Search Engineering

AI search visibility, without the false promises

AI answers are becoming a distinct layer of discovery. We can improve the conditions that make a page citable and measure whether citing happens. We cannot make an AI system cite you, and any agency claiming otherwise is selling something it does not control.

The problem

A new discovery layer with a lot of confident advice and little evidence

AI answer engines choose their own sources. That makes citation influenceable but not controllable, and the current market is full of claims about a mechanism nobody outside those systems fully understands. The practical question is narrower and answerable: is your site in a position to be cited at all?

  • Competitors appear in AI answers for queries where you are absent
  • You cannot tell whether AI-sourced traffic exists for your category, or how much
  • Existing SEO reporting stops before the point where citations happen
  • Structured information about your organisation is inconsistent across the web
  • Your most authoritative content is not the content the machines can read most easily
Who this is for

The people who usually bring us this problem

Head of SEO / SEO Director

You have been asked for an AI visibility plan and want it grounded in what can be measured rather than what can be asserted.

CMO

You need to know whether AI search is materially affecting your category yet, and what to do that will still be correct regardless of how the answer engines evolve.

Founder

You have been pitched AI search optimisation and want a straight answer about what is real.

What it costs

What this costs while it goes unfixed

Engineering faults are rarely confined to the engineering layer. These are the commercial consequences we see most often.

Being absent from answers is invisible in current reporting

Standard analytics shows you the traffic you received. It does not show you the answers you were excluded from, which is the larger and growing number.

Acting on speculation is expensive

Work undertaken for an unproven mechanism consumes the budget that would have fixed the crawl, performance and content problems that matter under both old and new discovery models.

The prerequisites are the same ones you already need

Crawlability, clean structure, accurate entities and machine-readable facts are required for AI citation and are also required for conventional search. That is the reassuring part.

What we do about it

Capabilities

Each of these is work we carry out, not an area we advise on.

Entity and identity consistency

AI systems resolve organisations as entities. Inconsistent naming, conflicting descriptions and unlinked profiles make that resolution harder and weaken everything downstream. This is foundational and largely determinate.

Machine-readable factual structure

Clear, structured, unambiguous statements of fact — supported by visible page content, not invented in schema. This is one of the few areas where the mechanism is legible and the benefit is defensible.

Content legibility for extraction

Answers are assembled from passages that can be lifted and quoted. Content structured so its key claims are self-contained and extractable is easier to use as a source than content that buries them.

Citation measurement

Track which queries return answers that cite you, which pages are cited, and how your citation share compares with competitors' — using third-party tooling, reported as external evidence rather than as a product result.

Topic and entity gap analysis

Establish which topics in your domain produce answers you are absent from, and whether the absence is a content gap or a legibility gap. Those have different remedies.

Discovery prerequisite enforcement

The unglamorous part that actually determines outcomes: crawlable, fast, well-structured pages with accurate metadata. Without this, nothing else in this field works.

How we work

Engineering methodology

The sequence is deliberate. The order is usually what determines whether the work holds or has to be repeated.

  1. Separate established practice from experiment

    We are explicit about which part of the work is well-founded and which is a reasonable bet. Claiming certainty about how AI systems select sources would be dishonest, and it would also make the work unfalsifiable and therefore useless.

  2. Do the determinate work first

    Crawlability, structure, entity consistency and factual clarity. These are beneficial under every plausible future, so they are never a wasted investment even if the specifics of AI search change.

  3. Measure with third-party evidence

    Citation data comes from external tooling and is presented as such. We distinguish platform actions, conventional search outcomes and AI citation outcomes as three separate things, because conflating them is how the field ends up over-claiming.

  4. Run it as a pilot on content that deserves to be cited

    The productive approach is not to submit an entire site. It is to select the URLs whose content is genuinely authoritative, remove whatever is blocking discovery, and then measure whether citations follow.

Deliverables

What an engagement produces

Documentation is a deliverable, not an afterthought. On most of these engagements a large part of the value is a defect report precise enough for another team to act on.

Assessment

  • Entity and identity consistency audit
  • Machine-readability assessment of priority content
  • Citation baseline across a defined query set
  • Competitor citation-share comparison

Implementation

  • Structured data and factual markup for priority content
  • Content restructuring for passage extraction
  • Discovery prerequisite remediation
  • Topic and entity gap closure

Measurement

  • Citation and cited-page tracking
  • Query-set monitoring over time
  • Separated reporting of platform actions, search outcomes and AI outcomes
Under the hood

Architecture and technology

What we measure

  • Queries returning answers that cite the site
  • Cited pages and their topic distribution
  • Citation share against a defined competitor set
  • Grounding queries and the topics third-party tools associate with them

What we change

  • Organisation and content entity signals
  • Structured, machine-readable factual markup
  • Passage-level content structure
  • Crawl and indexation prerequisites
Adjacent problems

If this is not quite your problem

These overlap at the edges. Sending you to the right page is more useful than having you work it out.

AI systems are not discovering the pages at all

That is a crawl and indexation problem before it is a citation problem. See Technical SEO.

You want the measurement and the acceleration run as a programme

See the AI Crawl & Citation Acceleration solution.

Questions

Frequently asked

Can you guarantee we get cited by AI systems?

No, and nobody can. The answer engines choose their own sources and do not publish their selection criteria. What we can do is improve the conditions that make citation possible, and measure whether citing happens. Any provider guaranteeing citations is either describing something they do not control or measuring something that is not what you think it is.

Is this worth doing now, or should we wait?

The foundational work — crawlability, clean structure, consistent entities, machine-readable facts — is worth doing regardless of how AI search develops, because conventional search needs exactly the same things. The experimental work is worth doing as a bounded pilot rather than a programme. Waiting costs you the foundation; over-investing costs you the budget you need for it.

How do you measure something that is still changing?

Imperfectly, and we say so in the reporting. Citation data today comes from third-party tools with real limitations, including coverage gaps and volatility. We track direction and share rather than treating any single reading as precise, and we distinguish what we did from what changed as a result.

Does this overlap with your technical SEO work?

It shares a foundation. If the site is not crawlable and well-structured, AI search optimisation has nothing to work with. In practice most engagements start with the technical prerequisites and add the citation-specific work once the foundation is sound.

Bring us the problem you have not been able to fix

Describe what is happening rather than what you think the cause is. If we are not the right people for it, we will say so.