a visitor who has scrolled 2,000 pixels down a service page should never have to scroll back to the top to make contact.

Test on Real Devices

Chrome DevTools device emulation is useful and it lies about performance. It renders on your laptop’s processor. Buy a mid-range Android handset, put it on a throttled connection, and use the site. The experience is usually sobering and it is what a meaningful share of your audience actually gets. Our design and development team tests on real hardware for exactly this reason.

On-Page Foundations: Titles, Headings and Metadata

This is the part everyone knows and most sites still get wrong, usually through inconsistency rather than ignorance.

Title Tags

The title tag remains one of the strongest on-page signals and the main thing that decides whether someone clicks. Put the primary keyword near the start. Keep it between 50 and 60 characters so it does not truncate in results. Make every title on the site unique. Write it for a person deciding between ten options, not for a crawler counting keywords.

Google rewrites titles it considers unhelpful, and a rewritten title is a signal that yours did not describe the page well enough.

Meta Descriptions

Meta descriptions are not a ranking factor and they heavily influence click-through rate, which is worth just as much. Write 150 to 160 characters, include the primary keyword naturally, and describe the specific benefit of clicking. A description that could sit on any page of the site is a wasted opportunity on all of them.

Heading Hierarchy

One H1 per page, describing what the page is about. H2s for major sections, H3s beneath them, and never a skipped level. Headings are how both Google and a scanning reader build a mental model of the page. On a long guide, the heading structure is the table of contents, and it should read as a coherent outline on its own.

Image Optimisation

Images are usually the largest files on a page and the biggest single lever on load time. Serve modern formats such as WebP or AVIF. Size images for the largest viewport that will display them rather than uploading a 4,000 pixel original and letting CSS shrink it. Lazy-load everything below the fold and nothing above it.

Write alt text that describes the image for someone who cannot see it. This is an accessibility requirement first and a ranking benefit second, and our guide to ranking images in Google covers the detail.

Writing Titles That Earn the Click

Titles are worth spending real time on because they are the cheapest lever you have. Here are four written for the same page, in ascending order of usefulness. “SEO Web Design | Agency Name” is the default most content management systems produce. It contains the keyword and gives a searcher no reason to choose it.

“SEO Web Design Services for UK Businesses” is better. It adds qualification and audience, and it still describes a category rather than a benefit. “SEO Web Design: Sites Built to Rank and Convert” adds the outcome. A searcher comparing ten results now has something to prefer.

“SEO Web Design: The 2026 UK Guide to Sites That Rank” adds recency and scope, signals depth, and fits inside the character limit. The pattern across all four is that specificity beats completeness. A title that tries to include every keyword ends up describing nothing.

A Practical Test

Write your title, then put it in a list with the titles currently ranking in the top five for your target term. Read the list as a searcher would, in about two seconds each. If yours does not stand out for a reason you can name, rewrite it.

This takes five minutes per page and improves click-through rate more reliably than most technical work.

Content Architecture: Service Pages, Location Pages and Blog Clusters

A site that ranks for competitive commercial terms does it with a structure, not with a page. This is the piece most often missing.

The Pillar and Cluster Model

The model is straightforward. A pillar page targets the main commercial term and covers the subject broadly. Cluster pages target the specific long-tail questions around it in depth. Every cluster page links up to the pillar, and the pillar links down to each cluster.

This does two things. It gives you pages that can rank for hundreds of specific queries you would never fit onto one page. And it concentrates topical authority on the pillar, because Google can see a coherent body of related work rather than an isolated page making claims.

A service page with no cluster beneath it is competing on assertion alone. A service page supported by fifteen genuinely useful articles is competing on evidence.

How Deep a Service Page Should Be

Long enough to answer everything a buyer needs and no longer. In practice, for a competitive UK commercial term, that is usually 1,500 to 2,500 words. Cover what the service is, who it is for, how it works step by step, what it costs or how pricing is structured, what results look like with real examples, who delivers it, and what happens when someone gets in touch.

Thin service pages are the most common structural failure we see, and they are the reason a page can collect 600,000 impressions and 23 clicks. This is the single most common finding in our SEO audits.

Location Pages Done Properly

Location pages have a bad reputation because most are spam. A template with the town name swapped in, published 200 times, is a scaled content problem and Google treats it as one. Done properly they work well. A location page needs genuinely local content: real projects in that area, named local clients where you have permission, specifics about the local market, and a real address or service area.

If you cannot write 600 unique words about serving that town, do not publish a page for it.

Blog Content That Supports Commerce

Blog posts should be chosen from keyword data and linked deliberately to the service pages they support. A post that ranks well and links nowhere useful is a vanity metric. Every post should carry at least one contextual link to the relevant service page, using anchor text that describes what that page offers. Our SEO copywriting team plans this at the outline stage rather than adding links afterwards.

A Worked Cluster

Here is how a cluster looks for a single competitive service term, using this discipline itself as the example. The pillar page is the service page, targeting the primary commercial term and covering the offer, the process, the evidence and the enquiry path.

Every cluster page links up to the service page with descriptive anchor text. The service page links down to the three or four most useful clusters. That structure lets you rank for the commercial term through the pillar and for several hundred long-tail queries through the clusters, while every one of those clusters passes authority and visitors towards the page that generates enquiries.

How Long a Cluster Takes to Work

Longer than most people are told. A cluster of six pages published over a quarter typically starts producing measurable movement on the pillar in the following quarter. The clusters themselves often rank faster than the pillar, because long-tail queries are less competitive. That is useful, because it gives you early evidence the approach is working while the commercial term is still climbing.

Publish the clusters first if you need to prove the model quickly. They rank sooner, they are cheaper to produce, and they build the internal links the pillar needs anyway.

Structured Data and Schema for UK Businesses

Structured data is code that tells search engines what a page means rather than leaving them to infer it from the words. It has become considerably more valuable as search has shifted towards AI-generated answers.

What to Implement First

Most UK business sites need four types and can ignore the rest until those are solid. Organization schema describes the business itself: legal name, logo, address, contact details and social profiles. It goes on every page, usually in the footer or head, and it is what feeds knowledge panels.

LocalBusiness schema extends Organization with opening hours, geographic coordinates and service areas. Any business with a physical location or a defined service area should have it. Service schema describes what each service page offers, who provides it and where. It is underused and it is the one that most directly supports commercial pages.

FAQPage schema marks up genuine question and answer content. It helps search engines extract your answers directly, which matters a great deal for AI Overviews.

Getting It Right

Use JSON-LD. Google prefers it, it sits in a single script tag, and it does not tangle with your HTML. Only mark up content that is actually visible on the page. Marking up content the user cannot see is a structured data violation and can earn a manual action.

Validate everything with the Rich Results Test before publishing and check the Search Console enhancement reports afterwards. Schema that fails validation does nothing at all, and it fails silently. We build this into every SEO web design project as a launch requirement. Structured data does not improve rankings directly. It improves how your result is displayed and how easily machines can extract your content, and both of those move click-through rate.

Why It Matters More in 2026

When a large language model assembles an answer, it needs to identify entities, relationships and facts quickly and reliably. Structured data hands it that on a plate. Sites with clean, comprehensive schema are being cited in generative answers more often than comparable sites without it. The relationship is not fully documented by Google and the pattern is consistent enough across our client base to act on.

What Good Schema Actually Contains

A common mistake is implementing the bare minimum required for validation and assuming the job is done. Validation only proves the syntax is correct. Useful Organization schema names the business, its logo, its full postal address, its telephone number, its founding date and every social profile it controls. That set of properties is what allows Google to connect your site to an entity it already knows about.

Useful Service schema names the service, describes it in a sentence that matches how buyers describe it, identifies the provider, and states the geographic area served. Where you have pricing structure rather than fixed prices, an offer range is more useful than omitting price entirely.

Useful FAQPage schema marks up questions people genuinely ask, phrased the way they ask them, with answers that stand alone without the surrounding page. The test is whether someone reading only your structured data would understand what the business does, where it operates and what it offers. If they would not, the schema is decorative.

Designing for Conversion, Not Just Ranking

A page at position three that converts nobody is a more expensive failure than a page at position thirty, because you paid to get it there. Conversion design and SEO pull in the same direction more often than people expect. Both reward clarity, speed and giving the visitor what they came for.

The Five-Second Test

Someone lands on your page from a search result. In five seconds they decide whether to stay. They need three questions answered in that time: what is this, is it for me, and what do I do next. If the top of the page is a stock photograph and a slogan, none of those are answered and the visitor returns to the results.

Put a clear headline that matches the search intent, a single sentence of supporting detail, and a visible call to action above the fold. The clever creative can follow underneath.

Matching the Page to the Intent

A visitor searching “what is SEO web design” wants an explanation. A visitor searching “SEO web design agency” wants to evaluate suppliers. Serving the same page to both loses one of them. Map intent to page type at the architecture stage. Informational queries get guides. Commercial investigation queries get comparison and evidence. Transactional queries get service pages with a direct path to enquiry.

Offering More Than One Way to Convert

Most business sites offer exactly one conversion: a contact form for people ready to buy now. That captures a small minority of visitors and discards the rest. Add a lower-commitment option for people who are researching. A downloadable guide, a checklist, a benchmark report or a diagnostic tool gives you a contact you can nurture rather than a bounce.

For a subset of visitors, a booked call is easier than a form. Our own schedule a meeting route exists because a meaningful share of people would rather pick a slot than write an enquiry.

Forms

Every field you add reduces completion. Ask for what you genuinely need to respond usefully and nothing else. Name, email and a message will do for most businesses. Phone number if you intend to call. Everything else can be gathered in the conversation that follows.

Multi-step forms often outperform single long ones, because the first step feels trivial and commitment builds. Our conversion rate optimisation work usually starts with form simplification because the gains are quick and measurable.

What Does Not Work

A few conversion tactics are widely recommended and reliably disappointing, and it saves money to know which. Exit-intent popups offering a discount on a considered B2B service rarely convert, because the objection was never price and the interruption reads as desperation. Live chat staffed by nobody is worse than no live chat. A widget that promises a reply and delivers one in four hours teaches visitors that the business does not respond.

Long landing pages built entirely from testimonials and logos, with no substantive detail about how the service works, perform badly with buyers who are actually evaluating. Social proof supports a decision and does not make one. Auto-playing video with sound loses more visitors than it persuades, and it damages Core Web Vitals at the same time.

Testing Without Enough Traffic

Most UK business sites do not have the traffic for statistically valid A/B testing on a service page, and running tests anyway produces confident conclusions from noise. Below roughly a thousand conversions a month, use sequential testing instead. Make one substantial change, run it for a full business cycle, and compare against the equivalent previous period while accounting for seasonality.

This is less rigorous and it is honest about its own limits, which is better than a test declared significant at 40 conversions. Where traffic is genuinely too low, move up the funnel. Qualitative research, session recordings and five user interviews will tell you more than any underpowered experiment.

Accessibility, and Why It Overlaps With SEO

Accessibility is a legal obligation, an ethical one, and an unusually efficient way to improve search performance at the same time. The overlap is not a coincidence. Screen readers and search crawlers both consume the page without seeing it, so what helps one generally helps the other.

Where the Two Overlap Directly

Alt text describes images for screen reader users and tells search engines what the image shows. Heading hierarchy lets screen reader users navigate by heading and lets crawlers understand document structure. A page with six H1s and no H2s is hard for both. Descriptive link text tells a screen reader user where a link goes when read out of context, and tells Google what the target page is about. “Click here” fails both.

Semantic HTML gives assistive technology and crawlers the same structural map. A div styled to look like a button is a button to neither. Captions and transcripts make video accessible and give search engines text to index from content that is otherwise opaque.

The UK Legal Position

The Equality Act 2010 requires service providers to make reasonable adjustments, and that has been understood to include websites. Public sector bodies have stricter obligations under separate regulations. The practical standard to build to is WCAG 2.2 at AA level. Most of it is straightforward: sufficient colour contrast, keyboard operability, visible focus states, labelled form fields and text that resizes without breaking the layout.

Building to AA from the start costs very little. Retrofitting it later costs a great deal. That asymmetry is the argument for putting it in the original brief, alongside the SEO and performance requirements.

A Practical AA Checklist

Most of WCAG 2.2 at AA level comes down to a manageable set of checks that a designer and developer can work through together before launch. Colour contrast needs to reach 4.5 to 1 for normal text and 3 to 1 for large text. This is the single most commonly failed criterion, and it is usually caused by light grey body copy chosen for aesthetic reasons.

Every interactive element must be reachable and operable by keyboard alone. Tab through the whole page without touching the mouse. If you get trapped in a modal or cannot reach the navigation, you have found a failure. Focus states must be visible. Many themes remove the default browser outline for tidiness and replace it with nothing, which makes the site unusable for keyboard navigation.

Form fields need programmatically associated labels, not placeholder text standing in for them. Placeholders disappear on typing and are not read reliably by assistive technology. Text must resize to 200% without loss of content or function, and the page must reflow at 320 pixels wide without horizontal scrolling.

Any video needs captions, and any audio needs a transcript. Run an automated checker first to catch the mechanical failures, then test manually, because automated tools catch roughly a third of real accessibility problems and confidently report the rest as passing.

Choosing a Platform: WordPress, Webflow, Shopify and Headless

The platform decision constrains everything afterwards, and it is usually made on familiarity rather than fit.

WordPress

WordPress runs a large share of the web and remains a reasonable default for content-led marketing sites. The SEO tooling is mature, the content editing experience is familiar, and developers are easy to find. The weakness is performance. A WordPress site with a page builder, twenty plugins and an off-the-shelf theme will struggle with Core Web Vitals, and fixing it usually means removing things people like.

Built carefully, with a lean custom theme and disciplined plugin use, WordPress performs perfectly well. Built the way most WordPress sites are built, it does not.

Webflow

Webflow produces clean markup and fast sites with good Core Web Vitals out of the box, and gives designers direct control without a developer for every change. The constraints are the CMS structure, which is less flexible for complex content models, and the cost at scale. For a marketing site of moderate size with a strong design requirement, it is often the better choice.

Shopify

For ecommerce, Shopify handles the hard parts well. The SEO limitations are real and mostly structural: a rigid URL structure that forces /collections/ and /products/ into paths, and duplicate content risk from products appearing in several collections. These are manageable with canonical tags and careful internal linking. They are not reasons to avoid the platform if you are selling online.

Headless

A headless build separates the content management from the front end, which gives you maximum performance and maximum flexibility. It also gives you maximum cost and a dependency on developers for changes a marketer could otherwise make themselves. It is the right answer for large, complex or highly bespoke sites, and overkill for a twenty-page service business.

How to Actually Decide

Write down who will edit the site day to day, how often, and what they need to change without help. Write down your performance requirement. Write down your content model. Then pick the platform that fits those three things. The best platform is the one your team will keep current, because an out-of-date site on a perfect stack still loses. We work across all four on search-led website builds.

Comparing the Options Honestly

The table below is a starting point rather than a verdict, because the right answer depends on your team more than on the technology.

The question that settles most of these is who makes a change on a Tuesday afternoon when someone spots a typo on a service page. If the answer is a marketer, choose a platform a marketer can operate. If the answer is a developer with a two-week sprint cycle, your content will go stale and no amount of technical performance will compensate.

The Migration Question

Businesses often ask whether they should move platforms to fix performance. Usually the answer is no. A badly built site on a fast platform is still a badly built site, and migration carries the ranking risks covered later in this guide. Most performance problems are caused by how the site was built rather than by what it was built on.

Move platforms when the platform genuinely caps what you need, when the current build is beyond economic repair, or when the editing experience is stopping your team keeping the site current. Those are good reasons. “Our developer prefers something else” is not.

The Redesign Trap: Rebuilding Without Losing Rankings

More UK businesses lose organic traffic to their own redesign than to any algorithm update. The pattern is predictable and entirely avoidable.

How the Damage Happens

The new site launches with different URLs and no redirect map, so every ranking page returns a 404 and the accumulated authority evaporates. Or the staging site was blocked from indexing and someone forgot to remove the noindex tag at launch, so the entire site disappears from search for as long as it takes anyone to notice.

Or the new design is beautiful and thin, having cut the 1,800-word service pages down to 400 words of elegant copy, and the pages no longer contain enough to rank. Or the tracking was rebuilt and the historical comparison is gone, so nobody can prove what was lost.

The Protective Sequence

Before anything is built, crawl the existing site and export every URL with its traffic, rankings and backlinks. This is your baseline and your redirect source list. Identify the pages that earn organic traffic and treat their content as a requirement rather than a starting point for the designer. Content can be improved. It should not be quietly deleted.

Build a one-to-one redirect map from every old URL to its closest new equivalent. Avoid redirect chains and never point everything at the homepage, which Google treats as a soft 404. Launch, then check indexation, redirects, Core Web Vitals and Search Console coverage on day one, day seven and day thirty.

Our website redesign SEO checklist sets this out step by step and is worth working through before any rebuild.

Expect a Dip

Even a well-executed migration usually produces a short-term fluctuation while Google recrawls and reassesses. Two to four weeks is normal. Knowing this in advance stops panic decisions in week two. What is not normal is a sustained decline past six weeks, and that is when you go back to the redirect map and the indexation report. Our ongoing SEO checklist covers the monthly maintenance that follows.

Designing for AI Overviews and Generative Search

A growing share of searches now return an AI-generated summary above the traditional results. That changes what a ranking is worth and what a site needs to do to earn attention. Google published a guide to optimising for generative AI features in May 2026, and the guidance in it is notably close to ordinary good practice rather than a new discipline.

What Actually Changes

The click-through rate for informational queries falls when an AI Overview answers the question directly. That is the uncomfortable part and pretending otherwise helps nobody. What rises in value is being the cited source inside that answer, and ranking for queries where the searcher still needs to reach a supplier. Commercial and transactional intent is far less affected than pure information seeking.

The practical response is to shift the content mix towards queries where a click is still required, while making the informational content extractable enough to earn citations.

How to Be Extractable

Answer the question directly in the first fifty words under the heading, before the context and the caveats. Systems assembling an answer take the clearest available statement, and burying it after three paragraphs of preamble means they take someone else’s. Use clear headings phrased as the questions people ask. Keep paragraphs short and self-contained so a passage can be lifted without losing its meaning.

Include specifics. Numbers, dates, named thresholds and concrete steps are more citable than general advice, because they can be verified and attributed. Implement the structured data covered earlier, which gives the machine an unambiguous account of what your page contains.

Experience Is the Differentiator

Google’s guidance on creating helpful, reliable, people-first content keeps returning to first-hand experience, and it is the hardest thing for a generated summary to replicate. Original data from your own client work, photographs of real projects, named case studies with real numbers, and opinions that could only come from doing the work are what distinguish a source worth citing from one worth summarising.

Our guides to Google AI Overviews and appearing in ChatGPT go further into the mechanics.

Measuring SEO Web Design: The Metrics That Matter

Most reporting on this work measures the wrong things, which is why it is so often cancelled before it pays off.

The Metrics Worth Watching

Impressions by page tell you whether Google considers the page relevant at all. Rising impressions with flat clicks means the page is being shown and rejected, which is a content and title problem rather than a technical one.

Average position by page tells you how far you have to travel. Movement from 43 to 25 produces almost no extra clicks and is real progress, and reporting it alongside clicks stops the work being judged as a failure at the halfway point. Click-through rate by position tells you whether your titles and descriptions are earning the click your ranking has bought.

Core Web Vitals field data tells you whether the technical work has reached real users, which lab scores do not. Organic sessions to key landing pages, and conversions from those sessions, are what the business actually cares about.

The Metrics Worth Ignoring

Total keyword count is noise. Ranking for 4,000 terms means nothing if none of them are commercial. Domain authority scores from third-party tools are estimates invented by those tools. They correlate loosely with performance and Google does not use them. Lighthouse scores, as covered, are diagnostics rather than outcomes.

Bounce rate on informational pages is frequently misread. Someone who reads a guide, gets their answer and leaves has been served well.

A Reporting Cadence That Survives Contact With a Board

Monthly reporting on a quarterly-to-annual discipline invites cancellation, because month one shows cost and no return. Report monthly on leading indicators: pages published, technical issues resolved, Core Web Vitals field data, impressions and average position. Report quarterly on lagging indicators: organic sessions, enquiries and pipeline.

Set expectations in writing at the start. Meaningful ranking movement on competitive commercial terms takes two to three quarters. Anyone promising otherwise is either lucky or selling something.

A Worked Reporting Example

Here is a ninety-day report for a priority page set, written the way it should be rather than the way it usually is. The baseline in week one: five priority pages, combined 84,000 impressions over the previous quarter, average position 38, 61 clicks, 2 enquiries.

At week twelve: combined 91,000 impressions, average position 19, 340 clicks, 11 enquiries. The impressions barely moved, which is expected and worth explaining in the report rather than hiding. Google was already showing these pages. What changed is that they moved from the fourth page of results to the second, and the click-through rate rose from 0.07% to 0.37%.

Average position was the leading indicator and it moved first. Clicks followed. Enquiries followed clicks. Anyone reading only the enquiry number at week six would have concluded the programme had failed.

Presenting This to People Who Did Not Commission It

Finance directors and boards see cost immediately and results late, so the report has to bridge that gap explicitly. Lead with the modelled value of the position change rather than the position change itself. “These five pages moved from page four to page two, which on current click-through rates is worth an additional 280 visits a month and roughly nine enquiries” lands where “average position improved 19 places” does not.

Show the baseline every time. A number without a comparison is not a result. Be explicit about what has not yet moved and when you expect it to. Reports that only contain good news stop being believed, usually at exactly the moment you need to be believed.

A 90-Day SEO Web Design Programme

This is the sequence we use when taking on a site with the impressions-without-clicks problem. It assumes an existing site rather than a new build.

Weeks 1 to 2: Measure and Diagnose

Crawl the whole site and export every URL with its status code, title, meta description, word count and internal link count. Pull sixteen months of Search Console data grouped by page and by query. Identify every page with high impressions and an average position between 8 and 50, which is the striking-distance set where improvement pays fastest.

Pull Core Web Vitals field data for mobile and desktop. Record the current state so you can prove movement later. The output is a ranked list of pages by opportunity, not a list of problems by severity. Those are different documents and only one of them gets acted on.

Weeks 3 to 4: Fix the Foundations

Resolve indexation problems first, because a page Google cannot index cannot benefit from anything else you do. Fix broken internal links, redirect chains and orphaned pages. Correct the heading hierarchy where it is broken. Implement or repair the four core schema types. Address the single worst Core Web Vitals failure across the templates rather than page by page, since template fixes apply everywhere at once.

Weeks 5 to 8: Rebuild the Priority Pages

Take the top five pages from the opportunity list and rebuild them properly. Expand thin content to genuinely cover the subject. Rewrite titles and meta descriptions for click-through. Add the internal links pointing at them from your highest-authority pages. Add a conversion path appropriate to the intent.

Five pages done properly beats fifty pages touched lightly, and it gives you a clean test of whether the approach works on your site.

Weeks 9 to 12: Build the Cluster and Measure

Publish three to five supporting articles around the priority pages, each linking up to the relevant service page. Re-run the crawl and the Core Web Vitals check. Compare impressions and average position against the week one baseline for the priority set specifically, not for the site as a whole, because site-wide averages will drown the signal.

Expect movement in position and impressions by week twelve. Expect clicks and enquiries to follow in the quarter after, not within the ninety days.

What Happens After Day 90

The ninety days prove the model. What follows is the part that compounds, and it is where most programmes quietly stop. Months four to six are for widening the priority set. Take the next ten pages from the opportunity list and apply what you learned from the first five, which will be faster because the template fixes are already in place.

Months six to twelve are for the content clusters. This is the slowest and highest-return work, because a mature cluster ranks for hundreds of long-tail queries and keeps passing authority to the commercial pages indefinitely.

Throughout, keep re-running the crawl and the Core Web Vitals check monthly. Sites decay. Plugins get added, images get uploaded at full size, redirects accumulate, and a site that passed in March frequently fails by September without anyone changing anything deliberately. Budget for maintenance from the start. A programme that builds beautifully and then stops being measured reverts towards its starting condition within about eighteen months.

Why the Cheapest Fixes Come First

There is a sequencing argument that gets lost in most proposals, and it decides whether a programme survives its first review. Technical fixes and title rewrites produce movement within weeks. Content depth and cluster building produce movement across quarters. Both are necessary and only one of them will be visible when someone asks what has changed after a month.

So do the fast work first, not because it matters more, but because it buys the time the slow work needs. A programme that opens with three months of content production and nothing to show is a programme that gets cut in month four.

Sequence the work so that every review has something to report, and keep the long-range items running underneath. That is a political argument as much as a technical one, and ignoring it is why good programmes get cancelled while worse ones continue.

How GLO Approaches SEO Web Design

We do not start with the site. We start with a Growth Strategy, because a site built before anyone has agreed who it is for is a guess with a design budget attached.

That work covers brand archetypes, tone of voice, commercial objectives and the audiences most likely to buy. For each audience we build vertical profiles: the job titles that sign off, the pain points that drive the decision, the objections that slow it down, and the lead magnets that earn an enquiry.

Only then do we draw the sitemap, and by that point the sitemap writes itself, because we know which audiences need which pages and what each page has to do. The build follows with the technical standard agreed up front rather than tested at the end. Performance budgets, schema, accessibility level and the redirect map are all decided before design begins.

After launch the programme continues, because search results move. You can see the shape of that work across our projects and the wider digital marketing services that sit alongside it.

Bringing It Together

A site that ranks and converts is built that way from the sitemap onwards. Retrofitting it is possible and it costs more and delivers less than getting the foundations right at the start. The points worth carrying away:

• Impressions without clicks is the defining symptom, and it means relevant but not chosen
• Architecture decided in week one constrains everything you can do afterwards
• Core Web Vitals thresholds are LCP under 2.5 seconds, INP under 200 milliseconds and CLS under 0.1, measured on real users at the 75th percentile
• Service pages need a cluster of supporting content behind them to compete on anything commercial
• Average position moves first, clicks follow, enquiries follow clicks, and reporting that ignores the sequence gets programmes cancelled early
• You can test the whole thesis on three pages in ninety days before committing to a rebuild

If your site is collecting impressions and not clicks, that is a structural problem with a measurable value attached. Book a Growth Strategy session and we will work through your Search Console data with you and size the opportunity before anyone talks about a rebuild.

Frequently Asked Questions

What Is SEO Web Design?

SEO web design is the practice of building a website so that search engines can crawl, understand and rank it while visitors can use it and convert. It combines technical foundations such as site speed, crawlability and structured data with content architecture and conversion design.

The distinguishing feature is sequence. In SEO web design the search requirements shape the sitemap, the page templates and the performance budget before design begins, rather than being added once the site is built. That ordering is what separates it from a conventional web design project with SEO bolted on afterwards.

How Long Does SEO Web Design Take to Show Results?

Technical improvements such as Core Web Vitals and indexation fixes can register within two to four weeks, because Google recrawls and reassesses relatively quickly. Ranking movement on competitive commercial terms is slower. Expect visible change in impressions and average position within eight to twelve weeks, and meaningful change in clicks and enquiries across two to three quarters.

The variable that matters most is your starting position. A page moving from 43 to 25 produces almost no extra clicks despite being real progress, while a page moving from 12 to 6 can double its traffic.

Is SEO Web Design Different From Normal SEO?

They overlap heavily and differ in scope and timing. Ordinary SEO generally works within the constraints of an existing site, improving content, links and technical issues on foundations someone else laid. SEO web design sets those foundations. It decides the URL structure, the template architecture, the performance budget and the content model, which are the things ordinary SEO cannot change without a rebuild.

If your site’s structure is sound, you need SEO. If your structure is the thing holding you back, you need SEO web design.

Do I Need a Full Rebuild to Fix This?

Usually not, and it is worth testing before committing the budget. Most sites with the impressions-without-clicks pattern have three or four structural problems affecting a handful of high-value pages. Take the two or three pages with the highest impressions and worst positions, rebuild those properly, and measure over ninety days. If they move, you have evidence for the wider investment and a model that works.

A full rebuild becomes the right answer when the platform itself caps performance, when the URL structure is fundamentally wrong, or when the template cannot support the content depth the market requires.

How Much Should SEO Web Design Cost in the UK?

Pricing varies widely and the honest answer is that it depends on scope rather than on any standard rate. A focused programme rebuilding a handful of priority pages on an existing site sits at the lower end. A full rebuild with new architecture, content and technical foundations sits considerably higher, and ongoing programmes are usually retained monthly.

The more useful question is what the pages are worth. Work out the impressions your key pages already earn and model what a realistic position improvement would produce. If the modelled return does not comfortably exceed the quoted cost, the scope is wrong or the pages are not the right ones.