Optimising for AI features begins with the same durable foundation: crawlable pages, original content, clear entities, contextual links and a genuinely useful experience.

The short answer

No special markup or isolated technique can guarantee inclusion in an AI answer. Build pages that satisfy a distinct intent, demonstrate real experience and can be discovered, understood and verified by people and machines.

The next step is not choosing a tool. It is clarifying the decision, ownership and evidence that the team will accept.

Decision model

01. Technical foundation

Verify status codes, canonicals, sitemap, mobile behaviour, performance and accessible HTML.

02. Information architecture

Assign one primary intent to each URL and connect pillars through descriptive internal links.

03. Content quality

Show first-hand experience, responsible authorship, sources, limits and substantive updates.

04. Entity and measurement

Keep organisation data consistent, match structured data to visible content and monitor Search Console.

Applying the model

01. Starting context

Start with a technical and editorial inventory: status codes, canonicals, indexability, sitemap, hreflang, performance, pages, queries and internal links. Separate defects that block discovery from opportunities to improve relevance. Preserve URLs with useful history and consolidate overlapping intent before creating more pages. A practical audit produces a URL-to-intent map and prioritised causes, not a generic checklist detached from the site and its commercial goals.

02. Controlled execution

Build clusters around problems and entities the organisation can explain from experience. Pillar pages define the domain, commercial pages answer service intent and articles support distinct research decisions. Contextual links explain relationships to readers and crawlers. Structured data should describe visible content; it does not create expertise. Visibility in AI-assisted search still depends on accessible pages, clear entities, original evidence, reliable sourcing and content that can be understood and cited.

03. Useful evidence

Use Search Console to monitor indexing, queries, pages and position trends, then connect those signals with analytics and confirmed leads. Review clusters and page cohorts rather than only domain averages. AI-answer visibility is difficult to attribute completely, so observe referrals, mentions and commercial questions without inventing precision. No local file, schema type or submission mechanism guarantees ranking or citation, and reported progress must preserve that limitation.

04. Decision threshold

Prioritise by impact, confidence and dependencies. Repair discovery and indexing barriers, strengthen commercial pages and only then expand the editorial library. Change a modification date only after substantive review. If two URLs answer the same intent, consolidate them instead of optimising them against each other. Authority develops through consistent expertise, links, maintenance and off-site reputation, not through publishing the largest possible number of keyword variants.

Scenario and working plan

01. Diagnostic example

A site may create separate pages for digital marketing, online marketing, marketing agency and marketing services without defining different intent. The result is cannibalisation, repeated content and confusing internal linking. Keyword mapping groups demand, assigns one primary URL and decides which variations belong naturally on the same page. Commercial pages answer service intent, articles support research decisions and links explain relationships. Consolidation can create a stronger source than continuously expanding the number of near-duplicate URLs.

02. Implementation plan

A 90-day plan starts with status codes, canonicals, sitemap, performance and commercial pages. It continues with priority clusters, author accountability, sources, structured data and internal links. Publish only material that adds experience or a useful model, then use normal discovery mechanisms and monitor Search Console. Review movement into the Top 20, 10 and 3 alongside confirmed leads. For AI Search preserve clear entities, citations and crawler access, but never treat llms.txt or schema as shortcuts to guaranteed visibility.

03. Decision log

To make the recommendations in “Modern SEO: visibility in Google and AI-assisted search” traceable, open a simple decision log before the first change. Record the observed problem, baseline, hypothesis, owner, evaluation window and the condition for stopping or continuing. Evidence should come from sources suited to the topic, while technical indicators remain separate from commercial outcomes. The first measure reviewed is index coverage and technical errors, without treating it in isolation from data quality, total cost and downstream effects. This turns a favourable dashboard into an explainable decision rather than a conclusion based on intuition.

04. Review and next decision

At the end of the cycle, compare the result with the baseline and record what changed, what remains uncertain and which side effects appeared. Check explicitly whether “The page is indexable and self-canonical” and “H1, title and description are unique” are true. If the evidence cannot support a conclusion, keep the hypothesis open instead of declaring success. The risk “Creating a page for every keyword variation” stays visible during review so that pressure to show progress does not replace analysis. Choose the next step only when the team can explain what it learned and why the new priority matters more than the alternatives.

Pre-implementation checklist

  • The page is indexable and self-canonical.
  • H1, title and description are unique.
  • The content serves a distinct search intent.
  • Author and organisation are visible.
  • Structured data represents the page accurately.
  • Contextual internal links point to the page.
  • Relevant crawlers are not unintentionally blocked.

What to measure

Metrics are defined before launch and separate technical signals from confirmed business outcomes.

  • index coverage and technical errors;
  • queries and pages grouped by topic cluster;
  • click-through rate and progress into the top 20, 10 and 3;
  • confirmed leads and observable AI referrals.

Mistakes and limits

  • Creating a page for every keyword variation.
  • Scaling content without original value.
  • Adding schema for information users cannot see.
  • Treating llms.txt as a ranking factor.
  • Changing publication dates without substantive review.

Conclusion

SEO and AI-assisted search reward the same editorial and technical discipline. Create a clear, useful and maintained source, then measure outcomes without promising indexing, citations or rankings.