Twelve weeks. Four search surfaces — Google organic, Google AI Overviews, Bing/ChatGPT, and Claude/Perplexity/Gemini. Every action below tagged with priority, effort and likely impact for a small Yorkshire trade business. No fluff, no generic "do good SEO" essays.
In 2026 the answer to "who fits granite worktops in Leeds?" is increasingly given by ChatGPT, Claude, Perplexity and Google's AI Overviews before a customer ever clicks a search result. The job is to be the answer, not just be in the results.
Each AI search engine cites different sources. ChatGPT runs off Bing's index. Perplexity pulls heavily from Reddit. Google's AI Overviews scrape direct answers from indexed pages. Knowing the rules per surface is the work.
Google Business Profile drives Local Pack (~32% of signal). Schema-rich pages with direct answers earn AI Overview citations. People Also Ask is captured by Q&A format <h2>+50-word answers.
87% of SearchGPT citations match Bing's top results. A Bing Webmaster + IndexNow + Bing Places setup unlocks every Microsoft-side AI surface. Currently NCW has none of these.
Source: Seer Interactive SearchGPT analysis, Feb 202540% of Perplexity citations come from Reddit; 47.9% of ChatGPT citations from Wikipedia. Authentic Reddit presence + a Wikidata entity = the highest-leverage moves for Perplexity/Claude visibility.
Sources: Semrush 2025 · Soar AI citation analysisThe current robots.txt blocks ClaudeBot, GPTBot, anthropic-ai, Google-Extended, PerplexityBot, ChatGPT-User and 40+ others. NCW is invisible to every AI answer engine by design.
Single highest-leverage SEO move in 2026. 87% of SearchGPT citations match Bing's top results. Bing Webmaster is free, IndexNow pushes URL updates to Bing in real-time.
Set up via bing.com/webmasters · IndexNow integration: 1 API call per published pagePrimary category: "Countertop store". List all services. Upload 30+ geo-tagged install photos. Add Q&A. Post weekly. Set service area (Leeds, Bradford, Halifax, Harrogate, York). GBP drives ~32% of local-pack signal.
Source: Sterling Sky 2025 Local Search Ranking FactorsInclude address, geo, openingHoursSpecification, areaServed as discrete City entities (Leeds, Bradford, Halifax…), aggregateRating, priceRange, sameAs to social + directory profiles.
30% higher AI Overview visibility per Frase 2025 study. Wrap the existing FAQ in FAQPage + Question + Answer markup. Each answer ≤50 words for People Also Ask capture.
Each material (granite, quartz, porcelain, ceramic) gets Service markup with provider, areaServed, offers (price range), category. Also a Product entry for each branded slab (Dekton, Atlas Plan) to earn product-results citations.
Each Google review re-rendered on-site with Review + author + datePublished + reviewRating. Earns stars-in-results in Google + signals authority to LLMs.
6-step process (sketch → slab → template → fabricate → install → handover) marked up as HowTo earns the rich step-by-step result format that AI Overviews lift verbatim.
LLMs pattern-match against Wikipedia-style introductions and lift them verbatim into answers. One 60-word paragraph stating what NCW is, where, since when, what materials, who runs it — written like a Wikipedia lede.
Wikipedia is the #1 source for ChatGPT citations (47.9%). Wikipedia article requires notability NCW doesn't yet meet — but a Wikidata entity (no notability bar) helps disambiguation across every AI engine. Includes type, location, founders, founded year.
Source: Soar AI citation analysis 2025/llms.txt + /llms-full.txtHonest read: ~10% adoption, no AI provider has confirmed they read it, Google has rejected the proposal. Implement anyway — it's a 30-minute belt-and-braces play that costs nothing and demonstrates AI-readiness to any client doing due diligence.
Demo file: /llms.txt · Rankability adoption data Q1 2026Reddit is 40% of Perplexity citations and a growing share of ChatGPT. Genuine, not promotional. Ryan should answer 1-2 worktop questions per week with his real account; never plug NCW directly. Reputation builds the citation.
Source: Semrush AI citation source analysis 2025Apple Intelligence + Siri search uses Apple Business Connect data. Free, ~20 min to claim. Same listing surfaces on Apple Maps + macOS Spotlight.
LLMs extract tables verbatim. Each material page gets a tidy table: thickness (mm), heat tolerance (°C), Mohs hardness, edge profiles available, lead time, £/m² start. Extraction-friendly format outranks paragraphs for "what's the heat tolerance of porcelain worktops" type queries.
LLMs increasingly weight authorial trust. A profile page with named owners, qualifications, IASME / trade memberships, decade-by-decade history. Linked via sameAs from LocalBusiness schema.
Not 25 thin combo pages. Hubs at /granite-worktops/, /quartz-worktops/, /porcelain-worktops/, /ceramic-worktops/, /dekton-worktops/. City hubs at /worktops-leeds/, /worktops-bradford/ etc. Only spawn city×material combo pages (e.g. /granite-worktops-leeds/) for the top 2-3 intent pairs that justify 700-900 unique words.
Drip, not batch. After every install, send a one-tap review-request link. Respond to each review within 48h including the city keyword. A business with 50 recent reviews outranks one with 500 stale.
Source: Sterling Sky 2025 — review recency outweighs total countName, Address, Phone must match exactly everywhere. Audit and fix: Bing Places, Apple Business Connect, Yell, Houzz, Checkatrade, MyBuilder, Rated People, FreeIndex, Cylex, ThomsonLocal, plus all material-supplier "find a fabricator" pages (Cosentino, Caesarstone, Atlas Plan).
One Post per week with a recent install photo + city + material. Posts are short-lived but they signal active business + feed AI Overviews with structured facts.
Every city hub embeds the Google Map with NCW's pin + lists 3-4 FAQs about that specific city ("how long does install take in Bradford? do you cover postcodes BD1-BD19?").
Free, trade-credible, high-DA. cosentino.com/dekton/find-a-fabricator, caesarstone.co.uk/dealer-locator, iris-ceramica.com/atlas-plan. Each one is a single-form-application backlink from a domain Google trusts enormously for the topic.
Highest DA among UK trade directories, and increasingly cited in Google AI Overviews for "best worktop installer near me" queries. ~£35/mo paid plan worth it for the citation alone.
Government-endorsed (TrustMark = free). Which? is paid but earns the Which? badge — high consumer trust signal for kitchen renovators.
UK trade marketplaces. Profile + 12+ projects each. Backlinks moderate-DA but the review-aggregation is what compounds — adds to E-E-A-T signal across every search surface.
One feature per year — community story, charity project, local-business interview. Earns a high-authority local backlink and surfaces in "local trade news" LLM answers.
Stone masonry / kitchen-fitting course sponsorship in exchange for a .ac.uk link. Long-tail SEO and brand-building.
Each pillar targets a high-intent buyer query AND is structured to earn AI Overview citations (direct answer first, structured tables, comparison sections):
One per recent install. Format: location · cabinet maker · slab · sq metres · edge profile · turnaround · 6-10 photos · customer quote. Each case study double-duty as SEO content for that material+city combo AND visual social proof.
Twenty-four queries to dominate. Tagged commercial (buyer intent) vs informational (research intent).
Compounding work. P0 unblocks → Bing & GBP → schema & entity paragraph → material hubs → city hubs → pillar content → review velocity → external citations. Each week's lift compounds on the previous.
Most of week-one's work is 30 minutes of cleanup that the current site is actively losing money on. The harder work — content, citations, review velocity — pays back over a year. Twelve weeks gets to a defensible position; a year gets to dominant.