Reading the site before we redraw it.

A small dealer in north Cheam asked us to redesign their site. Family-run, more than fifty years on the same forecourt. Could we have something to show quickly?

We said no — and then said yes, but with a short discovery pass first. We didn't open a design file. We didn't sketch a hero. We read the existing site, with six pairs of eyes, and produced a single brief from it. The redesign came later, with reasons.

What follows is what we actually do during that discovery window. Most agency "discovery decks" we've seen are a logo, a moodboard and a paragraph about brand pillars, and the client deserves better than that. So: what we look at, who looks at it, and what comes out the other side.

the businessKnowing it before redrawing it.

Discovery starts with a short list of plain questions about the business and the website. They protect against every guess you'd otherwise make later.

Industry

Used commercial vehicle dealer — vans-first

Location

North Cheam, Sutton

Heritage

More than fifty years on the same forecourt

Business size

Small independent — about nine visible vehicles at review time

Customers

Sole traders, small businesses, trades — vans first; private buyers second

Price band

£7,495–£12,495 + VAT, finance from around £150/month

Reputation

RAC-approved, FCA-regulated, 5★ but thin — two AutoTrader reviews, no Google count surfaced

Main competitor

A neighbouring multi-site dealer with 300+ stock, walking distance, owned by a 1904 motor group

Differentiator

Half a century on one forecourt — family-run, no-pressure, every credibility marker under one independent roof

Most of those answers were already on the website. They weren't loud. The heritage line appeared once, in body copy, mid-paragraph. The single most valuable thing this brand owned was a footnote.

That's discovery's first job: surface what's already true. Half of redesign is amplifying things the client has already paid for and is currently underplaying. We don't have to invent a story — we have to listen for the one that's already there.

six readersOne brief. Six disciplines.

We don't run discovery as one person. We split it across specialists, because the technical pass finds things the brand pass misses, and the brand pass finds things SEO misses. Each writes findings in their own voice; the lead stitches one document together at the end.

• Hannah — Client discovery & lead+

  Holds the brief. Covers business overview, audience, goals, conversion path, success measures, scope, risks. Owns the final document — and is the one who refuses to ship it if any answer is invented rather than evidenced.

• Ibica — Technical audit+

  Reads the site like a stack. Found a Hugo build six versions behind, an undocumented vendor framework holding the entire data layer, a finance calculator with the API key in the page source, two CDNs stacked on each other, and a reCAPTCHA key likely tied to the platform vendor's GCP project.

• Nathan — SEO & local discoverability+

  Found that the website's phone number and the AutoTrader phone number don't match. A NAP inconsistency that hurts local discoverability. Also found four URLs in the JSON-LD rendering as `https:/…` with a single slash — invalidates rich-result eligibility across the site, was a five-character template fix.

• Grace — Content & messaging+

  Caught the language drift — finance and warranty pages talked about 'cars' as often as 'vans', even though the business has been van-first for half a century. Generic CTAs everywhere: 'Read More +', 'Send', 'Next'. The page wasn't asking for the sale it was supposed to ask for.

• Sofia — UX & conversion+

  Read every journey on mobile first. The vehicle pages had no sticky bottom action bar, the header phone number opened a modal instead of dialling, and the inventory cards offered no way to call, enquire or check availability without a full page load. The mobile journey felt years behind the desktop one.

• Clara — Brand & visual+

  Wrote the cleanest sentence in the brief: 'The site has no brand system — only a logo and accidental colour. Visual ownership of the page currently belongs to the RAC and the finance companies, not the dealer itself.'

The point of splitting it isn't to produce more documents. It's to catch contradictions early. The SEO recommendation to add a service-area page collides with the content audit's worry about thin pages. The technical recommendation to migrate to a direct AutoTrader feed depends on the client owning the feed credentials — which the technical audit flagged as an open question rather than an assumption. Those conversations happen in writing, on a single ticket, while the client is still drinking their first cup of tea.

One brief leaves the studio. The rest stays in the ticket.

findingsRanked, not just listed.

Listing every problem on a website is easy — most automated audits will give you four hundred. The hard part is ranking. A list isn't a finding; a ranked list is.

Critical is launch-blocking — the kind of thing that loses money or trust on day one. High leaves CTR or conversions on the table but doesn't burn the launch. Medium is a slow leak.

This is the part where ranking pays for itself. Most of the launch-day arguments — "we're not paying for that, it's not important" — are easier to have when everyone is looking at the same word next to the same item.

brandA small, decided palette.

Sites without a brand system borrow colour from their partners. On the existing site, RAC orange and the finance partner blues were doing the heavy chromatic lifting. Strip the logo off and you couldn't tell whose website it was.

So we drew a small, owned palette, and a B option, and let the client see both.

We always show the client a B option. Not three, not seven — one well-argued alternative. The point isn't to let the client paint-pick. The point is to make sure the choice between A and B is the choice, instead of being argued about by accident for the next two months.

deliverablesThree documents, no slides.

Discovery ends with three things, all in plain text, all settled, tracked, version-controlled:

• The discovery brief. A single markdown document, roughly five thousand words, with assumptions labelled and open questions in their own section. Section sign-offs at the bottom, one per specialist.
• The HTML design concept. A working browser preview, two files, opens locally. Real type, real components, real responsive behaviour. Not a Figma image.
• The design-system markdown. Palette, type, spacing, motion, components, accessibility floor. Tokens are written in YAML inside the document; the components reference semantic tokens, so switching palette is a one-line change.

discovery/
├── discovery-brief.md       ~36 kb
├── design-system.md         ~31 kb
└── concept/
    ├── index.html           ~93 kb
    └── vdp.html             ~57 kb

None of it is decorative. The brief is signed off line-by-line in the ticket. The HTML is the thing developers build against. The design-system markdown is the thing future-us reads, six months later, to remember why anything is the colour it is.

Slides would be faster to produce. They would also be worse, because they hide the why.

restraintWhat we didn't do.

We didn't promise the redesign would add 30% more leads. We don't have a baseline yet, we don't know which months the dealer is currently soft, and we won't invent figures to sell ourselves.

Discovery sometimes lands on a different answer entirely — that the existing site doesn't need redrawing yet. Whether a rebuild is the right call is part of what the reading decides.

We didn't write the final copy. The brief tells the copywriter what each page needs to do; the words come once the structure is signed off.

We didn't propose a new CMS for a small dealer inventory. The stock is feed-driven; the inventory doesn't need an authoring tool, it needs a feed.

We did flag about thirty open questions — split into positioning, reputation, service experience, proof, technical and commercial — and put them in their own section of the brief. Each one is something we could have guessed at, but every guess in a discovery document is something that breaks later, in front of the client, in a meeting that costs everyone an afternoon.

The brief says, often, "Assumption — clearly labelled." That's the part of the document we're proudest of.

why we do thisA small upfront cost.

A short discovery pass is a small expense. It looks like a delay to a client who wanted to see designs straightaway. It is, in fact, the thing that lets every later week go faster.

When the visual designer starts, they're not staring at a blank page — they have ten constraints, three differentiators, and a written brand voice. When the developer starts, they know which vendor lock-ins to plan around. When the copywriter starts, they know what every page is supposed to do, not just say. When the client opens the first concept, the brief is the thing they've already signed off on — so the only debate is whether the rectangles match the words, not whether the words were right.

That's the part we put pride into. Not the rectangles. The reading that comes first.

If you've got a site that could use this kind of reading before it gets redrawn, we'd be happy to hear from you.