SEO Agency & SEO Services, From Research to AI Search

A good SEO agency does not start with keywords or backlinks. It starts with the people who buy from you, turns their questions into a page map, builds a site that search engines and AI assistants can read without effort, then measures and refines every month. That is the workflow RAIN Design Studio runs, end to end, for companies in the UK, Europe, the USA, Canada and the Gulf, from our studio in Casablanca, Morocco.

What our SEO agency does

Our SEO services cover the full chain: buyer research, keyword research, information architecture, technical SEO, content, structured data, local and international SEO, AI search, earned links and measurement. Because RAIN is also a design and engineering studio, we change templates, schema and server rules ourselves rather than sending your developers a list of issues to schedule.

We work remotely. We have no offices in London, New York or Dubai. What we do have is a documented method that we run on our own site first. This site’s homepage scored Performance 96, Accessibility 96, Best Practices 100 and SEO 100 in a Lighthouse run on 2 October 2026, with FCP 1.5 s, LCP 2.5 s, TBT 0 ms, CLS 0 and a Speed Index of 3.2 s. That is one lab test of one page, not a ranking, but it shows the standard we build to. The full toolchain is set out on our RDS Engine page.

The ten phases below are the complete workflow. Each one lists what we do, what you receive and why it matters.

Phases 1–3: research and architecture before any page

1. Buyer and market research

We start with the people who sign the contract. We read sales-call notes, CRM records, support tickets and reviews, and interview your sales team to list who buys, the job they need done, their objections and the questions they ask before committing. We also identify who actually wins your searches today, which is often not your business competitors but directories, publishers and marketplaces.

• Deliverables: buyer profiles, a question and objection inventory, a competitor search map.
• Why it matters: keyword tools show demand; buyers explain it. Pages written from real objections convert better and answer the questions people put to AI assistants. For B2B SEO especially, a handful of well-answered decision questions outweighs hundreds of generic posts.

2. Keyword research and search intent

We use Semrush per market database to measure volume, keyword difficulty (KD), cost per click and intent for each term, then add long-tail and question queries (“how much”, “vs”, “best for”) and the phrasings buyers type into ChatGPT or Perplexity. The goal is a list of searches that real buyers make and that you can realistically win. An excerpt from our own research, pulled on 1 October 2026:

Keyword	Database	Volume / month	KD	Decision
ai seo agency	US	3,600	28	Winnable: a dedicated US market page
ai seo agency	UK	1,600	22	Winnable: a UK page in British spelling
generative engine optimization	US	8,100	77	Hard: a service page plus a guide, a long game
agence seo offshore	FR	170	3	Small but easy: a French page for France-based buyers
agence seo casablanca	MA	170	n/a	Local: a French page for Morocco

• Deliverables: a keyword sheet per market with intent labels and priorities, plus an AI-prompt list.
• Why it matters: ranking for a term nobody buys from is vanity. Difficulty tells us where early wins are possible while harder terms mature.

3. Keyword-to-page map and information architecture

Every target query gets one URL, one H1 and one answer. We group pages into hubs and silos, check for cannibalisation (two pages chasing one intent), define breadcrumbs and slugs, and write internal-linking rules so every page is reachable and linked in context. On this site, the /seo/ map draws every page and every internal link as an interactive constellation, generated from the same content as the pages so it cannot go stale. On Panolux, the same kind of map reports 189 contextual internal links and 0 orphan pages.

• Deliverables: keyword → URL → template map, architecture diagram, internal-link rules, a redirect plan if you are migrating.
• Why it matters: competing pages split signals. A clear tree tells Google and language models which page is the answer.

Phase 4: the technical SEO foundation

Content cannot rank if it cannot be crawled, rendered and indexed. Our technical SEO pass covers:

• Crawlability and indexation: robots.txt, meta robots, stray noindex tags, crawl traps and orphan pages.
• Canonicals and duplicates: one canonical per page, with HTTP, www, trailing-slash and parameter variants consolidated by 301.
• Sitemaps: XML sitemaps generated from the page inventory with honest lastmod dates, plus image sitemaps where useful.
• hreflang: reciprocal language and region annotations with x-default.
• Status codes and redirects: no chains, no soft 404s, every legacy URL mapped.
• Rendering: static or server-rendered HTML rather than a client-rendered single-page app. GMAB’s old React SPA exposed one real URL; the rebuild exposes 67.
• Core Web Vitals: LCP within 2.5 s, INP within 200 ms and CLS within 0.1 for real users, mobile-first.
• HTTPS and accessibility: HTTPS with HSTS, semantic headings, alt text and WCAG 2.2 AA contrast.

Deliverables: an SEO audit with a severity and a fix for each issue, implemented in code where we have access, and a post-deploy check that every sitemap URL returns HTTP 200.

Why it matters: technical debt silently caps everything else. For new builds we default to static HTML on Astro, which any crawler can read on the first request.

Phases 5–6: content that answers, and structured data

5. Content that answers

Every key page opens with a two- or three-sentence answer, then the detail. We write for E-E-A-T (experience, expertise, authoritativeness and trust): named authors, real examples, sources, original data where you have it, comparison tables, and FAQs drawn from the question inventory in phase 1. Each page shows a visible “last updated” date. AI tools help us research and draft, but a senior human reviews and signs off every page. On GMAB, a post-build audit fails the build if any of about 45 previously invented phrases appear, so unsupported claims cannot slip back in.

• Deliverables: content briefs, written or edited pages, FAQ sets, a refresh calendar.
• Why it matters: search engines and AI answer engines both favour pages that answer precisely. Generic copy is neither ranked nor quoted.

6. Structured data and entity SEO

We model each page as one JSON-LD @graph: Organization, WebSite, Service, Product, Article, FAQPage and BreadcrumbList, linked by @id, with a consistent name, address and phone (NAP) and sameAs links to your real profiles. Schema is generated from the same data as the visible content, so the two cannot drift. Google limited FAQ rich results in 2023, but the markup still describes your questions and answers clearly. On Media Progetti, 66 of 69 built pages carry JSON-LD, with FAQPage on 59 and LocalBusiness on 12.

• Deliverables: a schema model, implementation, and validation in Google’s Rich Results Test and the Schema.org validator.
• Why it matters: entities remove ambiguity about who you are and what you sell, for Google and for language models.

Phase 7: local and international SEO, done honestly

Local SEO. A Google Business Profile belongs to a real location or a real service area, so we set one up or clean one up only where you genuinely serve customers. City pages earn their place only when each has different content. On shi we built 155 city pages across 9 Moroccan cities, and they are not a uniform grid: Casablanca and Rabat have 25 each, Khouribga 5, each city carrying only the services offered there. On GMAB, cities are modelled as a Service with areaServed, not as fake branch offices.

International SEO. Each market gets its own keyword research, because buyers phrase things differently: UK buyers search “bespoke software”, US buyers “custom software”. We set hreflang, local spelling, currency and language per market. Our own 30 international market pages, such as AI SEO agency for UK businesses, came from one research pass.

• Deliverables: a market and city page plan, an hreflang matrix, Business Profile setup where a real location exists.
• Why it matters: Google’s spam policies target doorway pages. Honest local pages last.

Phase 8: AI search and generative engine optimisation

A growing share of buyers ask an assistant before they search. We publish llms.txt and llms-full.txt generated from your site’s data, write facts in short citable units, and set a deliberate robots.txt policy for GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot and Google-Extended. That policy is a business decision: Origin Element, a London consultancy, chose to allow PerplexityBot and Google-Extended while disallowing GPTBot and ClaudeBot. We then track a fixed set of buyer prompts in ChatGPT, Perplexity, Gemini and Google AI Overviews and log which sources are cited.

• Deliverables: crawler policy, llms.txt files, a prompt list and a monthly citation log.
• Why it matters: a site these systems can read is easier to cite, though no one can guarantee a citation. Our generative engine optimisation service and the guide What is generative engine optimization? go deeper.

Phases 9–10: authority and measurement

9. Off-page SEO and digital PR

We earn links with assets worth citing: original data, useful guides and tools, partner and supplier pages, industry associations, and the directories your buyers actually browse. We also reclaim unlinked mentions. We do not buy links, use private blog networks or run link-exchange schemes; they break Google’s spam policies and can cost a domain its visibility.

• Deliverables: an outreach target list, linkable assets, a monthly log of earned links and mentions.
• Why it matters: authority still separates pages of similar quality. Shortcuts put the whole domain at risk.

10. Measurement and iteration

We connect Google Search Console and GA4, track rankings per market and log AI citations. Every month we report impressions, clicks, queries, conversions and indexation, then act: pages with impressions but low click-through get new titles, pages sitting just off page one get expanded, and decaying pages get refreshed with new facts and a new date.

• Deliverables: a dashboard, a monthly report and the next month’s plan.
• Why it matters: SEO compounds through iteration. The loop never stops.

SEO timeline: weeks 0 to 12 and beyond

Weeks	Phases	Main output
0–1	Kick-off and baseline	Search Console and GA4 access, Semrush baseline, full crawl
1–3	1–2: buyer and keyword research	Question inventory, keyword sheet per market
2–4	3–4: page map and technical audit	Keyword-to-URL map, prioritised audit
4–8	4–6: fixes, schema, first content	Technical fixes shipped, JSON-LD graph, priority pages built or rewritten
6–10	7–8: local, international and AI search	Market or city pages, hreflang, llms.txt, crawler policy
8–12	9–10: links and measurement	Outreach under way, dashboard, first monthly report
12+	The loop	Refreshes, new pages, earned links, monthly reporting

Projects kick off within 3–5 business days of signature. If the work includes a new marketing site, the build typically takes 3–6 weeks and the research and architecture phases run inside it.

What you get each month

Each month	Growth retainer	Scale retainer
Hours	45	100
Active projects	Two	Unlimited
Search Console, GA4, rank and AI-citation report	Yes	Yes
Technical monitoring and fixes in code	Yes	Yes
New or refreshed pages	Priority pages within the hours	Larger, multi-market content programmes
Schema, sitemap and llms.txt upkeep	Yes	Yes
Link-earning outreach	Focused on core topics	Broader, across markets
Syncs and response time	Twice weekly, 24 h	Daily, same day

Why RAIN: design, engineering and SEO in one team

Most SEO companies audit and recommend; most web agencies build and leave. RAIN does both. One senior team designs the pages, writes the code and owns the search results it can influence, using the same RDS Engine toolchain on every project. Proof from our case studies:

• Media Progetti: 68 French pages with 306 FAQ pairs in schema, and about 15 legacy URL rules consolidated into 301s. According to the client delivery statement dated 28 May 2026, a Semrush site audit of 78 crawled pages reported 100% site health and 100% AI search, with 0 errors. We have not re-measured it since.
• GMAB: a React SPA that Semrush found ranking for one unrelated keyword with no organic visits (22 September 2026) became 67 indexable pages with JSON-LD on every page. We have no post-launch data yet.
• shi: 215 pages, 187 of them generated from one SEO dataset, with fr-MA hreflang and a 212-URL sitemap.
• Panolux: a July 2026 crawl found 13 pages, no sitemap, no llms.txt and missing HSTS; the rebuild has 28 indexable pages and fixes for each.

None of these case studies claims rankings or traffic we cannot verify. French-speaking teams in Morocco can read our agence SEO au Maroc page.

SEO pricing and honest expectations

• Fixed-scope projects from $10,000: an SEO audit with implemented fixes, a migration, or a rebuild of a site search engines cannot read.
• Growth, $7,500 per month: 45 hours, two active projects, twice-weekly syncs, 24-hour response.
• Scale, $15,000 per month: 100 hours, unlimited active projects, daily syncs, same-day response.

Prices are in USD; we can invoice in GBP, EUR or AED on request. What to expect: technical fixes often show in Search Console within weeks of a recrawl, but competitive rankings build over months, and the curve compounds. We guarantee the work, the deliverables and honest reporting. We do not guarantee positions or AI citations, because no agency controls them.

Book a free 15-minute SEO call

Tell us your market, your buyers and the searches you want to win, and we will tell you what we would do first. Book a free 15-minute call: projects start at $10,000, retainers at $7,500 per month.