nixal.ai

AI Citation Counts Hide the Commercial Interests Behind the Sources

By EmmaPublished Updated 10 min read

Most AI visibility reports stop at mentions, citations, and domains. That tells you where your company appeared — but not who publishes the sources buyers see, what commercial relationships sit behind them, or what action the evidence calls for.

In our review of 62 cited domains, the visible source layer included affiliates, retailers, competitors, manufacturers, experts, communities, and creator content — very different relationships presented through the same citation interface.

An AI answer connects to a single citation layer, which expands into six differently labelled source cards: expert, community, affiliate, retailer, competitor, and manufacturer. 'Third-party = independent' is crossed out; the takeaway is to check who shapes the answer.
The citation interface can make very different source relationships look alike.

What you'll learn#

This article helps you:

  • see why counting cited domains does not describe who is shaping AI product answers about a company;
  • distinguish observable ownership, retailer, affiliate, competitor, and manufacturer relationships in a visible source mix;
  • decide where the next investment belongs — in stronger owned information, customer proof, independent coverage, or corrective market presence.

Key findings#

Finding 1 — The visible third-party layer contained multiple, materially different publisher and commercial relationships. In this sample, ChatGPT and Perplexity cited expert and health organizations, communities, affiliate and commerce media, retailers, competitor brands, a manufacturer or supplier, and creator video. These categories describe observable ownership and disclosures. They do not establish accuracy, causal influence, or how any engine weighted a source internally.

Finding 2 — In the 16 advice-style sources, most had a visible product-commerce relationship or an unstable status. Of the 16 sources classified as advice-style content, 11 had confirmed affiliate disclosures, 2 were sellers or manufacturers publishing advice-style content, 1 had a site-wide affiliate program while the cited article did not disclose, 1 cited page had changed and was excluded from incentive claims, and 1 had no visible product-commerce incentive found. Advice-style presentation did not, in this sample, indicate observable independence.

Finding 3 — Commercial source types appeared more often in the selected solution- and brand-aware queries than in the selected problem-aware queries. In this query set, expert and community sources were more common on problem-aware questions, while affiliate media, retailers, and competitor and manufacturer pages appeared more often on solution- and brand-aware questions. This is an observation in the selected queries, not a general market rule.

Finding 4 — A citation can persist in an answer after the destination changes or disappears. One cited page redirected to unrelated content at review time, and one cited YouTube video was unavailable because the account had been terminated. Source decay is an observable risk in this dataset; it is not evidence about publisher incentives.

Finding 5 — Source ownership and incentives are a diagnostic layer. They describe whose commercial interests are visible in the answer. They do not determine accuracy or prove that any specific source caused a recommendation.

What this means for your company#

Knowing which sources appear in an AI answer is not the same as knowing who is shaping it. Different observable relationships imply different business responses.

  • Where the answer relies on unclear or weak official facts, the gap may call for stronger owned information — clearer product pages, specifications, safety and compatibility details, and evidence that reviewers or publishers can check.
  • Where the answer is built on vendor-, competitor-, or manufacturer-authored comparison content, credible customer proof and independent coverage may be needed alongside owned pages.
  • Where retailers and affiliate media dominate a shopping question, examine how those sources represent the product on the pages AI cites, so the company understands how it appears in shopping-oriented answers.
  • Where a citation depends on an unstable or repurposed page, source decay means a snapshot of "we are cited" should not be treated as permanent.

None of this proves that a change to any of these layers will produce a citation or recommendation gain. The point of a source-relationship review is to decide, question by question, which evidence layer is worth improving next.

Detailed observed results#

We opened all 62 domains and classified each by its own stated signals — an affiliate disclosure, a shop, an About page, a nonprofit mission, a stated business description. We used only verifiable evidence and flagged anything we could not confirm. The named companies below are used to describe observable publisher, ownership, affiliate, retailer, or competitor relationships; we are not inferring intent, dishonesty, or accuracy.

The visible source categories#

To an AI answer, these all appeared as citation links under a confident answer. To a reader, they represent different ownership, disclosure, and commercial relationships.

Source categoryExample cited domainObservable relationship
Expert / health organizationCornell Vet, iCatCare, VCA, PetMDeducational or health information; no product commerce visible on cited pages
CommunityReddituser discussion; no product commerce visible on cited threads
Affiliate / commerce mediaForbes Vetted, WIRED, CNN Underscored, cats.comaffiliate disclosures on cited pages
RetailerChewy, Best Buy, Walmartsells the product
Competitor brandPetSafe, PetKit, Catitsells its own product
Manufacturer / supplieraitakon.comdescribes a pet-products manufacturing business (B2B)
Creator videoYouTubemixed: sponsored/affiliate, brand-owned, charity, or unavailable

The categories are descriptive classifications, not judgments. Affiliate, retailer, competitor, and manufacturer content is not necessarily inaccurate; expert and community content is not automatically neutral. The point is that these very different relationships enter the answer through the same citation interface.

The 16 advice-style sources#

We isolated the sources that read as advice or comparison content — review media, roundups, "best of" and comparison blogs, seller or manufacturer advice blogs, and personal blogs — from the rest of the 62-domain dataset. Of those 16 advice-style sources:

  • 11 had confirmed affiliate disclosures on the cited page.
  • 2 were sellers or manufacturers publishing advice-style content: catboxy.com is a Shopify store whose About page describes a product line, and aitakon.com describes a B2B pet-products manufacturing business.
  • 1 (petscare.com) had a site-wide affiliate program, although the cited article itself did not disclose.
  • 1 (aptpaws.com) had a cited page that had changed at review time and was excluded from incentive claims.
  • 1 (pawsitive-purrpose.com) is an advice-style personal blog with no visible product-commerce incentive found.

In this sample, most advice-style sources had a visible product-commerce relationship or a changed status; one had no visible product-commerce incentive found. "No visible product-commerce incentive found" is not the same as complete independence — it describes what we could and could not verify from the source's own stated signals.

The incentive mix shifted with buyer stage in this sample#

The blend of source types changed with the buyer stage of the selected queries. This is a pattern observed in the selected query set, not a universal law.

Query stage (this set)Common source types
Problem-aware (why won't my cat drink, are fountains safe)expert and community sources, alongside some corporate and seller content
Solution-aware (best, quietest, easiest to clean)affiliate and commerce media, retailers, competitor and manufacturer pages
Brand-aware (review, vs, alternatives)brand-owned pages and affiliate review media

Two observations stood out. Problem-aware answers included fewer commerce-oriented sources but were not purely non-commercial: vet and health sources sat next to corporate pet-health content, a pet-service marketplace, and seller blogs. And within this query set, confirmed affiliate review sources appeared only on shopping-oriented queries, not on health questions.

A possible reader takeaway is that in the selected query set, closer-to-purchase questions carried more visible product-commerce relationships in the source layer. Whether this pattern holds across other categories, queries, engines, users, and dates was not tested here.

One pattern worth a careful note#

This was one run per query, so treat the following as an observation, not a measurement. The two engines leaned on different parts of the source mix in this sample. In the ChatGPT runs, the commerce-oriented layer skewed toward brand, competitor, and product pages. In the Perplexity runs, the third-party layer was broader — more review roundups, Reddit, YouTube, and small sites. Same citation interface, different observable source texture. In a separate experiment on a different category, we observed a related pattern in video specifically: the YouTube clips AI cited most in that test were low-view affiliate comparisons.

Source decay is a separate, smaller finding#

One cited URL no longer holds the article the AI answer quoted; it now redirects to an unrelated tool. Separately, one cited YouTube video was unavailable because the account had been terminated. Even where an AI answer cited a real source at the time of retrieval, that citation layer can change or disappear later.

Source decay is a source-integrity observation. It is separate from the commercial-incentive classification above and does not tell us anything about publisher intent.

How we checked this#

Nixal has no commercial relationship with PETLIBRO. This analysis was not commissioned, sponsored, or reviewed by the company. PETLIBRO was selected as a public, outside-in example, and the review is based solely on publicly accessible AI answers and cited sources. It does not evaluate PETLIBRO or its products.

We selected 15 cat-water-fountain queries across three buyer stages — problem-aware, solution-aware, and brand-aware — and ran each once on Perplexity and once on ChatGPT 5.5 in June 2026. That produced 62 unique cited domains, 138 answer-level domain appearances, and 281 URL mentions, after excluding image-only sources. We kept the 62-domain count, the 138 answer-level appearances, and the 281 URL mentions as distinct measures.

We then opened every domain and classified it using only verifiable signals: an affiliate disclosure on the cited page, a shop, an About page, a stated business description, or a nonprofit mission. Where a signal was unclear, we flagged rather than guessed. One cited page had changed at review time and is excluded from incentive claims; one cited YouTube video was unavailable and is treated as source-decay evidence rather than incentive evidence.

The full classified dataset (62 sources)#

Every domain the two engines cited, classified by observable ownership or commercial relationship from its own stated signals (affiliate disclosure, shop, About page, nonprofit mission, stated business description), sorted by how many answers cited it. Commercial interest badges: Sells = sells a product, Affiliate = discloses affiliate commission on the cited page, None found = no visible product-commerce incentive found. aptpaws.com is excluded from incentive claims (its cited page had changed at review time).

Show all 62 classified sources
DomainAnswersURLsEngineQuery layerSource roleCommercial interestNote
petlibro.com1140BothSolution, Brandowned brand (subject)SellsSubject brand's own store; 40 URL mentions in this sample
cats.com716BothSolution, Brandspecialist review mediaAffiliateAmazon/Chewy affiliate outbound on fountain page
reddit.com79PerplexityProblem, Solution, BrandcommunityNone foundUGC; no product commerce visible in cited threads
youtube.com611PerplexityProblem, Solution, Brandcreator videoVaries7 videos: 4 affiliate/sponsored, 1 brand channel, 1 charity (Woodgreen), 1 unavailable (account terminated)
petsafe.com517ChatGPTSolution, Brandcompetitor brandSellsCompetitor brand store
forbes.com511BothSolution, Brandreview media (commerce)AffiliateForbes Vetted; stated affiliate policy
wired.com58BothSolution, Brandreview media (commerce)AffiliateWIRED Reviews; commerce links + affiliate policy
chewy.com58ChatGPTSolution, BrandretailerSellsOnline pet retailer
vcahospitals.com57BothProblem, Solutionexpert authorityNone foundVet hospital network
bestbuy.com47BothSolution, BrandretailerSellsRetailer
petkit.com46ChatGPTSolution, Brandcompetitor brandSellsCompetitor brand
vet.cornell.edu46ChatGPTProblemexpert authorityNone foundUniversity vet school
catit.com45ChatGPTSolution, Brandcompetitor brandSellsCompetitor brand
icatcare.org44ChatGPTProblemexpert authorityNone foundCat welfare nonprofit
uahpet.com36PerplexityProblem, Solutioncompetitor brandSellsFountain brand publishing advice blog (brand-as-publisher)
oneisall.com36ChatGPTSolution, Brandcompetitor brandSellsCompetitor brand
walmart.com33ChatGPTSolution, BrandretailerSellsRetailer
vetstreet.com213PerplexityProblem, Brandexpert/health mediaNone foundVet/pet health media; no cat-fountain affiliate evidence on cited page
zoetispetcare.com25PerplexityProblemcorporate health / animal-health brand contentNone foundZoetis corporate/brand content; own products and rewards program; no cat-fountain affiliate evidence on cited page
cats.org.uk25BothProblemexpert authorityNone foundCats Protection charity
petsciencereview.com24PerplexitySolutionreview/roundup mediaAffiliateDedicated affiliate-disclosure page; earns commission via pet product recommendations
petmd.com24BothProblemexpert authorityNone foundVet health media
kittyspout.com23PerplexitySolutioncompetitor brandSellsBrand store; utm_source=Perplexity affiliate params observed
pawspik.com22ChatGPTSolution, Brandcompetitor brandSellsBrand
thesprucepets.com22ChatGPTBrandreview mediaAffiliateDotdash; affiliate roundups
catgearcanada.com19PerplexitySolutionreview/roundup mediaAffiliateAffiliate Disclaimer + Amazon.ca affiliate links / may earn commission
petscare.com18PerplexityProblempet service / marketplaceAffiliate?Site-wide Affiliate Program (creators earn commission); cited article itself lacked product affiliate disclosure; no named author
yourmultiversepet.store13PerplexitySolutioncompetitor brandSellsBrand store (.store)
petwant.com13PerplexitySolutioncompetitor brandSellsBrand
pawjoykwt.com13PerplexitySolutioncompetitor brandSellsBrand
petbarn.com.au13PerplexityProblemretailerSellsPet retailer blog
similarweb.com13PerplexityBrandother (data)n/aTraffic-comparison data tool
catster.com12PerplexitySolutionreview mediaAffiliatePet review media; affiliate roundups
aptpaws.com12PerplexitySolutionunverifiableExcludedCited URL now resolves to unrelated content; excluded from incentive claims (source-decay observation)
aitakon.com12PerplexitySolutionmanufacturer / supplier-owned contentSells (B2B)About page describes a pet-products manufacturing business (Shengzhou, China); cited page is a manufacturer guide; no affiliate links claimed
petcube.com12PerplexityProblemcompetitor brandSellsBrand
foothillpethospital.com12PerplexityProblemexpert authorityNone foundVet hospital
catinaflat.co.uk12PerplexityProblempet service marketplace blogNone foundCat-sitting service blog (marketplace-owned); no product-affiliate evidence in cited context
catboxy.com12PerplexityProblemseller-owned blogSellsShopify store; About page describes an own product line; brand blog publishing advice-style content
fetchyfriends.com12PerplexityBrandreview/roundup mediaAffiliateAffiliate disclaimer + Amazon deal links; no named author
cnn.com12PerplexityBrandcommerce mediaAffiliateCNN Underscored press note: earns from purchases via affiliate networks
aquapurr.com12ChatGPTSolutioncompetitor brandSellsBrand
apps.apple.com12ChatGPTSolutionapp storen/aApp rating signal
thirstycatfountains.com11PerplexitySolutioncompetitor brandSellsBrand (ceramic fountains)
smarthomeexplorer.com11PerplexitySolutionreview/roundup mediaAffiliateArticle contains affiliate links; Amazon Associates disclosure / may earn commission
goodhousekeeping.com11PerplexitySolutionreview media (commerce)AffiliateHearst; affiliate roundups
whisker.com11PerplexityProblemcompetitor brandSellsBrand (Litter-Robot maker)
rover.com11PerplexityProblempet service marketplace + blogAffiliateRover affiliate disclosure; also marketplace-owned
closerpets.co.uk11PerplexityProblemcompetitor brandSellsBrand
catlinkus.com11PerplexityProblemcompetitor brandSellsBrand (CatLink)
pawsitive-purrpose.com11PerplexityBrandpersonal/advocacy blogNone foundNamed author; links to Petlibro; no affiliate disclosure on cited page; one advice-style source with no visible product-commerce incentive found
support.petsafe.net11ChatGPTSolutioncompetitor brandSellsPetSafe support docs (owned)
store.enabot.com11ChatGPTSolutioncompetitor brandSellsBrand store (raw subdomain kept)
pioneerpet.com11ChatGPTSolutioncompetitor brandSellsBrand
cdc.gov11ChatGPTSolutionexpert authorityNone foundGovernment health
bissell.com11ChatGPTSolutionother (brand-adjacent blog)None foundCleaning brand blog on fountain safety; sells cleaners not fountains
esfi.org11ChatGPTProblemexpert authorityNone foundElectrical safety nonprofit
centerforpetsafety.org11ChatGPTProblemexpert authorityNone foundNonprofit
play.google.com11ChatGPTBrandapp storen/aApp rating signal
petsmart.com11ChatGPTBrandretailerSellsRetailer
petco.com11ChatGPTBrandretailerSellsRetailer
homerunpet.com11ChatGPTBrandcompetitor brandSellsBrand

FAQ

Are third-party AI citations necessarily independent?

Not in this sample. In this review, the visible third-party layer included expert organizations, communities, retailers, affiliates, competitors, a manufacturer, and creator content. "Third-party" is a reporting category we used to describe non-brand sources; it is not a platform interface label, and it does not describe one uniform commercial relationship.

Does commercial interest mean a source is inaccurate?

No. This review does not evaluate accuracy. Affiliate, retailer, vendor, competitor, and manufacturer content can be well-researched and useful. Understanding the observable commercial relationship helps a reader weigh what interests are represented — it does not label the source as wrong.

Why should a company examine publisher ownership?

Because different observable relationships imply different responses. A weak owned page, a vendor-shaped comparison layer, a retailer-dominated shopping surface, and an unstable citation each suggest different next steps — stronger owned facts, better customer proof and independent coverage, retailer and affiliate visibility work, or awareness that a citation snapshot can decay.

What can this review tell us about buyer-stage differences?

In the selected query set, expert and community sources appeared more often on problem-aware questions, while affiliate media, retailers, and competitor and manufacturer content appeared more often on solution- and brand-aware questions. Treat this as an observation in this query set, not a general market rule.

Can cited sources change or disappear?

Yes. In this dataset, one cited page had changed at review time and one cited video was unavailable. Companies should not treat a citation snapshot as permanent; the destination behind an AI citation can be edited, replaced, or removed.

AI Visibility MeasurementAI SourcesPerplexityChatGPTAI Search Insights

Keep reading