The Owned-Citation Gap: What Five AI Visibility Scans Reveal About Who AI Actually Recommends
ibs Fulcro's AI search analysts ran five independent Velivo Radar scans across five unrelated categories, cement, oral care, health drinks, paint, and whisky, tracking how often ChatGPT and related AI engines actually recommend real, named brands versus their competitors. The same structural pattern showed up in every single one.
Key Takeaways
- Rank and sentiment are rarely the real constraint. In four of five scans, the brand being measured ranked well or was viewed positively whenever it did appear in an AI answer. The actual constraint was whether it appeared at all.
- Owned-domain citation share tracks directly with AI recommendation coverage. The one brand in this set with double-digit owned-citation share (UltraTech Cement, 13%) also had by far the strongest coverage (38%) and the only #1 average ranking. The three brands at under 1% owned-citation share (Boost at 0.2%, JSW Paints, Imperial Blue) all sat at single-digit-to-mid-20s coverage, well behind their category leader.
- This is a publishing gap, not a product or brand-perception gap. In every scan, the category leader's own domain was cited repeatedly and converted those citations into brand mentions at 70-90%+ rates. The brands being measured had almost no citable content of their own for AI engines to draw from.
Why We Ran Five Scans Instead of One
A single AI visibility scan tells you about one brand in one category. It doesn't tell you whether a finding is specific to that brand's product quality, marketing spend, or reputation, or whether it reflects something structural about how AI engines build answers in general. ibs Fulcro's AI search analysts ran five independent scans using Velivo Radar across five categories with no overlap in buyer, product, or competitive set: cement, ayurvedic oral care, children's health drinks, decorative paint, and mass-market whisky.
Each scan tracked a real, named brand against its real, named competitors, running actual prompts a buyer might type or ask, against live AI engines, and recording exactly what came back: whether the brand was mentioned, where it ranked when mentioned, what tone the AI used, and which sources the AI cited to build that answer.
The Five Scans at a Glance
| Category | Brand Measured | Prompts Scanned | Coverage | Owned-Citation Share |
|---|---|---|---|---|
| Cement | UltraTech Cement | 50 | 38% | 13% |
| Health drinks | Boost | 30 | 53% | 0.2% |
| Whisky | Imperial Blue | 25 | 28% | 0% |
| Ayurvedic oral care | Dabur Red | 20 | 15% | 11% |
| Decorative paint | JSW Paints | 50 | 8% | 0.6% |
Coverage is the share of scanned prompts in which the brand appeared in the AI engine's answer at all. Owned-citation share is the share of all citations behind those answers that came from the brand's own domain. Source: Velivo Radar scan output, ibs Fulcro AI search analysts.
A methodology note before the findings: four of these five scans ran on ChatGPT only, since the other tracked engines returned no data for these particular prompt sets at scan time, and each ran one response per prompt rather than a repeated, trended measurement. Every figure below should be read as a real, specific point-in-time reading of one model's retrieval behaviour, not an audited, permanent fact about any brand. Where a scan's own findings noted this limitation directly, we've preserved that caveat rather than smoothing it away. These scans also report average position when mentioned rather than a full visibility score; see our research on why that metric alone can be misleading for the more rigorous alternative we now use.
Finding One: The Brand With the Most Owned Citations Also Had the Best Coverage
UltraTech Cement was the only brand in this set with double-digit owned-domain citation share (13%, or 24 of 186 total citations). It was also the only brand in the set with strong coverage: 38% of scanned prompts, an average position of 1.05 when it appeared (effectively always ranked first), and 68% positive sentiment. Its nearest rival, JK Cement, appeared in only 22% of prompts.
At the other end, Boost, JSW Paints, and Imperial Blue all recorded owned-citation shares at or below 1%, at or near zero in two of the three cases. Their coverage sat at 53%, 8%, and 28% respectively against category leaders sitting well ahead of them (Horlicks at 82% coverage for Boost's category; Asian Paints at 92% for JSW's; Royal Stag at 56% for Imperial Blue's).
Owned-Citation Share vs. Category Leader's Coverage
| Brand | Owned-Citation Share | Brand's Coverage | Category Leader | Leader's Coverage |
|---|---|---|---|---|
| UltraTech Cement | 13% | 38% | (itself) | 38% |
| Boost | 0.2% | 53% | Horlicks | 82% |
| Imperial Blue | 0% | 28% | Royal Stag | 56% |
| Dabur Red | 11% | 15% | Patanjali Dant Kanti | 40% |
| JSW Paints | 0.6% | 8% | Asian Paints | 92% |
Source: Velivo Radar scan output.
Finding Two: When the Brand Does Appear, It Usually Ranks Fine
This is the pattern that shows up most consistently, and it's the one worth taking most seriously. In four of the five scans, the tracked brand's own average ranking, when it did appear in an AI answer, was competitive or better than its rivals. UltraTech averaged position 1.05. Imperial Blue averaged 1.86, ahead of Blenders Pride's 2.08. Dabur Red averaged 1.67 and carried 100% positive sentiment across every mention. JSW Paints was the one exception, averaging 3.5 and carrying no positive mentions at all, but even there, the scan's own conclusion was explicit: “this is not a ranking problem, it is an absence problem.”
Put together, this means the brands in this set were not being penalized by AI engines for being worse products, cheaper, or less trusted. They were simply not being retrieved as an answer to begin with, for most of the questions buyers were asking.
Finding Three: Owned Pages Convert Citations Into Mentions at a Very High Rate, When They Exist
Across every scan that had any owned-domain citations to measure, the conversion rate from “cited” to “named as the answer” was consistently strong: 88% for UltraTech's own domains, 78% for Dabur's, and reported as directly correlated in every deck that measured it. The structural insight is not that AI engines are hostile to smaller or newer digital presences. It's that when a brand's own content is part of what the model retrieves, the model tends to use it. The problem for four of these five brands wasn't that their owned content performed poorly. It's that there was almost none of it in the retrieval set to begin with.
Finding Four: The Categories That Return Zero Are Usually the Highest-Value Ones
In both the UltraTech and JSW Paints scans, the categories returning zero brand mentions were not niche or low-value questions. UltraTech was entirely absent from waterproofing, putty, tile-fixing, crack-filler, and fast-track cement prompts, five real, commercially significant product categories where a different kind of brand (Asian Paints, Sika, Pidilite) had already established the citable content. JSW Paints was absent from interior paint, waterproofing, wood finish, metal enamel, and putty prompts, covering 33 of its 50 scanned prompts, including interior paint, the single largest spend category in its industry.
For Dabur Red and Imperial Blue, the pattern was similar but split by intent rather than product category: both brands performed reasonably on brand-discovery and shelf-availability questions, and returned zero on recommendation and trust questions specifically, the exact questions closest to an actual purchase decision.
Why this distinction matters: a brand that's absent from low-value, informational prompts has a minor visibility gap. A brand that's absent specifically from the highest-commercial-intent questions, or the specific product categories carrying the most revenue, has a gap sitting directly on top of its P&L.
What This Means for How AEO Should Actually Be Prioritised
The consistent, cross-category takeaway is that AI search visibility work should start with a genuinely unglamorous question: does the brand have indexable, citable content answering the specific questions buyers are asking, on its own domain, in a form an AI engine's retrieval system can actually find and quote? Every scan in this set found some version of the same root cause: a real product, a real market position, and almost nothing published to let a model cite it as the answer.
This reframes AEO priority away from a pure content-volume or SEO-ranking exercise and toward a more specific one: identify the highest-value, highest-intent questions in a category, check whether the brand's own domain currently has an indexable answer to each one, and treat every gap as a direct, addressable visibility loss rather than a general content backlog item.
Read Each Scan in Full
References
- All figures in this article are drawn directly from Velivo Radar scan output, an AEO and GEO tracking, analytics, and recommendation platform used by ibs Fulcro's AI search analysts. No client-supplied or confidential data was used in this research; all brands named are scanned as publicly observable entities in AI engine responses.
If you want our team to build and calibrate an AEO prompt universe for your enterprise, explore our AI Search Visibility (SEO / AEO / GEO) Services.