Case Study: The Commercial AI Citation Blindspot

Answer Capsule

An analysis of AI citations in the Medical Alert Systems market found that the sources dominating broad AI conversations were substantially different from the sources appearing most often when AI systems were evaluating products for specific buyer situations.

We compared a broad Ahrefs Brand Radar dataset containing 26,773 AI responses and 151,601 citations with a commercially focused research dataset drawn from 15 Medical Alert Systems consensus studies and 4,534 observed citations.

The results showed:

• Only 2 of the top 10 most-cited domains overlapped
• The top 10 commercial citation sources captured 47.6% of citations, compared with 23.8% in the broader dataset
• Several specialized publishers increased their citation share by 10x, 20x, or more inside commercial recommendation prompts
• MobileHelp moved from #147 in the broad dataset to #9 in the commercially focused dataset
• YouTube fell from #1 to #49
• Reddit fell from #6 to #48
• Mayo Clinic and CDC disappeared from the commercial citation dataset entirely

The implication is important:

A website can have modest visibility across an industry broadly while occupying a disproportionately important position when AI systems are helping consumers decide what to buy.

This case study measures observed citation behavior. It does not prove that any cited source caused an AI system to make a recommendation.

The Problem: Broad AI Visibility Can Hide Commercial Citation Authority

Publishers and brands are increasingly monitoring how often they appear in ChatGPT, Google AI experiences, Perplexity, Gemini, Claude and other AI-generated answers.

But a basic citation count leaves out an important variable:

What was the user asking when the citation occurred?

Consider the difference between:

“What is a medical alert system?”

and:

“What is the best medical alert system for a senior who lives alone?”

Both belong to the same industry.

Commercially, however, they are very different questions.

One is informational.

The other places an AI system much closer to a product evaluation or purchasing decision.

Our research asked whether those two types of prompt environments produce the same citation ecosystem.

They did not.

The Study

We compared two separate AI citation datasets within the Medical Alert Systems market.

Dataset 1: Broad Medical Alert AI Citations

Ahrefs Brand Radar provided a large topic-level dataset containing:

26,773 AI responses

151,601 citations

16,881 unique domains

62,779 unique URLs

This represents a broad AI information environment surrounding medical alerts and related topics.

Dataset 2: Commercial Recommendation Citations

The comparison dataset came from 15 Medical Alert Systems consensus studies produced through Aging in Place Index.

Rather than attempting to measure every conversation surrounding medical alerts, these studies examined defined consumer situations and product-selection questions.

The dataset contained:

4,534 citation occurrences

264 unique domains

1,690 unique URLs

The studies covered buyer situations such as living alone, automatic fall detection, GPS and mobile use, affordability and other product-specific selection scenarios.

The two datasets are different in size and construction.

That is intentional.

The objective was not to determine which dataset was more comprehensive.

It was to ask:

Does the observable citation environment change when the prompt universe becomes more commercially focused?

Finding #1: Only 2 of the Top 10 Citation Domains Overlapped

The difference was dramatic.

Among the 10 domains cited most frequently in each dataset, only two appeared in both:

NCOA.org

and

Google.com

That produced:

Top-10 overlap: 2 of 10

Jaccard similarity: 0.11

Expanding the comparison produced only modest increases in overlap:

This means the websites dominating broad AI discussion of the industry were substantially different from those appearing most frequently within focused product-recommendation research.

Finding #2: Commercial AI Citations Were Far More Concentrated

The difference was not simply about which websites ranked.

It was also about how concentrated the citation market became.

The top 10 domains in the broad Ahrefs dataset accounted for:

23.8%

of observed citations.

In the commercially focused dataset, the top 10 domains accounted for:

47.6%

of observed citations.

That is nearly half of all citations concentrated among only 10 domains.

This raises an important hypothesis:

AI citation environments surrounding commercial recommendations may be materially narrower than the broader information environment surrounding an industry.

That hypothesis needs to be tested across additional industries before treating it as a general rule.

But if the pattern persists, its commercial significance could be substantial.

Finding #3: Specialized Sources Surged as Prompts Moved Toward Product Evaluation

Several publishers that appeared relatively minor in the broad citation dataset became leading sources in the commercially focused research.

Commercial Citation Rank Lift

The MobileHelp result is particularly illustrative.

Across the broad citation universe, MobileHelp represented approximately 0.06% of citations and ranked #147.

Inside the commercially focused dataset, it represented approximately 3.44% of citations and ranked #9.

That is a 58.6x difference in citation share.

A broad visibility report could therefore make MobileHelp appear relatively insignificant.

A commercial-intent citation analysis tells a very different story.

Finding #4: Broad Informational Sources Collapsed in Commercial Prompts

The reverse pattern was equally revealing.

Some of the largest citation sources in the broad dataset nearly disappeared once the research moved toward product selection.

BROAD AI CITATION ENVIRONMENT

YouTube — #1

Reddit — #6

Wikipedia — #11

NIH — #22

Mayo Clinic — #38

↓

COMMERCIALLY FOCUSED CITATION ENVIRONMENT

SeniorLiving.org — #1

NCOA — #2

SafeWise — #3

Bay Alarm Medical — #4

The Senior List — #6

The individual declines were substantial.

This does not mean these websites lack authority.

It indicates that general informational prominence and commercially focused citation prominence are different measurements.

The Prompt-Bleed Problem

The comparison uncovered another issue that can affect broad AI visibility measurement:

Prompt bleed.

A broad category such as “medical alerts” can expand into semantically related subjects that are not part of the commercial market a company is actually trying to understand.

In the Ahrefs dataset, highly ranked domains included wireless Lifeline subsidy and telecommunications properties such as:

• Tag Mobile
• TruConnect
• Assurance Wireless
• Verizon
• other Lifeline-related properties

Seven of Ahrefs' top 10 domains were absent from the focused commercial dataset.

Those citations may be completely legitimate responses to prompts associated with the broader semantic topic.

But they may have limited relevance to someone asking:

Which medical alert company is AI most likely to recommend?

This creates a measurement problem.

The larger the prompt universe becomes, the more possible it is for adjacent informational topics to influence a category-wide visibility score.

For companies making commercial decisions, prompt relevance can matter as much as dataset size.

Commercial Citation Lift

Commercial Citation Lift compares a source's share of citations in a commercially focused prompt environment against its share in a broad industry citation environment.

Formula

Commercial Citation Share ÷ Broad Citation Share

For example:

SafeWise:

5.36% ÷ 0.29% = approximately 18.3x

MobileHelp:

3.44% ÷ 0.06% = approximately 58.6x

A high Commercial Citation Lift does not prove that the source caused AI recommendations.

It identifies something more specific:

The source is disproportionately present within commercially focused AI citation environments compared with the broader topic.

That distinction can be extremely useful.

Why This Matters for Publishers

Imagine two publishers.

Publisher A

Receives enormous numbers of AI citations across broad informational questions.

Publisher B

Receives fewer citations overall but repeatedly appears when consumers ask AI systems:

• which provider to choose
• which product is best
• how competitors compare
• which service fits a particular buyer
• which company offers a specific feature

Which publisher owns the more commercially valuable AI citation footprint?

A global citation count cannot answer that question.

A prompt-conditioned citation index can begin to.

For publishers, this creates a new intelligence layer around:

Citation Share

How frequently is the publisher cited?

Commercial Citation Share

How frequently is it cited around buying-oriented questions?

Citation Lift

Does its prominence increase or decrease near commercial decisions?

URL-Level Authority

Which individual articles generate the citations?

Prompt Coverage

Which types of questions cause those articles to appear?

Platform Breadth

Which AI platforms cite them?

Brand Associations

Which companies are being discussed when the publisher is cited?

Why This Matters for Brands

Brands face the inverse problem.

Knowing that an AI platform recommends a competitor is useful.

But the next question is considerably more actionable:

What sources repeatedly appear within the information environment surrounding that recommendation?

Suppose a competitor consistently appears alongside citations from five specialized review publishers.

That does not prove those publishers caused the recommendation.

But it identifies an observable evidence environment worth investigating.

The brand can then ask:

• Are we included in those articles?
• Is our information accurate?
• How are we positioned against competitors?
• What independent evidence do those publishers rely upon?
• Which important commercial sources do not cover us at all?
• Are outdated claims appearing repeatedly?
• Which pages are cited across multiple AI systems?

This changes AI visibility analysis from simply measuring outputs to beginning to map the external information environment surrounding those outputs.

Why This Matters for Review and Editorial Publishers

The implications may be particularly important for review websites.

A publisher's value in AI search may not be adequately represented by traditional traffic metrics or broad AI mention counts.

A specialized article could receive relatively modest human search traffic while being repeatedly cited within high-value AI recommendation prompts.

That creates a new type of publisher intelligence:

Which parts of our editorial library occupy commercially important AI citation environments?

For publishers, that information could potentially inform:

• editorial investment
• content maintenance
• category strategy
• sales positioning
• sponsorship packaging
• research priorities
• competitive intelligence
• audience and advertiser conversations

The citation data itself does not determine how a publisher should monetize that position.

It provides the evidence needed to understand where that position exists.

The Strategic Blindspot in Broad AI Monitoring

Broad AI monitoring answers an important question:

How visible are we across the topic?

But commercial citation intelligence asks another:

How visible are we when the question matters economically?

Those should not be assumed to produce the same answer.

The Medical Alert Systems data demonstrates that they can produce radically different source rankings.

That distinction may become increasingly important as consumers move from traditional search results toward conversational product research.

From AI Visibility to Citation Intelligence

A commercially focused Citation Index can map the citation environment from the atomic observation upward.

Each observed citation can retain its relationship to:

AI Platform

→ Prompt or Study

→ AI Response

→ Company Being Evaluated

→ Citation Occurrence

→ Individual URL

→ Publisher / Domain

This makes it possible to answer questions that a simple mention counter cannot.

For example:

Publisher-Level Questions

Which of our domains and URLs are cited?

Which AI systems cite us?

Which product categories cite us most?

Which buyer questions produce our citations?

Which brands are associated with those citations?

Where are competing publishers cited instead?

Brand-Level Questions

Which publishers appear around our recommendations?

Which sources appear around competitor recommendations?

Which commercial prompt clusters contain us?

Where are we absent?

Which source relationships recur across multiple AI platforms?

What This Research Does Not Yet Claim

Citation frequency is not the same as source influence.

A citation attached to an AI answer does not automatically mean:

“The AI made this recommendation because of this source.”

AI systems may combine retrieved information, prior model knowledge and multiple sources.

For this reason, the current Citation Index measures:

Observed Source Usage

It does not yet claim to measure:

• causal influence
• sales impact
• recommendation causality
• source sentiment
• claim-level influence
• unique informational contribution

Those are separate questions requiring deeper source-to-answer analysis.

The distinction makes the methodology more useful, not less.

First determine:

What gets cited, where, how often and in connection with what?

Then investigate:

Which of those sources appear to materially affect the answer?

An Important Methodology Note

Because broad third-party monitoring datasets and proprietary research datasets may define individual citation observations differently, exact citation-share comparisons should be reconciled at equivalent counting units before being treated as standardized benchmark metrics.

The large differences in domain rankings and concentration are the primary descriptive findings.

Commercial Citation Lift should currently be treated as an analytical comparison rather than a finalized industry-standard score.

The Larger Hypothesis

The Medical Alert Systems analysis raises a much larger research question:

Do AI citation markets become more concentrated as prompts move closer to commercial decisions?

If similar patterns emerge in industries such as:

• insurance
• personal finance
• credit cards
• mortgages
• SaaS
• healthcare
• travel
• home services
• consumer technology

then the implications would be significant.

A brand might not need to influence the entire information ecosystem surrounding its industry.

It may need to understand a substantially smaller set of sources that repeatedly appear within the specific AI conversations closest to its customers' decisions.

That remains a hypothesis.

The next step is testing it across additional industries.

What a Commercial Citation Intelligence Pilot Can Measure

A focused pilot can begin with a single commercially important category.

For a publisher, brand or marketplace, the study can map:

• most frequently cited domains
• most frequently cited individual URLs
• citation share
• response appearances
• AI-platform coverage
• buyer-intent prompt coverage
• brand and company associations
• competing publisher visibility
• commercial citation lift
• citation concentration
• cross-platform overlap
• areas of strong citation presence
• citation gaps

The objective is not simply to generate a larger AI visibility report.

It is to identify:

where a company's content sits inside the information environment AI systems expose when consumers are evaluating commercial choices.

Key Takeaway

The Medical Alert Systems study revealed a clear citation blindspot.

Broad AI citation leaders were not necessarily commercial AI citation leaders.

Only 2 of the top 10 domains overlapped between the two datasets.

Commercially focused citations were almost twice as concentrated among the leading sources:

23.8% broad citation share

versus

47.6% commercial citation share

And several sources increased their citation prominence dramatically:

SafeWise — 18.3x

Caring — 16x

Lively — 31.2x

MobileHelp — 58.6x

At the same time:

YouTube fell from #1 to #49

Reddit fell from #6 to #48

and several major informational health authorities disappeared from the commercially focused citation dataset.

The strategic conclusion is not that broad AI visibility data is wrong.

It is that broad visibility and commercial citation visibility measure different things.

For companies attempting to understand their position in AI search, that distinction may be one of the most important measurements to make.

About LLM Authority Index

LLM Authority Index develops research and measurement systems for understanding AI recommendations, citation environments and brand authority.

Its research is designed to answer four increasingly important questions:

What do AI systems recommend?

Which sources do they cite?

Which sources repeatedly appear around commercially important questions?

How are brands positioned within those citation environments?

The objective is to move beyond simple AI mention tracking and begin mapping the information environments surrounding machine-generated recommendations.

Commercial Citation Intelligence Pilot

For publishers, marketplaces and brands with substantial editorial footprints, LLM Authority Index can begin with a single vertical rather than attempting to index an entire website.

The pilot is designed to answer:

Where are you being cited by AI systems?

Which of your pages are being cited?

Which commercial questions generate those citations?

Which competing sources dominate where you are absent?

And how does your citation footprint change when AI conversations move closer to a buying decision?

That is the difference between simply measuring AI visibility and understanding commercial AI citation intelligence.