Case study  ·  observational tracking of public AI answers

Apple wins the answer.
The tech press wins the citation.

Eight months of weekly tracking, 676 stored analyses. In pure market questions — “best laptops” — apple.com is cited in under 1% of answers. The visibility is carried by the review ecosystem the engines trust.

76 → 89%
share of voice in AI answers
6 / 823
market answers citing apple.com
223 → 307
domains citing in the category
508
one-off sub-queries: the open tail
01

The climb.

Share of voice, monthly averages of weekly scans.

25507510081%92%
02

Market questions vs product questions.

Three brand-free prompts and one product-named prompt, measured separately. Even for Apple, the market questions are answered from third parties.

Tracked promptTypeStored answersOwn site citedRate
“Best iPhone to buy in 2026”Brand15213%
“What’s the best laptop for professionals in 2025?”Market28741%
“what are the best laptops for designers 2025?”Market28721%
“what are the best laptops?”Market24900%
Under 1% own-site citations in market questions — for the most famous brand on earth. The engines assemble “best laptop” answers from reviewers, not from apple.com. Nobody escapes the source game.
03

The citation ledger — before and after.

The raw accounting, market questions only, branded prompts excluded. Every number counted from the stored answers, not estimated.

Metric — market questions onlyBefore (Nov–Jan)After (Feb–Jun)Change
Stored answers analysed670686
Total citations inside those answers8,0428,075stable
Unique domains cited372530+42%
Citations of apple.com3 (0.04%)16 (0.2%)×5 (tiny base)
apple.com URLs appearing as sources35+2
Even Apple’s own domain is 0.2% of market citations. The graph is the game: 372 → 530 domains, and the review sites inside it decide the answer.
04

The market’s source graph.

The domains the AI cites for “best laptops” — the review ecosystem, plus retail.

SourceBeforeAfter
google.com106314
laptopmag.com326234
youtube.com227185
creativebloq.com144100
pcmag.com6993
rtings.com19071
bestbuy.com4051
reddit.com3747
wired.com1638
tomsguide.com2935
laptopmag, pcmag, rtings, creativebloq — this is who actually writes the AI’s answer. Being in their round-ups is worth more than any page on your own site. That pitch list is Get Cited’s job.
05

The cause is in the citations.

Same windows, every citation counted from the stored runs: 10,102 before, 21,121 after. No single engine moved — the sources writing the answer did.

SourceShare beforeShare afterChange
laptopmag.com10.9%6.8%-4.1 pts
techradar.com6.3%2.2%-4.1 pts
rtings.com7.5%4.2%-3.3 pts
theverge.com2.6%0.5%-2.1 pts
google.com7.8%17.1%+9.3 pts
forbes.com0.4%3.0%+2.6 pts
cnn.com0.1%1.0%+0.9 pts
Specialist reviewers lost share to aggregators and mainstream press — google.com alone doubled to 17%. A niche reviewer sometimes crowns a Dell; a mainstream “best laptops” list always carries the MacBook. When the source mix generalizes, the category default wins. apple.com stayed at 0.2% — the site didn’t earn this.
06

The proof: the only thing that removes the MacBook.

2,711 stored runs on the laptop questions; the brand appears in 99% of them — so the interesting signal is what correlates with the missing 1%. For every stored run we check two things in the same answer: which sources it cites, and whether the brand’s name appears in the answer text. Comparing runs where a source is present vs absent isolates each source’s effect. Correlation measured across runs — not a causal claim — sources with enough runs to matter only.

Source cited in the runRunsBrand in answer
when cited
When absentEffect
laptopmag.com · best-laptops lists (2 URLs)921100%98%+2 pts
creativebloq.com · laptops for designers638100%98%+2 pts
pcmag.com/picks/the-best-laptops329100%98%+2 pts
lenovo.com · business-laptops pages15984%99%-15 pts
dell.com · business laptops shop2080%99%-19 pts
pcworld.com · budget laptops list2090%99%-9 pts
Nothing in the editorial ecosystem removes the MacBook — every reviewer list carries it at 100%. The only sources that correlate with Apple missing from the answer are rival manufacturers’ own pages: when the engines cite lenovo.com or dell.com directly, Apple’s presence falls to 80–84%. Vendor content in the citation graph is a weapon — and Apple’s rivals are already using it.
07

The fan-out layer: where the gaps live.

Every prompt a user asks explodes into sub-queries the engines run silently. We track each one. Most appear once and vanish — nobody owns them yet. That long tail is the gap map.

677
sub-queries tracked
2
reach multi-engine consensus
508
appear once — the long tail
75%
of the battlefield is unclaimed
Sub-query the engines actually ranTimes seenEngine consensus
“best laptops for designers 2025”349×4 engines
“best laptops for design 2025”312×4 engines
“best laptops 2026 reviews”105×2 engines
“best laptops for creative professionals 2025”85×1 engines
Product-named sub-queries excluded. Two market sub-queries dominate with 4-engine consensus — and 508 appear once. Even in a category this mature, 75% of the battlefield is unclaimed.
08

The playbook this case proves.

The moveWhat it doesIn the suite
Get into the round-ups AI readslaptopmag-tier reviewers decide the market answersGet Cited + AI Links
Answer-first newsroomOwn press pages quotable enough to be citedContent Hub
Product pages with clean structureWin the product & navigational questions outrightEcommerce SEO
09

The schema autopsy: what the winners run under the hood.

We fetched the cited pages and extracted their structured data (JSON-LD), live. The pattern maps one-to-one to which questions each page wins.

Cited pageSchema foundWhat it wins
laptopmag.com/best-laptops-for-graphic-designersArticle + ItemList + BreadcrumbMarket questions — the round-up recipe (349× fan-out winner)
apple.com/macbook-proProduct + OrganizationProduct questions
gucci.com/store/725-fifth-avenueStore“Near me” questions (74+ citations)
gucci.com capsule & category pagesnone foundOnly win when the brand is already named
zara.com (any page)unreachable — 403 to all botsNothing: the engines cannot read the site
The winners are not a mystery. Article + ItemList wins markets. Product wins products. Store wins near-me. No schema wins nothing you didn’t already own — and a 403 wins nothing at all. Writing the first is Content Hub; deploying the rest is Action Center; catching the 403 is the LLM Readiness audit.

Method: live fetch of each cited URL with JSON-LD extraction (Aug 7, 2026). Gucci pages verified via Wayback snapshot — gucci.com blocks datacenter traffic. zara.com returned 403 to browser, GPTBot and PerplexityBot user-agents alike.

Run the same measurement on your brand.

Free scan, live in ~15 seconds. Your visibility, your gaps, the sources AI cites instead of you.

Scan my brand free →  Book a demo

Observational study of public AI answers, measured weekly by AI Rankia (Nov 2025 – Jun 2026). Per-engine presence rates; early-window samples are smaller. Brand names are trademarks of their respective owners, referenced for identification only; no affiliation or endorsement implied.