Ask an AI engine for the best luggage brand in India, and Samsonite comes up more than any other name. Across the five engines we test, it is the most visible brand in the category. Then ask a harder question: when a model recommends it, whose pages did it actually read? Samsonite's own website accounts for 0.53% of the citations in the category. It is the most recommended brand in Travel and Luggage, and its own website is barely in the citations, at the same time.
That split is the through-line of the August 2026 Cited Index. AI is already deciding which brands get recommended to Indian buyers, and it is not reading your website the way you think it is.
What the Cited Index measures
The Cited Index is a free, public benchmark of how Indian brands show up in AI-generated answers. We run a fixed library of non-branded, buyer-intent prompts, the things a real customer types, like "best CRM for a small sales team" or "best sunscreen for oily skin", across five AI engines: ChatGPT, Perplexity, Gemini, Google AI Overviews, and Google AI Mode. Then we measure two different things.
The first is AI Visibility: the percent of AI answers in which a brand gets named. That mention rate is one of the Cited 8, the metrics we track for AI visibility. The second is citation share: of all the structured citations an engine attaches to its answers in a category, what fraction resolve to the brand's own website.
These are not the same measurement, and treating them as one is the most expensive mistake a marketing team makes here. A brand can be named in nearly every answer and still have almost none of the underlying citations pointing to its own pages. Samsonite is exactly that: high AI Visibility, near-zero citation share. It is visible, not cited.
What's new in the August 2026 edition
This edition ranks 339 brands across 12 categories, drawn from 8,493 citations that the five engines attached to their answers. Two categories are new this time: Higher Education and Online Learning. The Index now runs on a bi-monthly cadence, so the next edition lands in two months.
One thing changed under the hood, and it matters if you followed earlier editions. We rebuilt how we read citations from the ground up. Earlier editions inferred sources from the prose of an answer, which over-counted brand names that happened to appear in a sentence and missed the actual links. This edition reads the exact structured citations the engines return with their answers. The practical rule: brand AI Visibility scores are comparable to earlier editions, citation figures are not. If you saw an older Cited number claiming a single marketplace drove a fifth of all citations, that was the old method, and it does not hold on the real data.
Finding 1: the most recommended brands' own sites are barely cited
Want to know how your brand scores on these same metrics?
We'll run 20 prompts across 3 AI platforms and send your report within 24 hours.
We call this borrowed visibility, and it shows up in 56 of the 112 brand-and-category angles we analysed. Being named on pages you do not own is the pattern in this Index, not the exception.
Samsonite is the clearest case. It is named in 51.4% of Travel and Luggage answers, the highest mention rate in the category, yet its own website is just 0.53% of the category's citations. When an engine names Samsonite, it is usually reading a listicle, a Reddit thread, or a rival's comparison page, not samsonite.com. American Tourister sits right behind it, second in the category, with its own site just 0.4% of the category's citations.
That visibility is real, and it is rented. It rests on pages the brand has no control over. If a publisher reworks a "best luggage" roundup, or a forum thread gets buried, the brand's presence in AI answers moves with it, and the brand never touched the lever.
There is an upside version of the same mechanic. Mokobara, a young Indian D2C luggage brand, is the biggest riser in Travel and Luggage this edition: its mention rate rose 11.4 percentage points. Because mention rates are comparable across editions, that gain is a real climb, not a measurement artefact. A challenger can close on incumbents faster in AI answers than it ever could on a page of Google results.
The read for a category leader: if most of your AI visibility is borrowed, your first job is not more owned content. It is knowing which third-party pages are carrying you, and making sure a challenger does not out-place you on them. We went deeper on that off-site reality here.
Finding 2: where AI reads changes completely by category
This is the finding that should reorganise how you spend. First, a quick disambiguation, because the phrase "citation share" is doing two different jobs. Finding 1 measured a single brand's own-site share: how much of a category's citations point to that one brand's domain. This finding measures the category's brand-owned share: how much of a category's citations point to any brand's own website, every brand in the category added together. Same citation pool, different cut.
Across the Index, that category brand-owned share runs from 90% at the top to 27% at the bottom.
| Category | Brand-owned citation share |
|---|---|
| HR & Payroll | 90% |
| CRM & Sales | 84% |
| Conversational AI Platforms | 84% |
| Pet Care | 62% |
| Online Learning | 59% |
| Health & Wellness | 51% |
| Higher Education | 45% |
| Baby Care | 43% |
| Skincare & Beauty | 42% |
| Personal Grooming | 37% |
| Travel & Luggage | 36% |
| Audio & Wearables | 27% |
Travel and Luggage sits near the bottom at 36%. That 36% is the combined figure for every brand's own domain in the category, spread across roughly 29 brands, which is exactly why even the category leader, Samsonite, lands at just 0.53% on its own. A thin, shared 36% divides into a lot of near-zero individual shares, and Finding 1 is what that looks like for one brand.
Read the two ends of the table together. In HR and Payroll, 90% of what AI cites is brand-owned. In Audio and Wearables, only 27% is; the rest is earned media, aggregators, and community threads. In business software, AI assembles its answer largely from what brands publish about themselves, so the work sits on your own pages. In consumer categories, it does not, and off-site placement decides the outcome.
The HR pattern is not the quirk of a single engine. Brand-owned citations there run between 88% and 98% across all five engines, spread across five different brand domains. It is a structural feature of how the category gets answered, not one model's preference.
One caution, because it is easy to over-read. This is a category-level pattern about where the work sits, not a law for any single brand. A specific HR software brand can still have a citation gap on its own site, and a specific D2C brand can still win with owned content. The Index tells you which playbook the category rewards; your own tracking tells you where you personally stand. We broke down the on-site versus off-site allocation call here.
Finding 3: in some categories, directories outrank every institution
New this edition, Higher Education shows the sharpest version of a pattern worth naming: aggregator gatekeeping. When students ask AI about colleges, the answer is assembled substantially from directory sites, not from the colleges themselves.
Shiksha is the single most-cited domain in the category at 6%. Education directories together account for 21% of all citations, more than any individual institution's own website. Even BITS Pilani, the most visible institution among those we measured, has almost none of its citations pointing to its own pages. The answer a student reads is built largely on third-party directories that sit between the institution and the reader.
Two honest caveats here. Higher Education is the hardest category to capture completely, and the reason is the names. Institutions go by long formal titles, acronyms, abbreviations, and everyday short forms, often several for the same college, which makes them easy to under-count. So read this as the institutions we measured, not a ranking of every college in India. And "barely cited" is the honest phrasing: it means few of an institution's own pages were resolved, not that it has no presence at all.
The takeaway travels well beyond education. In any category where directories or marketplaces sit between you and the buyer, getting listed and well-reviewed on those intermediaries can move your AI visibility more than another page on your own site.
Finding 4: in consumer categories, the source AI reads most is often a video or a social post
Look at what actually sits at the top of the cited-source list in consumer categories, and it is usually not a brand, a marketplace, or a magazine. It is a creator platform.
YouTube is the single most-cited domain in Travel and Luggage (10.4%) and in Audio and Wearables (11.4%). Instagram is the single most-cited domain in Skincare and Beauty (10.7%) and in Baby Care (7.6%). In each of those four categories, one creator platform is cited more than any individual brand's own website, and more than any marketplace.
The pattern runs across the consumer side of the Index. YouTube is also a top source in Personal Grooming (7.4%), Higher Education (5.1%), Skincare and Beauty (5%), Online Learning (4.1%), and Pet Care (3.8%). Instagram shows up strongly in Personal Grooming (7.2%) and Health and Wellness (5.2%).
Read this as a source fact, not a visibility score. It tells you where the engines go to read, not who they end up naming. But the two connect directly. If a brand's presence in the reviews, comparisons and unboxings on YouTube and Instagram is thin, then a large share of what AI reads in its category is being written by other people, often about other brands. Creator and social content used to be aimed at the human scrolling the feed. It is now also a primary source the engines read to decide who to recommend, which makes it the clearest form of borrowed visibility: your standing in AI answers rests on pages you neither own nor control.
Finding 5: the sources are more concentrated, and more engine-specific, than you think
Two facts sit under every category, and both cut against a common assumption.
First, concentration. In Pet Care, five domains account for 36% of everything AI cites. In HR and Payroll, five domains account for 26%. The list of places that matter for your category is short and targetable, which is good news once you know the five.
The name most people expect at the top of that list is not there. Across the entire Index, amazon.in is 3.07% of all citations. It is a source, not the source. AI answers about Indian brands are assembled from a wider and more specific set of pages than "it all comes from Amazon" would suggest.
Second, the engine matters as much as the category. Take the same Travel and Luggage prompts. Gemini pulls 67% of its citations from brand-owned pages, while Perplexity pulls 23% from them. Same category, same prompts, opposite source mixes. A brand tuned for one engine's habits is not automatically tuned for another's. We mapped that engine-by-engine divergence here.
One trap to sidestep when you read any share-of-citations number, including ours. Google AI Overviews lists about 10 sources per answer; Google AI Mode lists about 3. So a cell where one engine "supplies more citations" often just means that engine shows more slots per answer. A share of citations measures the format of an answer, not which engine matters most to your buyers.
What this means for your brand
Start with the diagnosis, using the 3-Layer AI Visibility Stack. Layer 1 is Discoverability: can AI crawlers reach your content. Layer 2 is Citability: when they reach it, do they cite it. Layer 3 is Authority: when they cite you, how well are you positioned against rivals. The Index is a Layer 2 and Layer 3 readout at market scale.
Three moves fall out of the four findings.
Separate your visibility from your citations. If you are named often but your own domain is rarely cited, you have borrowed visibility. Find the third-party pages carrying you before a challenger takes them.
Match the effort to your category's source mix. If you sell business software, your own pages are most of what AI reads, so make them answer buyer questions cleanly and completely. If you sell to consumers, the citations live off-site, so earned media, marketplace listings, and community presence do more per rupee than another blog post on your own domain.
Check yourself engine by engine. A single average hides the 40-point spreads that sit between engines. The fix for your Perplexity gap is often a different fix from your Gemini gap.
None of this shows up in your Google Analytics or your rank tracker. It is a different measurement, and it needs a different instrument.
Methodology
The August 2026 Cited Index is built from 314 non-branded, buyer-intent prompts run across ChatGPT, Perplexity, Gemini, Google AI Overviews, and Google AI Mode, producing 1,569 answers and 8,493 structured citations across 339 brands in 12 categories. AI Visibility is the percent of answers in which a brand is named, averaged across the five engines. Citation share is the fraction of a category's structured citations that resolve to the brand's own domain. Every category-level citation figure in this piece is drawn from between roughly 550 and 845 of those 8,493 citations, which is the range across the twelve categories. Citation figures are not comparable to editions before August 2026, which used a different extraction method; brand visibility scores remain comparable across editions.
Where to go from here
See your category in the live Index. It is free and public, with every brand, score, and source breakdown at getcited.in/cited-index. Travel and Luggage, HR and Payroll, Skincare, CRM, and eight more categories are all there, including the two new ones.
Then look at your own brand. Cited is an AI search optimization platform that tracks the same mention rate, position, sentiment, and source mix across ChatGPT, Perplexity, Gemini, and the two Google surfaces, applied to your prompts and your competitors. Start free at dashboard.getcited.in/signup and see where AI is reading you from.
If your brand shows up in AI answers today, whose pages is it reading to say your name?