News

Topic Visibility Monitoring Update

Kojable now helps teams understand AI visibility topic by topic, revealing where a company is well represented, where its evidence is weak, and where competitors or third-party sources may be shaping the answer instead.

A single visibility score can provide a useful overview, but it cannot explain why a company appears strongly for one question and disappears from another.

Topic Visibility Monitoring brings that detail into view.

  • Topic Monitoring
  • AI Visibility
  • 5 min read
  • News

What changed

See AI visibility at the topic level

Companies are rarely represented consistently across every product area, use case, category, audience, and buyer question.

A company may appear prominently when an AI system answers one question but be missing from a closely related answer. Its website may receive citations for one topic while competitors supply most of the evidence for another.

Kojable’s Topic Visibility Monitoring helps teams break portfolio-level visibility into the individual topics they are actively tracking.

For each topic, teams can understand:

  • How many AI responses were observed
  • How many responses were relevant to the intended topic
  • Whether the company was mentioned
  • Whether an owned page was cited
  • How citation activity was distributed
  • Which domains appeared most often
  • Which platforms and models contributed observations
  • How much evidence supports the result
  • When the topic was first and most recently observed

Answer: teams can see exactly where their AI representation is strong, weak, uncertain, or missing instead of treating visibility as one undifferentiated score.

Why it matters

Buyers do not research companies using one generic question

AI-mediated discovery happens through many different questions.

A buyer might ask:

  • Which tools solve a particular problem?
  • Which companies serve a specific industry?
  • What is the best approach for a particular use case?
  • How do two products compare?
  • Which vendor is credible for a certain requirement?
  • What alternatives exist?
  • Which company has evidence supporting a specific claim?

A company’s visibility can change significantly depending on how the question is framed.

Strong company-level visibility can hide weak topics

An overall score may be supported by a small number of high-performing topics while other strategically important areas remain invisible.

One weak topic can affect an important buying journey

A company may be well represented in broad category answers but absent from the specific use case or requirement that matters to a qualified buyer.

Competitors may dominate only certain conversations

A competitor does not need to lead everywhere to influence discovery. It may be especially visible for one high-value topic, audience, product capability, or comparison question.

Different platforms and models can produce different results

The same topic may generate different company mentions, cited sources, and answer structures depending on the platform or model involved.

Topic-level analysis helps teams see those differences rather than assuming that one observation represents the entire AI landscape.

Topic measurements

Understand what is happening within each tracked topic

Kojable brings several measurements together for every topic.

  1. Relevant responses

    The number of observed AI responses that meaningfully match the intended topic.

    This helps answer: How much usable evidence do we have for this topic?

  2. Topic Citation Share of Voice

    The share of citation activity within the topic that points to the company’s owned domains and pages.

    This helps answer: How much of the supporting evidence for this topic belongs to us?

  3. Topic citation coverage

    The proportion of relevant answers for the topic that include at least one citation to an owned company page.

    This helps answer: How often does our own evidence support answers about this topic?

  4. Topic answer mention rate

    The proportion of relevant answers that name the company, an approved brand name, or an owned domain.

    This helps answer: How consistently are we included when this topic is discussed?

  5. AI opportunity

    The estimated demand associated with the observed questions and answers for the topic.

    This helps answer: Which visibility gaps may matter most commercially?

  6. Distinct owned URLs

    The number of different company pages cited within the topic.

    This helps answer: Do we have broad supporting coverage, or does visibility depend on one page?

  7. Top cited domain

    The domain receiving the most citation activity for the topic.

    This helps answer: Who currently supplies the strongest body of evidence?

  8. Platform and model coverage

    The platforms and models represented in the available observations.

    This helps answer: Is the result consistent across AI systems, or concentrated in one source?

Relevant answers

Separate meaningful topic coverage from incidental mentions

Not every result containing a keyword is genuinely about the intended topic.

The same word or phrase can appear:

  • In another industry
  • As an incidental reference
  • As part of an unrelated company or product name
  • Inside a broader answer without meaningful relevance
  • In a question that lacks enough context to classify confidently

Kojable distinguishes between:

  • Accepted relevant responses
  • Ambiguous responses
  • Excluded responses

The platform preserves the underlying observations while calculating customer-facing topic metrics from responses that are sufficiently relevant.

Where an observation is excluded or treated as ambiguous, teams should be able to understand why.

Reasons may include:

  • Strong topic match
  • Weak topic match
  • Wrong industry
  • Ambiguous name or phrase
  • Incidental mention
  • Insufficient context

Answer: teams can understand whether a topic result is based on genuine market relevance rather than a noisy keyword match.

From topic to evidence

Investigate the questions behind the measurement

Topic-level visibility should not become another unexplained score.

For each tracked topic, teams can investigate the evidence behind the result, including:

  • Matched questions
  • Bounded answer excerpts
  • Company mentions
  • Owned pages receiving citations
  • Competitor pages receiving citations
  • Other influential domains
  • Source positions
  • Platform and model
  • First observed timestamp
  • Most recent timestamp
  • Relevant and excluded response counts
  • Collection diagnostics
  • Confidence state

This makes it possible to understand why a topic performs the way it does.

A team can move from:

“Our visibility is weak for this topic.”

to:

“We are mentioned in some answers, but competitors receive most of the citations and our relevant product page is not being used as evidence.”

That is a more useful starting point for action.

Platform and model differences

Understand where representation changes

AI systems do not always describe a company in the same way.

For a single topic:

  • One platform may mention the company frequently
  • Another may omit it
  • One model may cite an owned page
  • Another may rely primarily on publications
  • One answer set may include competitors
  • Another may describe the category without naming vendors
  • One system may have substantial available evidence
  • Another may have sparse or unsupported coverage

Kojable surfaces these differences so teams can understand whether visibility is broadly consistent or dependent on a particular platform or model.

The purpose is not to manufacture a single universal answer.

It is to reveal where representations differ and where those differences deserve investigation.

Confidence

Every topic result should show how much evidence supports it

Topic-level analysis can vary significantly in sample size.

Some topics may have many relevant observations. Others may have only a small number or no supported responses during the collection period.

Kojable distinguishes between confidence states such as:

  • Sufficient evidence
  • Directional evidence
  • Very sparse evidence
  • No indexed responses
  • No relevant responses
  • Unsupported coverage
  • Partial collection
  • Setup required

A sparse result should not be presented with the same certainty as a topic supported by a substantial sample.

Unsupported or unavailable data should not be displayed as 0% visibility, and an incomplete collection should not be treated as a complete measurement.

Percentages should remain connected to their underlying numerators, denominators, sample size, and collection status.

How teams can use it

Turn topic visibility into practical priorities

Topic monitoring helps teams decide where to investigate rather than simply producing another dashboard score.

Teams can use it to:

  1. Identify strategically important topics where the company is absent

    Focus on questions connected to important products, audiences, use cases, and buying journeys.

  2. Find topics where the company is mentioned but not cited

    Investigate whether owned pages clearly explain the subject and provide evidence AI systems can use.

  3. See where competitors supply most of the supporting evidence

    Compare the depth, clarity, specificity, and accessibility of competitor content.

  4. Find topics supported by too few company pages

    Reduce dependence on a single URL by improving the broader information architecture around the topic.

  5. Compare visibility across platforms and models

    Determine whether the company’s representation is consistent or concentrated.

  6. Review topics with weak or ambiguous relevance

    Refine the tracked topic or examine whether the phrase is being interpreted differently from the company’s intended market context.

  7. Prioritize using opportunity and evidence together

    A weak topic with limited demand may require less attention than a weak topic connected to substantial buyer interest.

What comes next

Monitor how individual topics change over time

Topic Visibility Monitoring establishes a detailed baseline for the questions and topics that matter to the company.

As comparable observations accumulate, teams will also be able to understand:

  • Which topics are gaining visibility
  • Which topics are losing visibility
  • Where new owned citations appear
  • Where owned citations are no longer observed
  • Where competitors are gaining supporting evidence
  • Where platform or model representation has shifted
  • Whether changes reflect performance or a change in scope or available data

Next step

Find the topics where AI representation needs attention

See where your company appears, where your pages are being cited, where competitors are stronger, and which topic gaps deserve deeper investigation.