Diamond Stone: the home page and landing pages
Date: 2026-11-11. Revision 1. Part of the website plan (docs/WEBSITE.md). Drawn in
docs/website-map.html sections 14 and 15; screens in docs/ui-shell-mockup.html
under "Website: the home page, top to bottom" and "Website: landing pages".
How the home page is built, part by part and why in that order; and a system of landing pages made from the same few blocks. Research was done on 11 November 2026. Each point says how strong its evidence is, because much of what is written about landing pages is sellers' claims. Nothing here has been tested with visitors.
1. What the evidence supports, and what it does not
Supported
| Finding | Source and its limits | What we do |
|---|---|---|
| A home page must say what the company does in one plain sentence, give clear starts for the main one to four tasks, and show real content | Nielsen Norman Group. Old (2002), never withdrawn | Section 2 |
| A rotating banner is ignored. About 1 % of visitors clicked one, and 84 % of those clicks were on the first slide; moving panels are read as advertisements | One university's figures (2013); NN/g observation | One still photograph. Nothing rotates |
| Anything that moves by itself for more than five seconds needs a pause control | The accessibility standard (WCAG 2.2.2) | No auto-playing video or carousel |
| On a phone, a menu hidden behind an icon is used far less: 57 % of cases against 86 % when some links are also visible | NN/g, 2016, 179 people | A row of visible links under the header, as well as the menu |
| Navigation by audience ("I am an architect") is usually a problem: people fit several groups or none, and fear missing something. Acceptable only as a secondary route, labelled "For architects" | NN/g, 2015. Four stone and tile suppliers' home pages read today all lead with product and use, not audience | Changed from my first draft, which led with three audience doors. Stones and uses lead; "For…" links come lower |
| Most viewing time goes to the first screen or two, then falls away | NN/g eye tracking, 2018. Desktop only; no phone figure was found | Parts are ordered by importance; nothing vital is left for the bottom |
| A familiar layout with low visual clutter is judged more appealing within a twentieth of a second | Tuch and others, 2012. Measures appeal, not enquiries | A plain, expected layout |
| Faster pages on phones go with more people reaching the form | A study paid for by Google (2019, 37 brands). Correlation | The speed budget in docs/WEBSITE.md section 6; the main photograph is loaded first and is a real image in the page |
| People trust what others say about a company more than its own testimonials | NN/g | Proof leads to reviews on the company's public profile, not to quotes we chose |
| Pages for regions, uses or audiences that are near-copies funnelling to one page are treated by Google as spam | Google's policy, read directly | Section 4: each landing page needs something of its own and a place in the menu |
| No automatic redirect by country or language; one address per language | Google, read directly | A suggestion the visitor can dismiss, never a redirect |
| With this little traffic, most A/B tests cannot reach a result: about 13,000 visitors per version are needed to detect a one-fifth improvement on a 3 % rate | A standard calculation, not a study | Section 6: five-person tests first |
Not supported: sellers' claims, treated here as things to try
Removing the menu from a campaign page (one seller's small tests; a counter-example lost a third of its results). Fewer form fields always winning (one well-known test lost 14 % by cutting fields). A "typical" landing page rate of 6.6 % (from pages built on a seller's own tool). Sticky buttons, wording rules for buttons and headlines, "one goal per page" ratios, and personalisation gains. Figures about software buyers applied to stone buyers.
The finding that cuts against a brand rule
In NN/g's study of business buyers, price was the most wanted of 28 kinds of information, and its guidance where exact prices cannot be shown is to give a typical price for a typical case, or a range. The study is old (2006) and not about stone, but it is the strongest evidence found on what such buyers want.
The brand rule is "never lead with price", and nothing here leads with it. But showing nothing is a different choice. This plan already shows the sample's price and "what changes the price". Whether to add a typical case ("a 200 m² floor in honed Desert Rose, delivered in Muscat, was quoted at about …") is a decision for the owner (section 8). The earlier documents disagreed on this; the evidence leans toward showing something.
2. The home page
One long page. Each part answers the next question a stranger asks, in the order they
ask it. The one action, "Request samples", stays at the right of the action bar throughout,
with the word "WhatsApp" at the left (docs/SITE-DESIGN.md section 21).
| # | Part | The visitor's question | What it shows | Filled from |
|---|---|---|---|---|
| Header | Short logo, language, search, menu; and beneath it four visible links: Stones, Uses, Journal, Contact | Fixed | ||
| 1 | Opening | What is this? | One still slab photograph, full width. The heading "Omani marble and limestone" and one line: "Tiles, slabs and cut-to-size stone, supplied from Oman." | The owner chooses the photograph each season |
| 2 | Stones | What do you have? | A row of four to six stones as photographs, then "All stones" | The catalogue; order set by marketing |
| 3 | Uses | Will it suit my job? | Four tiles: floors, walls, bathrooms, outdoors. Each leads to its by-use page | By-use pages that exist |
| 4 | Proof | Can I trust you? | Two or three figures from our own records, with their period; a link to public reviews | The system; the owner approves each figure |
| 5 | How buying works | How do I buy, and what will it cost? | Four steps, the first with the sample's price; a link to the full page | Store settings |
| 6 | A project | Has it been used well? | One project: photograph, stone, area, what was delivered | Projects with consent; otherwise "stone in use" the company owns |
| 7 | People ask | What do others want to know? | Three real questions, then "All answers and guides" | The most-read answers |
| 8 | For… | Is there something for people like me? | Three plain links: for architects, for traders, for my home | Secondary, by the evidence |
| 9 | A person | Who will I deal with? | A named person, languages, reply time, WhatsApp and call | By the visitor's region where known; otherwise the default |
| 10 | A place | Where are you? | Showroom and yard, open now or not, directions | The company record |
| 11 | The company | Is this real? | Legal name, registration and tax numbers, legal links, the other language | The company record |
What it does not have: a rotating banner; video that plays by itself; a pop-up on arrival; a newsletter box; an audience chooser at the top; a price as the headline; a form.
For someone returning: the same page for everyone, as search engines see it. A strip is added on the phone from what it already holds: "Your list: Desert Rose, Sahara Beige", with "Samples of both". For a known customer: "Open your orders".
The four starting tasks (NN/g: one to four): request samples; browse the stones; find a stone for a use; ask a person.
In Arabic the page mirrors, with the menu at the right, where one small study found Arabic readers rated sites higher. Evidence on Arabic design is thin; this is tested with Arabic readers before launch.
3. Blocks
The home page and every landing page are built from the same fourteen blocks. A block is filled from records, never typed into the page, so a landing page cannot disagree with the stone page.
| Block | Shows | Filled from |
|---|---|---|
| Opening | A photograph, a title, one line | The page |
| Answer | Two lines that answer the page's question | An Answer record, or written for the page |
| Stones | A row or grid of stones | Chosen, or by rule (colour, use) |
| Stones that pass | A table of stones with the values that matter for a use | Facts |
| Uses | Tiles for each use | By-use pages |
| Steps | Numbered steps | "How buying works" |
| Figures | Two or three numbers with their period | The system, approved |
| Project | One project | A Project record |
| People ask | Two to four answers | Answers |
| A person | The contact for this page | People, by region or language |
| A place | Showroom, hours, directions | The company record |
| Offer | The sample, with price, delivery and credit | The sample product |
| Downloads | Datasheets and other files | Downloads |
| The company | Legal facts and links | The company record |
Each block is one of the 96 components or a pattern made from them
(docs/COMPONENTS.md). Fourteen is a limit, not a start: the earlier plan in
docs/planning/02 listed several dozen, which is more than one person can keep
consistent.
4. Landing pages
A landing page is any page built to receive a particular kind of visitor. Two families, with different rules.
Pages meant to be found in search
They live in the menu, are indexed, and never end.
| Kind | Example | Must have of its own | One action |
|---|---|---|---|
| By use | Marble for bathrooms | The reasoning for that use, the values that matter, which stones pass and why | Samples for that use |
| By market | Delivering to the United Arab Emirates | Route, usual days as a range from our own deliveries, papers, who clears, currency, a contact for that country, a project there | A quote, delivered |
| For an audience | For traders and importers | What that buyer asks first, and files or terms made for them | Talk to sales |
| By kind of project | Stone for hotels | At least two real projects of that kind | A quote |
| A group of stones | Beige marbles | What tells them apart, side by side | Samples of the group |
| A comparison | Desert Rose or Sahara Beige | Both stones' own values and photographs at the same scale | Samples of both |
The rule that keeps these honest, and inside Google's policy: a page is refused at the publish check if it differs from another only by a word. It needs its own photographs, data or project, and a place in the menu. "Marble for kitchens" cannot be "Marble for bathrooms" with the room changed. Drawn in the mockup as a blocked page.
Pages for a campaign or a code
They receive people from one advertisement, post, message, leaflet or printed code.
| Kind | Example | Rules |
|---|---|---|
| Campaign | "Desert Rose, in your hand this week" for an advertisement in the Emirates | The same words as the advertisement. One action. Has an end date. Points search engines to its parent page, so it does not compete with it |
| Trade fair or event | A code on the stand | Names the person they met, the stones shown, one offer. Ends a month after |
| Sample, card, brochure, crate | The code on a sample | Opens the stone or person, knowing which sample or card (docs/WEBSITE.md section 16) |
| A link from sales | A page sent in WhatsApp | Any page, with the sender named at the top |
- A campaign page keeps the logo, the language switch and a way to the rest of the site. Removing the menu entirely is a claim, not a finding; it is one of the first things to test.
- Every campaign page has an end. On that date it redirects to its parent page.
- Every link and code is made in one place in the staff app (drawn), so each enquiry knows the advertisement, post or leaflet it came from. Nobody hand-writes tracking codes.
- The form asks three things (name, phone, what they need). The evidence that shorter forms always win is weak; what matters is that each field is plainly needed.
- WhatsApp is offered beside every action. Its use in Oman is near-universal (a national survey, unverified); no evidence was found on its use for business buying in the Gulf, so its share of enquiries is measured, not assumed.
What may change for a visitor, and what may not
| May | May not |
|---|---|
| A strip added on the phone from what it holds: their list, their last stone | Different content for search engines than for people |
| A dismissible suggestion: "This page is also in Arabic" | Redirecting by country or browser language |
| The contact person chosen by region, where the visitor has said it or arrived on a market page | Country-specific facts that exist only in a strip. They must also live on the market page, at its own address |
5. Running it
Under Reach → Website → Landing pages. Three screens are drawn.
| Screen | Does |
|---|---|
| Landing pages | Every page with its kind, state, visits and enquiries; campaign pages with days left; pages blocked for being too alike |
| A page | Blocks in order, moved by drag or by arrows; each block filled from records; the page's kind, one action, whether it may be indexed, when it ends |
| Campaign link | Makes the tracked link and the printed code for a place (a social platform, WhatsApp, print, email) |
Publishing uses the same checks and second-person review as every other page
(docs/WEBSITE.md section 5), plus three: the page has its one action; it is not too
like another page; a campaign page has an end date and a parent.
Automatic: a campaign page redirects on its end date and tells the search engines; a block that draws on a changed fact is rebuilt; a landing page with no enquiries in 90 days is listed for review; each enquiry carries the page and the link it came from.
Advertising is not planned here. If it is used, each platform checks the page an advertisement leads to: it must load, match the advertisement, and be reachable in the country targeted. Click-to-WhatsApp advertisements and each platform's rules are to be looked at then. Consent: Oman's data protection law has been enforced since February 2026 and requires explicit consent; whether campaign tracking and cookies need a notice is a question for the lawyer before any campaign runs.
6. Finding out what works
| Method | When | Why |
|---|---|---|
| Five people from each kind of buyer use the home page on their own phone while we watch | Before launch, and after any large change | Finds most problems; the only method that works at low traffic |
| The five-second check: show the opening, hide it, ask "what does this company do?" | With each new opening photograph and sentence | A method, not a score. If they cannot say, the sentence is wrong |
| Arrivals through to enquiry, by page and by source | Always | Already planned (docs/WEBSITE.md section 19) |
| What people tap first on Home; how far they scroll | Monthly | Shows which parts earn their place |
| A split test | Only on pages with enough visits, and only for a large change | Most would never reach a result. Cloudflare can split a static page when one is worth doing |
What to try first, since the evidence does not settle them: the menu on campaign pages; the order of Proof and Uses; WhatsApp or form as the main action in each market; a typical-case price, if the owner allows one.
7. Speed and access, for the opening
- The opening photograph is a real image in the page, loaded first, never delayed, under 200 KB.
- Text is never part of an image. Text over the photograph keeps its contrast by sitting on a tinted band, checked over the lightest and darkest part.
- Space for the photograph is reserved, and fonts are matched, so nothing jumps as it loads.
- Nothing moves by itself.
8. Decisions
- A typical-case price on "How buying works". The evidence leans toward yes; the brand rule is unchanged either way, since it would never lead.
- The opening. Proposed: the heading "Omani marble and limestone" and the line "Tiles, slabs and cut-to-size stone, supplied from Oman."
- Which figures to show as proof, and who approves them each quarter.
- Which landing pages first. Proposed: four by-use pages, one market page (the Emirates), "For architects", "For traders".
- Whether advertising is planned at all. It decides whether campaign pages are built in the first release.
- Home owners (still open from
docs/WEBSITE.md): it decides the third "For…" link.
9. Sources
Read directly: Nielsen Norman Group on home page guidelines, business-buyer usability, scrolling and attention, carousels, hidden navigation, audience-based navigation, trust, showing prices, and how many test users; Tuch and others 2012; the Notre Dame carousel figures; WCAG 2.2.2; web.dev on the largest content paint and on the Google-commissioned speed study; Unbounce's 2024 benchmark report; HubSpot on navigation; Google's spam policies, duplicate addresses, localised versions and multi-regional guidance; Google Ads destination policy; a legal note on Oman's data protection law; Cloudflare's split-testing guide; the home pages of Polycor, Porcelanosa, Stone Source and Caesarstone. From summaries only (unverified): Gartner, 6sense and TrustRadius buyer figures; form-length studies; WhatsApp use in Oman; the Arabic menu study; Lindgaard and others 2006. No evidence was found on which kinds of landing page bring enquiries for a stone supplier.
Name pages (added 11 November 2026)
Superseded on 12 November 2026 by "Name pages, changed" below and
docs/CATALOGUE.md"Revision 2". Kept for history; do not build from it.
A seventh kind of page meant to be found: a name page, for a market name that is
ambiguous, is used for stones from several places, or has real search demand. It
explains the name and leads to the stone. It is subject to the same rule as every
other landing page: it must say something the stone page does not
(docs/CATALOGUE.md section 5).
Look (added 11 November 2026)
The home page and landing pages take the look in docs/SITE-DESIGN.md. The eleven
parts and fourteen blocks are unchanged; "blocks" are editorial sections, not cards.
Name pages, changed (11 November 2026)
The explanatory name page is withdrawn. A market name now has an address that shows the
stone's own page headed by that name, with nothing explained
(docs/CATALOGUE.md, "Revision 2").
Matching (added 11 November 2026)
The home page gains one part, placed second, after the opening: a laid-out floor with
"Tiles and slabs that match" and three lines (one lot, one finish, one thickness),
leading to "How we match". It is also a block for landing pages. See
docs/MATCHING.md section 7.