Most conversations about AI search run on anecdotes. One brand noticed it got recommended by ChatGPT; another noticed it vanished from Perplexity. To see the shape of the thing, you have to measure it at scale.
So we did. The June 2026 Cited Index ran 226 Indian brands across ten categories through five AI engines (ChatGPT, Perplexity, Gemini, Google AI Overviews, and Google AI Mode) and captured every source the engines cited: 760 citations in total. This is the map of where AI actually gets its answers about Indian brands.
The headline number is a near-even split. Across everything, AI citations go 37% to brand-owned websites, 37% to marketplaces and aggregators, 23% to community and social sources, and just 3% to earned media. That average is tidy, and it is misleading, because it is the sum of two completely different games.
The map splits in two: D2C and B2B SaaS
Break the 760 citations down by category and the neat 37/37/23/3 falls apart. The source AI reaches for is decided, first, by what kind of business you are.
| Category | Citations | Brand site | Marketplace | Community | Earned |
|---|---|---|---|---|---|
| Audio & Wearables | 112 | 20% | 49% | 29% | 2% |
| Personal Grooming | 100 | 21% | 57% | 22% | 0% |
| Baby Care | 86 | 24% | 52% | 20% | 3% |
| Skincare & Beauty | 91 | 35% | 33% | 27% | 4% |
| Health & Wellness | 80 | 36% | 34% | 20% | 10% |
| Pet Care | 88 | 43% | 36% | 19% | 1% |
| Travel & Luggage | 101 | 42% | 33% | 26% | 0% |
| Conversational AI | 49 | 78% | 0% | 20% | 2% |
| CRM & Sales | 32 | 69% | 0% | 22% | 9% |
| HR & Payroll | 21 | 71% | 0% | 19% | 10% |
Read the marketplace column top to bottom. In the seven D2C categories it runs from 33% to 57%. In the three B2B SaaS categories it is zero. Not low. Zero.
In Indian D2C, when an AI engine answers a buying question, it is usually reading a marketplace. Amazon.in is the single most-cited domain in almost every consumer category, and in personal grooming, marketplaces and aggregators account for 57% of all citations. Your own website is the minority source, and often a small one.
In Indian B2B SaaS, the game inverts completely. Brand-owned sites are 69% to 78% of citations, and marketplaces contribute nothing. When AI answers a question about CRM or HR software, it reads the vendors' own documentation, comparison pages, and product content, plus a layer of community. The much-discussed review platforms, the G2s and Capterras, do not appear in the top cited domains for any of the three SaaS categories.
So the first question is not "how do I get cited by AI." It is "am I a D2C brand, in which case the citation lives on Amazon, or a B2B brand, in which case it lives on my own domain." The answer changes almost everything you would do next.
The one layer that shows up everywhere: community
There is a single exception to the D2C-versus-B2B divide, and it is worth its own line. Community and social sources are 19% to 29% of citations in every category, consumer and B2B alike. What sits inside that layer shifts, though: in the consumer categories it is led by Reddit, a top-three cited domain in nearly all of them, with YouTube close behind and Instagram alongside. In B2B SaaS, YouTube and LinkedIn carry it instead.
Whatever else is true about your category, a meaningful slice of what AI says about your brand is being lifted from community sources you do not own, a forum thread or a video or a LinkedIn post. That is the most consistent finding in the entire dataset, and the one most brands have no plan for. We have written before about why Reddit is the GEO play most Indian D2C brands are ignoring; this data shows the wider community layer matters across the board, not just in consumer categories.
The seat almost nobody has: earned media
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.
Now look at the last column. Earned media, the news and editorial coverage that public-relations budgets are built to produce, runs from 0% to 10%, and it is near zero across most of D2C. It is highest in the B2B categories and health, where trade press and research still carry some weight, and it is effectively absent everywhere else.
This is not an argument that PR is worthless. It is a narrow, specific finding: as a direct source of AI citations in these Indian categories, traditional earned media is the weakest of the four source types. The citation weight has moved to owned pages, marketplaces, and community, and the budget has mostly not followed.
And then there is the engine
Category decides the first cut. The engine decides the second. The same Indian brands, asked the same questions, get cited from different sources depending on which engine answers: ChatGPT leans on marketplaces, Google's AI Mode leans on community, and the two Google surfaces pull from opposite ends of the web. Stack the two layers together and "AI search" is not one surface to optimise. It is your category's source mix, shifted again by each of five engines.
What the map tells you to do
The instinct is to ask for the one tactic that wins AI citations. The data says there is no such tactic, because there is no single source. What there is, is a location, and it is decided by two things you can look up: your category and the engines your buyers use.
If you sell D2C, your marketplace presence is not a sales channel you happen to also have. It is a primary AI-citation surface, and it should be managed like one. If you sell B2B SaaS, your own site is doing the citation work, so documentation depth and comparison content are where the citations live, not a review-platform listing. And in both cases, the community layer, the forum threads and the videos, is carrying a fifth to nearly a third of your citations whether or not anyone in your company is watching it.
The map is not the same for everyone. That is the point. The first step is not a tactic. It is finding out which map you are on.
Methodology. 760 citations from the June 2026 Cited Index, across 226 Indian brands in 10 categories and five engines (ChatGPT, Perplexity, Gemini, Google AI Overviews, Google AI Mode). Each cited source domain is classified into one of five types (brand-owned site, marketplace/aggregator, community/social, earned media, other) and pooled market-wide by citation count. Because it is market-wide, it uses no owned-versus-competitor split. Category totals range from 21 citations (HR & Payroll) to 112 (Audio & Wearables); the smaller B2B categories are directional. This is a market-wide Index measurement and is distinct from a single brand's Citation Mix inside the Cited dashboard. Shares are summed counts, not averages of per-brand ratios.
Want to know how your own brand shows up in AI answers? Get a free AI Visibility Report across ChatGPT, Gemini, and Perplexity.