Does AI cite every brand it recommends?

An answer qualified when it ranked at least two brands. AI attached at least one cited page to every ranked brand in 37,986 of 227,755 qualifying answers, or 16.6784%. The other 189,769 answers left at least one ranked brand without an attached cited page.

This narrows existing citation-coverage research to the whole ranked shortlist. Semrush starts with domain appearances, while Ahrefs starts with brand mentions and asks whether the answer links to the brand's own website. This study asks whether every ranked option in one answer has an attached cited page. An attachment does not prove that the page supports the recommendation or mentions the brand.

Every ranked brand had an attached cited page in 37,986 answers, or 16.6784%. Some but not all ranked brands did in 122,927 answers, or 53.9733%. No ranked brand had one in 66,842 answers, or 29.3482%.

A single answer-level citation count hides two different gaps: partial coverage and no coverage for any ranked brand. Separate those states before comparing engines, prompts, or competitors.

ChatGPT Search attached no cited page to any ranked brand in 48,465 of 114,933 answers, or 42.1680%. Google AI Mode did so in 18,377 of 112,822 answers, or 16.2885%. The rate at which every ranked brand had a cited page was 15.2672% on ChatGPT Search and 18.1161% on Google AI Mode.

A citation-completeness audit should cover both engines. The combined result hides a large difference in answers where no ranked option has an attached cited page. This comparison describes the observed answers and does not identify why the engines differ.

AI cited every ranked brand in 9,894 of 31,176 two-brand answers, or 31.7360%. The rate was 22.6882% for three brands, 15.1855% for four to six, 6.2292% for seven to nine, and 3.4071% for 10 or more.

Compare citation completeness within similar shortlist lengths. A long recommendation list creates more opportunities for an uncited brand. This descriptive pattern does not show that list length caused the difference.

Among the 122,927 answers that cited some but not all ranked brands, 24,010, or 19.5319%, left five or more ranked brands without an attached cited page. Another 35,139 missed one brand, 27,614 missed two, and 36,164 missed three or four.

Partial coverage is not always a one-brand gap. Count the uncited ranked options before deciding whether an answer needs a focused source review or a full-shortlist audit.

Among industries with at least 1,000 qualifying answers, Sustainability reached 22.4652%, with 226 of 1,006 answers citing every ranked brand. Manufacturing reached 21.2594%, Professional Services 20.4561%, Software 14.5606%, Collaboration 13.3568%, and Apps 13.1990%.

Use an industry baseline before treating a completeness rate as unusual. The result shows where the observed rates differed. It does not show that industry caused the difference.

LinkedIn had an attached cited page in 2,998 of 6,025 ranked appearances, or 49.7593%. Among the ten most frequent brands, Datadog followed at 37.3101%, while GitHub had the lowest rate at 25.4601%. Alphabet had the most appearances, with 13,996 across 2,811 prompts and a 27.5150% citation rate.

Brand-level citation rates identify where to inspect repeated gaps. They do not show whether one answer cited every option, whether the page mentioned the brand, or whether the recommendation was accurate.

The consolidated brand-family method returned 16.6784%. Using exact brand identities without parent consolidation returned 16.2354%. Starting on June 1 returned 16.6807%. The two independent headline query shapes also returned the same 37,986 numerator and 227,755 denominator.

The study excludes 24,667 answers that ranked only one brand because citation completeness is automatic or absent rather than a shortlist comparison. Each cited page and root brand counts once per answer. The observed prompt set is not a random sample of all AI use, and the study makes no longitudinal movement claim.

AI cited every ranked brand in 37,986 of 227,755 multi-brand answers, or 16.6784%. ChatGPT Search cited no ranked brand in 42.1680% of qualifying answers, compared with 16.2885% on Google AI Mode. Complete coverage fell from 31.7360% for two-brand answers to 3.4071% for answers ranking 10 or more brands.

Track every ranked brand and its attached cited pages by prompt and engine. Separate complete, partial, and no-coverage answers. Compare answers with similar shortlist lengths. Inspect the cited page before treating an attachment as support. Rerun the same fixed-window method next quarter before calling a change movement.