Cordley Solutions: A Brochure Site With One Job, Passing Supplier Due Diligence

Most websites have a dozen jobs. This one has exactly one. Cordley Solutions is a brand-new UK consumer-brands distribution business, and its website exists to pass supplier due diligence: to convince an established brand, sitting on the other end of an introduction email, that this young company is registered, real, run by named people, and worth a trading relationship. Everything else is secondary. This case study covers the build we shipped in three days in July 2026, and in particular the decision at its heart: the kickoff audit that measured the client's own brand palette against WCAG 2.2, found the accent colour failing at 2.59:1, and dropped it, with the client's agreement, before a single pixel was designed. It is a study in doing the measuring first, and, because the site is days old, it deliberately claims nothing about rankings that has not happened yet.

The Client: A New UK Consumer-Brands Distributor

Cordley Solutions Limited was incorporated in 2026 (Companies House number 17195360), with its registered office in Kington, Herefordshire. The business sources established consumer brands and supplies them to buyers through leading online marketplaces, working across the household, health and beauty, personal care and FMCG categories. Our contact throughout was Reiss Johnson, a director named publicly, alongside his co-director, on the company's about page.

Reiss came to us with something rarer than a budget: a genuinely considered brief. He had designed the company logo himself, specified a three-colour palette, and supplied five websites he admired as design references. He also gave us the single most useful sentence a web design client can give, which is what the site is actually for. Not traffic, not e-commerce, not brochure vanity. The site's job is to be the thing a potential supplier finds when they check the company out.

We build for this sector often enough that wholesale distribution websites have their own page in our portfolio, and the pattern repeats: a distributor's website is read most carefully not by customers but by the brands it wants to buy from. That reader arrives sceptical. Design for that reader and everything else follows.

Why a Young Company's Website Is Its Due-Diligence Pack

Here's the thing about a company that is months old: it has no trading history to point at, no decades of reviews, no case studies of its own. When a brand manager receives an approach from an unfamiliar distributor, the first thing they do is search the company name. What they find in the following ninety seconds decides whether the conversation continues. The website is not marketing at that moment. It is the due-diligence pack.

So the build order came straight from what a sceptical supplier checks. The Companies House number is rendered in a trust card directly beneath the homepage hero, not buried in the footer small print. The registered office is stated plainly. Both directors are named, with real backgrounds, on the about page. And, crucially, the copy refuses to inflate. Cordley's own where-we-are-now section says it outright: "We are at the beginning. The company is registered, the groundwork is done, supplier conversations are underway, and we are building the first relationships now."

That sentence was a deliberate strategy, and it took a conversation to land. The instinct of nearly every new company is to look bigger: plural "teams", invented client counts, stock photos of warehouses they don't run. The problem is that the due-diligence reader is precisely the reader who checks. A brand manager who searches the company number and finds a 2026 incorporation date will not be angry that the website said "we are early"; they will notice that it told the truth, and they will weight everything else on the site accordingly. Honesty is not the risky option here. It is the only strategy that survives contact with a reader who verifies. We wrote the whole supplier partnerships page in that register: "we would rather earn a few good ones than chase everybody."

We have written before about why Google rewards exactly this kind of non-commodity honesty, and it applies double when the audience is human due-diligence rather than an algorithm.

The Palette Audit: Brand Colours vs WCAG 2.2

Now the decision this case study exists to document. The brief specified three brand colours: navy #0F2744, silver #8A9199, and an accent orange, #F58220, on white and off-white grounds. Standard practice in most studios is to take the palette as given and design with it. We ran an accessibility audit on it first, at kickoff, on 16 July 2026, before any design work. I have audited enough inherited sites to know what it costs to discover a contrast failure after the design is built, so on new builds the palette goes under the meter on day one. The full method, including the runnable contrast script and the colour families that usually fail, is written up in our guide to testing brand colours against WCAG 2.2.

WCAG 2.2 AA sets two thresholds that matter here. Body text needs a contrast ratio of at least 4.5:1 against its background. UI components, button borders, form controls, need at least 3:1.

"The visual presentation of text and images of text has a contrast ratio of at least 4.5:1"

W3C, Understanding Success Criterion 1.4.3: Contrast (Minimum), WCAG 2.2

The measurements were blunt:

• Orange #F58220 on white: 2.59:1. Fails AA for body text (4.5:1 needed) and fails for UI components too (3:1 needed). It cannot legally carry text and it cannot even outline a button.
• White text on an orange fill: also 2.59:1 (contrast is symmetric). So the single most obvious use of an accent colour, an orange call-to-action button with a white label, is not AA-legal either.
• Silver #8A9199: 3.19:1 on pure white, dropping to 2.94:1 on the site's off-white ground, where it fails even the 3:1 UI threshold. The client's second colour could not be light-ground text at all.
• Navy #0F2744: 15.06:1 on white and 13.92:1 on the off-white. Comfortably past the stricter 7:1 AAA threshold. The client's primary colour was, measurably, excellent.

One rescue for the orange existed on paper: darkening it to #B34F00 brings it to 5.20:1, which passes. We rejected that too, because at #B34F00 it is no longer the colour the client chose; it reads brown. An accent that has to be disguised to be usable is not an accent worth keeping. Navy text on an orange fill does pass at 5.81:1, but restricting a brand colour to one contrived combination is a workaround looking for a problem.

The silver had a better ending. Rather than banning it, we followed the logo's own logic: in the client's mark, the silver wings sit against navy, never against white. So silver became a dark-ground and graphics colour, where it measures 4.73:1 on navy and passes, and a darker steel grey, #5B6672, was introduced to carry secondary text on light grounds at 5.41:1. The palette the site ships passes AA on every pair that actually renders, most of it at AAA.

One Colour Too Many: Measuring Five Reference Sites

Contrast maths alone rarely wins a palette conversation, because the client hears "your colour is wrong" and digs in. So we did the second half of the audit: we measured his taste. Reiss had supplied five reference sites he wanted the design to feel like: LARQ, Ruggable, Patchworks, WatchHouse and Gymshark. Instead of describing them, we extracted their computed styles with a headless browser and pulled the actual colour tokens each one renders.

The finding made the argument for us. Not one of the five uses an accent orange. Not one uses an accent colour in the way the brief imagined at all. Their premium feel comes from restraint: navy or near-black ink, greys, off-white grounds, big type, generous whitespace. Better still, the measurements validated the client's own instincts with uncanny precision: LARQ's navy is #153a5b against his #0F2744, and Patchworks' grey is #868f99 against his #8A9199. He had independently chosen almost exactly the palette his favourite sites use.

That turned the conversation on its head. The finding was never "your palette is wrong". It was "your instincts are right, you added one colour too many". Shown the contrast table beside the reference measurements, Reiss chose to drop the orange entirely, over darkening it, over restricting it, without us having to push. The evidence did the persuading, which is the whole point of measuring before designing. In three days there was no time for a redesign loop, and the audit is why none was needed.

Design Restraint as the Premium Signal

With the palette settled, the design system followed the same discipline. One content width on every page. Two typefaces: Manrope for display and body, IBM Plex Mono for eyebrows, category labels and the company number, a nod to the fact that distribution is a business of SKUs, references and catalogues. The client suggested Inter and Poppins; we talked him out of both, because they are what a design looks like when no type decision was made. Imagery is AI-generated (a months-old company has no photography), art-directed to be light, airy and desaturated so it sits inside the two-colour system instead of fighting it.

Restraint also meant deleting our own ideas. The logo's silver wings were tried as a recurring inline marker and dropped when, at small sizes, they read as a dropdown arrow. A decorative angled image crop was cut because it skewed the product photography. On a site whose reader is deciding whether a company is credible, every flourish is a liability; quiet competence is the aesthetic. The type scale was deliberately kept confident rather than shouting, because an oversized hero on a young company's site reads as swagger, and the due-diligence reader is allergic to swagger.

The same discipline that drives our WordPress design work applied here even though, full disclosure in line with our editorial standards: this site is hand-coded PHP, not WordPress. Seven pages, no CMS requirement, aggressive performance targets; hand-coding was the honest tool. The content model, schema and discovery patterns transfer directly to the WordPress builds that make up most of our work, exactly as they did on the Brewood Removals rebuild.

Organization Schema, Not LocalBusiness: The Registered-Office Call

Structured data on this build involved a decision most agencies get wrong by default: what the business is, in schema terms. The reflex for any UK company with a postcode is LocalBusiness markup, because it feels richer and every SEO checklist rewards it. We deliberately did not use it, and the reasoning matters.

Cordley Solutions' only public address is its registered office in Kington. It is a filing address, not premises a customer can walk into. Google's business representation guidelines are unambiguous about this pattern:

"If your business rents a physical mailing address but doesn't operate out of that location, also known as a virtual office, that location isn't eligible for a Business Profile."

Google, Guidelines for representing your business on Google

A distribution business that trades through online marketplaces has no shopfront and no service area in the local-search sense. Marking it up as LocalBusiness would tell search engines a story the business cannot back with a Google Business Profile, and inconsistency between schema and reality is exactly the kind of signal that erodes machine trust. So the site ships a detailed Organization node instead: legal name, founding date, the Companies House number as an identifier, both directors as founder Persons with their credentials, and knowsAbout entries for the product categories. Everything in the markup is checkable against a public register, which, on a site built for due diligence, is the point. The schema tells the same story as the copy: real company, registered office, named people, no local-business pretence.

AI Discovery Files From Day One on a Zero-History Domain

The second structural investment was the full set of AI Discovery Files at the site root from launch day: llms.txt, a brand-matched human-readable llms.html, ai.txt, ai.json, identity.json, brand.txt, faq-ai.txt, developer-ai.txt and robots-ai.txt, implementing the AI Discovery Files specification published by the AI Visibility Definition.

For an established business, discovery files improve how AI systems describe something they already know. For a zero-history company the case is stronger: they are close to the only source. When a brand manager asks ChatGPT or Gemini "who are Cordley Solutions?", there are no news articles, no reviews, no Wikipedia entry to draw on. Whatever structured, honest, machine-readable information the company publishes about itself is effectively the whole corpus. We watched AI assistants pick up a new site within days on the Lockerfella build, which hit number one on ChatGPT and Gemini inside three weeks; the groundwork that made that possible is the same groundwork laid here. And, in keeping with a case study built on re-checkable claims, this one takes a minute to verify: run the domain through the free AI Visibility Checker, or open Cordley's AI visibility profile, where the site's verified discovery-file status is public.

One implementation detail worth stealing: the files are for AI crawlers, not for Google's search results pages, so robots.txt disallows them for Googlebot specifically while leaving every AI crawler, including Google-Extended, free to read them. An earlier blanket disallow would have blocked the very crawlers the files serve; scoping the rule to Googlebot keeps raw JSON out of search results without defeating the files' purpose. Small file, easy to get subtly wrong.

The Search Strategy: Ranking for the Queries That Matter

What should a brand-new distribution business even try to rank for? The honest answer shaped the whole content plan, and it starts by conceding most of the map. Head terms like "wholesale distributor UK" belong to established directories and household names, and a domain registered weeks ago will not take them off anyone. Chasing them on day one would mean stuffing the site with content its reader does not need, which would damage the one thing the site must do well.

The queries that actually matter to this business, in order:

• The company name. "Cordley Solutions" is the search every supplier makes after an approach. The site must own it completely: every page title carries the brand, the Organization schema nails the entity, and the discovery files give AI assistants a first-party answer. This is the query where losing is not survivable, and the only one that matters in week one.
• Verification long-tails. The company name plus "Companies House", "reviews", "directors". The about page and the schema are built to satisfy these before the searcher has to leave Google.
• Commercial terms, earned slowly. "UK consumer brands distributor", "consumer brands distribution UK", supplier-side phrases like "sell to UK online marketplaces". The supplier enquiry page and the category pages target these with real copy rather than keyword walls, and they will accumulate authority at the pace the business accumulates history. No shortcuts exist, so the site does not pretend to take one.

Everything a searcher needs to complete due diligence, including getting in touch, sits one click from any landing point. The information architecture is the strategy; the rankings, where they come, will be a consequence.

What We Can't Claim Yet: Rankings and Traffic

Our case studies carry this section by policy, and this one needs it more than most. The site went live on 18 July 2026. At the time of writing that is four days ago. So, plainly: we claim no rankings, no traffic, no leads, no AI citations. Not because the early signs are bad, but because at four days old there are no signs of any kind worth reporting, and a case study that reaches for them anyway is inventing its ending. We took the same position on the Brewood Removals rebuild, days after its launch, and the discipline is now house style.

Launch week did hand us one honest war story, though. The contact form's hidden anti-spam field, named like a real company-website field, was being silently filled by browser autofill, which made the handler treat genuine submissions as bot traffic. The page said "message sent"; nothing sent. We caught it in the logs within hours of launch, renamed the field to something autofill ignores, verified the fix end to end, and folded the lesson into how we build every form since. Case studies that only contain victories have had their defeats edited out; this one keeps the receipts.

What is verifiable today: the live site, the palette that passes AA on every rendered pair, the Organization schema you can check against Companies House record 17195360, the nine discovery files at the root, and the launch date in this study. When real Search Console data exists, for the brand query first, this page will be updated with it, and the update will be logged, whichever way the numbers point.

Frequently Asked Questions