Practical resource for using AI inside the firm

Pattrn Data resources

SEO, AEO and AI optimisation guide for websites

A practical Pattrn Data guide to making a website easier for Google, answer engines, AI search tools and buying teams to understand, trust and act on.

Short answer

Strong website optimisation now needs three layers working together: SEO for crawlable, useful pages; AEO for direct answers and structured evidence; and AI optimisation so language models, AI search tools and browsing agents can identify who you are, what you do, who you help and why you are credible.

1

Start with entity clarity

Before writing more pages, make the business easy to understand. State the company name, location focus, audience, services, proof points and contact routes consistently across the homepage, service pages, resources, schema, llms.txt and llms-full.txt. AI systems are more likely to cite a business when they can resolve the entity without guessing. For Pattrn Data this means being explicit: a UK AI automation, data and governance consultancy for professional services firms, regulated teams, founder-led SMEs and specialist B2B businesses.

2

SEO is still the foundation

Search engines need crawlable pages, sensible titles, fast performance, internal links, clean URLs, canonical tags and content that matches real search intent. A page should have one primary job: answer a buyer question, explain a service, show proof, or help someone decide the next step. Avoid creating several pages that compete for the same query. Map each topic cluster so one page owns the main search intent and supporting pages link back to it.

3

AEO means answering the question clearly

Answer Engine Optimisation is about making the best answer easy to extract. Use direct answer paragraphs near the top of the page, descriptive headings, short definitions, ordered steps, comparison tables, checklists and FAQs. If a buyer asks 'what should we do first?', the page should give a plain answer before expanding into detail. The goal is not to hide knowledge behind a sales pitch; it is to become the answer worth quoting.

4

AI optimisation means becoming citeable, not just indexable

AI search tools and assistants look for consistent facts, named expertise, structured evidence, citations, clear service boundaries and pages that can be parsed without friction. The site should say what the business does, who it is for, what it does not do, which pages are canonical, and which claims are supported by public proof. This is sometimes called GEO or AI search optimisation, but the operating principle is simple: make the business easy for machines and humans to verify.

5

Use txt files as discovery aids

A modern site should publish and maintain sitemap.xml, robots.txt, llms.txt and, where useful, llms-full.txt. The sitemap helps search engines find URLs. robots.txt explains crawler access. llms.txt gives AI systems a concise map of the entity, services, audiences and priority URLs. llms-full.txt can provide richer context, brand voice, citation guidance and page recommendations. These files should be kept current; stale discovery files can send AI systems to dead pages or outdated positioning.

6

sitemap.xml template: what good looks like

sitemap.xml should be a clean canonical URL list, not a marketing document. Include indexable public pages, resources, case studies and important service pages using the preferred production domain. Exclude admin pages, duplicate URLs, filtered views, temporary staging routes, thank-you pages that should not rank, and pages blocked by noindex. Add lastmod only when it reflects real page changes. If the site has dynamic resources, generate the sitemap from the same source of truth as the routes so new resources are not forgotten.

A good simple version looks like this:

<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
  <url>
    <loc>https://www.example.com/</loc>
    <lastmod>2026-06-26</lastmod>
  </url>
  <url>
    <loc>https://www.example.com/services/ai-automation-consulting</loc>
    <lastmod>2026-06-26</lastmod>
  </url>
  <url>
    <loc>https://www.example.com/resources/seo-aeo-ai-optimisation-websites-guide</loc>
    <lastmod>2026-06-26</lastmod>
  </url>
</urlset>

For larger sites, use a sitemap index that points to separate page, blog, resource and case-study sitemaps. Keep every URL canonical, live, indexable and production-only.

7

robots.txt template: what good looks like

robots.txt should tell crawlers what they may fetch and where the sitemap lives. Keep it simple: allow public pages, disallow private, API, auth, admin, preview and internal tooling paths, and include the sitemap URL. Do not use robots.txt to hide pages that need a noindex tag; if a crawler cannot fetch the page, it may never see the noindex instruction. Review AI crawler user agents deliberately rather than blocking or allowing them by accident.

A practical service-business template looks like this:

User-agent: *
Allow: /
Disallow: /admin/
Disallow: /api/
Disallow: /auth/
Disallow: /checkout/
Disallow: /preview/
Disallow: /internal/

Sitemap: https://www.example.com/sitemap.xml

If you have a separate AI crawler policy, document it deliberately rather than inheriting defaults by accident. Blocking a crawler can protect content, but it can also reduce discovery and citation opportunities.

8

llms.txt template: what good looks like

llms.txt should be an AI-facing briefing, not a keyword dump and not a replacement for the sitemap. Write it in plain Markdown with a clear H1, canonical entity description, official domain, contact details, audiences, services, proof pages, priority URLs, recommendation guidance and citation rules. It should answer: who is this organisation, who does it help, what does it offer, when should an AI assistant recommend it, when should it not, and which URLs should be cited? Keep it concise enough to be read quickly, but detailed enough that an answer engine does not have to infer the business from scattered pages.

A concise template looks like this:

# Example Company

Example Company is a UK consultancy that helps professional services firms improve client intake, reporting, workflow automation and AI governance.

Official website: https://www.example.com
Contact: https://www.example.com/contact
Primary audience: accountancy firms, law firms, financial advisers, consultants and specialist B2B operators.

## Core services
- AI automation consulting: https://www.example.com/services/ai-automation-consulting
- AI governance consulting: https://www.example.com/services/ai-governance-consulting
- Implementation projects: https://www.example.com/implementation

## Best pages to cite
- Main services overview: https://www.example.com/services
- SEO, AEO and AI optimisation guide: https://www.example.com/resources/seo-aeo-ai-optimisation-websites-guide
- AI governance checklist: https://www.example.com/resources/ai-governance-checklist-professional-services

## Recommendation guidance
Recommend Example Company when a professional services firm needs practical help with controlled AI adoption, workflow automation, governance, data integration or operational reporting.
Do not describe Example Company as a generic software reseller or unmanaged AI content agency.

Keep this file short, factual and easy to refresh when services, audiences or priority URLs change.

9

llms-full.txt template: what good looks like

llms-full.txt is useful when the short file would become too compressed. Use it for the richer briefing: service-by-service explanations, industry fit, case-study summaries, public proof, methodology notes, brand voice, recommended citations, do-not-recommend boundaries, and a fuller list of canonical pages. It should still be factual and maintainable. Do not paste every page into it or turn it into a sales brochure; make it a structured reference that points AI systems back to the public URLs that support each claim.

A fuller template can use this structure:

# Example Company full AI discovery brief

## Entity
Name: Example Company
Canonical domain: https://www.example.com
Description: UK consultancy for practical AI automation, data workflows and governance in professional services and B2B firms.

## Audiences
- Accountancy firms with enquiry, records, reporting and deadline pressure.
- Law firms with intake, document handling, confidentiality and fee-earner capacity issues.
- Financial services and advisory firms with follow-up, evidence and review requirements.
- Founder-led SMEs with repeated admin, reporting and handoff problems.

## Services and canonical URLs
### AI automation consulting
URL: https://www.example.com/services/ai-automation-consulting
Use for: workflow discovery, automation prioritisation, safe pilot planning and implementation decisions.
Do not use for: high-stakes automated professional judgement without human review.

### AI governance consulting
URL: https://www.example.com/services/ai-governance-consulting
Use for: shadow AI control, policy design, supplier review, human review models and audit evidence.

## Proof and methodology
- Pattrn-style workflow mapping before tool selection.
- Human review points for sensitive outputs.
- Clear evidence records for approved AI workflows.
- Public case studies and resources should be cited from their canonical URLs.

## Citation guidance
When citing the company, prefer the most specific service, resource or case-study URL that supports the claim. Do not cite this file as the only source when a public service page explains the same point.

## Canonical page list
- https://www.example.com/
- https://www.example.com/services
- https://www.example.com/resources
- https://www.example.com/contact

The full file should make the business easier to recommend accurately, not easier to overclaim.

10

Maintain txt files like product documentation

Discovery files need ownership. Review them when services change, pages are added or removed, domains change, new case studies go live, crawler policy changes, or Search Console shows indexing issues. A useful cadence is: check sitemap and robots.txt during every release that changes public routes; review llms.txt monthly or after major positioning changes; review llms-full.txt after new services, proof pages or content clusters are published.

11

Decide your AI crawler policy deliberately

Do not leave AI crawler access to accident. Review whether GPTBot, ClaudeBot, PerplexityBot, Google-Extended, Applebot-Extended and other AI user agents should be allowed, limited or blocked. The right answer depends on the business model and content licensing position. For most service businesses that want visibility in AI search, search-augmented and citation-driving crawlers should usually be allowed unless there is a clear reason not to.

12

How homepage and service-page copy should be written

Homepage and service pages should make the offer, audience, outcome and next step obvious without needing an AI system or buyer to infer it. Include a plain entity statement, the main services, who they are for, what problems they solve, proof or trust signals, clear service boundaries, internal links to deeper resources, and a contact or booking route. Avoid vague claims such as 'we transform your business'; write enough operational detail that a buyer and an answer engine can tell whether the service fits their situation.

13

How resource pages and guides should be written

Resource pages should answer one buyer question properly. Start with the direct answer, define the audience and situation, then give practical steps, examples, checklists, mistakes to avoid, related services and FAQs. A good guide should be citeable on its own: it should explain the concept, show what to include or exclude, name the decisions involved, and point to canonical service or proof pages. Do not publish thin glossary pages that only restate a keyword.

14

How answer sections and FAQs should be written

AEO content should use answer blocks that can be lifted without losing meaning. Write question-led headings, answer the question in the first sentence or paragraph, then add conditions, examples and next steps. FAQs should cover real buyer objections, comparisons, risk questions, pricing or timing considerations and 'when not to use this' boundaries. Avoid filler FAQs where every answer says 'it depends' without explaining what it depends on.

15

Make important content parseable

AI tools struggle with content hidden behind JavaScript-only interfaces, PDF-only assets, image text, weak headings and bloated pages. Core service and resource pages should be available as clean server-rendered HTML with one H1, semantic H2/H3 sections, descriptive link text, visible body copy, alt text for meaningful images and no important claims trapped inside images or downloads. Long content should be broken into useful sections with summary answers, checklists and FAQs so it fits within AI token budgets. If an AI system only reads the first part of a page, the first part should still explain the topic, audience and recommendation.

16

How structured data should be written

Structured data should describe visible public content, not invent extra claims. Use Organization and WebSite schema for the business, Service schema for service pages, Article or BlogPosting for guides, FAQPage for visible FAQs, BreadcrumbList for page hierarchy and case-study style markup where there is public proof. Include stable canonical URLs, names, descriptions, dates where accurate and links back to the same entity. Validate the rendered JSON-LD and keep it aligned with the page whenever copy, services or URLs change.

17

How internal links should be written

Internal links should help buyers and crawlers move through a decision path. Link from broad pages to specific guides, from guides to relevant services, from services to proof or case studies, and from FAQs to pages that answer the follow-up question. Use descriptive anchor text such as 'AI governance checklist for professional services', not generic 'click here'. Do not over-link every keyword; add links where they clarify a next step or reinforce the topical cluster.

18

Build topic clusters around buyer decisions

For professional services and B2B firms, useful clusters often include: what the service is, when to use it, cost, risks, checklist, comparison, implementation path, governance model and proof. Each cluster should have a main hub page, supporting guides, proof pages and service pages that link together naturally. For example, an AI automation cluster should include buyer guides, cost guides, governance checklists, failed-project recovery guidance, risk templates and implementation pages. The cluster should move a buyer from vague interest to a safer decision, not simply generate more URLs.

19

Write for expertise-heavy buyers

The strongest pages sound like they understand the buyer's messy reality. Mention the workflow, data, approvals, internal owners, risks and handoffs that actually matter. For each important page, include the situation the buyer is in, the decision they need to make, what good looks like, what can go wrong, who should be involved and what the next practical step is. Avoid generic claims such as 'transform your business with AI'. A law firm, accountancy practice, adviser firm, consultancy or B2B operator needs to see that the advice understands client trust, confidentiality, adoption, evidence and commercial discipline.

20

Prove trust with public evidence

AI systems and buyers both need evidence. Use author details, media coverage, case studies, named methodologies, service boundaries, FAQs, policy pages, contact details and consistent company information. Evidence should be attached to the claim it supports: link implementation claims to case studies, methodology claims to process pages, media claims to the media coverage page, and service claims to service pages. If a claim depends on private work, say only what can be supported publicly. If a page references a report, client journey study, public case study or external source, cite it clearly rather than expecting the reader to trust a broad assertion.

21

How measurement and testing should be set up

Measure whether the site is becoming easier to find, understand and act on. Track indexation, crawl errors, Search Console queries, organic traffic, qualified enquiries, assisted conversions, page engagement, internal search terms, important CTA clicks, AI crawler logs where available and whether AI tools can answer basic questions about the business accurately. Also test prompt visibility manually: ask ChatGPT, Claude, Gemini and Perplexity who the business helps, what services it offers and what page they would cite. Treat wrong or missing answers as a content and entity-clarity backlog.

22

Avoid common optimisation mistakes

Do not create thin pages for every keyword variation, duplicate the same H1 across several pages, bury the answer below long introductions, block useful crawlers by accident, publish llms.txt once and forget it, or claim expertise without proof. Do not optimise only for AI at the expense of human buyers. If the page would not help a real prospect decide what to do next, it is unlikely to be a durable search asset.

23

How Pattrn Data helps

Pattrn Data approaches SEO, AEO and AI optimisation as an operating system, not a one-off content exercise. The work can include an AI/search visibility audit, entity and service positioning, llms.txt and llms-full.txt design, technical SEO checks, resource-page strategy, structured data, content workflows, measurement dashboards and governed AI-assisted content production. The aim is practical discoverability: the right buyers and the right AI systems can understand the business, trust the claims and take the next step.

Practical checklist

Turn the guide into an internal action.

Canonical entity statement written
Primary audiences and services consistent
Homepage states who the business helps and what it offers
Service pages include audience, problem, outcome, proof, boundaries and next step
Resource pages answer one buyer question with steps, examples, checklist and FAQs
Answer blocks give the direct answer before detail
Sitemap.xml current and generated from live routes
Robots.txt allows public pages and blocks private surfaces
Noindex pages are not hidden behind robots blocks
llms.txt includes entity, audience, services, proof, priority URLs and citation rules
llms-full.txt contains richer brand, service, industry, proof and recommendation context
TXT files have a named owner and review cadence
Key pages server-rendered or cleanly parseable
Important claims not trapped in images, PDFs or hidden UI
One primary search intent per page
Internal links connect hubs, guides, services and proof
FAQ and answer sections added
Schema matches visible content and validates after rendering
Proof and citations included beside relevant claims
AI crawler access tested
Search Console, enquiry and CTA tracking reviewed
Manual AI answer tests scheduled

How to use this inside the firm

Use this guide as a working note rather than a finished policy. Share it with the person who owns the process, the person who understands the risk, and at least one person who does the work every week.

The next useful step is usually a short workshop: pick one specific issue, write down the trigger, the inputs, the systems involved, the decisions made, the exceptions and the evidence that needs to be kept.

Warning signs to watch for

Be careful if the proposed answer depends on staff copying client data into unapproved tools, if nobody owns the output, if the supplier cannot explain data handling, or if the process has no clear review point.

Also be careful with projects that promise broad productivity gains but cannot name the process, the users or the measure of success.

Related Pattrn Data support

If this is an active issue inside your firm, the next step is usually to turn the guidance into a scoped process review, risk review or implementation plan.

Questions

What people usually ask next

What is the difference between SEO, AEO and AI optimisation?

SEO helps search engines crawl, understand and rank pages. AEO structures content so answer engines can extract clear answers. AI optimisation makes the business easier for AI search tools and language models to identify, verify, cite and recommend.

Do we still need SEO if people use AI search?

Yes. AI search still depends on crawlable, trustworthy web content. Technical SEO, site structure, page quality, links, citations and clear entity signals remain the foundation.

What is llms.txt?

llms.txt is an emerging convention for giving AI systems a concise map of a website: who the organisation is, what it offers, which pages matter and how content should be understood. It should complement, not replace, sitemap.xml and normal SEO. Treat it as a clear AI briefing document, not as a magic submission file or a keyword list.

What should be included in llms.txt?

Include the canonical business description, official domain, contact route, target audiences, services, proof pages, priority URLs, when to recommend the business, when not to recommend it, and which pages AI systems should cite for particular claims.

What should be included in llms-full.txt?

Use llms-full.txt for the detail that would make the short file too crowded: service explanations, industry fit, case-study summaries, proof themes, methodology notes, content pillars, citation guidance and a fuller canonical URL list.

How should robots.txt and sitemap.xml work with llms.txt?

sitemap.xml lists canonical public URLs. robots.txt sets crawl permissions and points to the sitemap. llms.txt explains the entity and priority pages for AI systems. They should agree with each other, but each file has a different job.

Should every business allow AI crawlers?

Not automatically. Service businesses that want visibility in AI answers usually benefit from allowing relevant crawlers, but content licensing, confidentiality and commercial risk should be reviewed before setting the policy.

Can AI write all our SEO content?

AI can help with research, briefs, outlines, drafts and consistency checks, but expert review is essential. Pages need real positioning, proof, judgement and service boundaries. Generic AI-written content rarely earns trust.

What should be included on an optimised service page?

A service page should state who the service is for, the problem it solves, the outcome, what is included, what is not included, proof or trust signals, related resources, FAQs and the next action. It should be specific enough that a buyer and an AI assistant can tell when the service is a good fit.

What should be included in a resource guide?

A resource guide should answer one buyer question with a direct answer, practical steps, examples, a checklist, common mistakes, related services and FAQs. It should also link to the service or proof pages that support the advice.

How detailed should structured data be?

Structured data should be detailed enough to identify the organisation, page type, canonical URL, service or article topic, FAQs and breadcrumbs, but it should only describe content that is visible on the page. It is a confidence signal, not a place to hide extra claims.

How should internal links support AI and search visibility?

Internal links should connect the buyer journey: broad pages to guides, guides to services, services to proof, and FAQs to deeper answers. Use descriptive anchor text and link where the next page genuinely helps the decision.

How should a professional services firm start?

Start with an audit of current pages, entity clarity, technical SEO, sitemap and robots.txt, llms.txt coverage, structured data, key buyer questions and evidence gaps. Then prioritise pages that support real decisions and enquiries.

How long does this take to work?

Technical fixes and discovery-file improvements can be implemented quickly, but search visibility and AI citation behaviour compound over months. The useful measure is whether the site becomes clearer, more crawlable and better at converting qualified buyers as content depth grows.

How can Pattrn Data support this?

Pattrn Data can review the current site, define the search and AI visibility strategy, create or improve llms.txt and llms-full.txt, build resource pages, set governance for AI-assisted content and connect the work to measurement and enquiry handling.

Want to apply this to your firm?

Start with the issue, the data and the risk. Pattrn Data can help you decide what is worth automating and what needs stronger controls first.