Sourcelift, a generative engine optimization firm founded by a senior professional from a major Silicon Valley technology company, released its first AI Shelf Index on October 7, 2026. The index put 40 common US holiday gift questions to an AI assistant on October 5, 2026, and analyzed which websites the assistant cited in its answers. The results challenge a core assumption of digital marketing: that strong traditional search rankings translate into AI visibility.
What the AI Shelf Index Found
The index produced several findings that should reframe how brands think about AI search visibility:
- 72% of AI answers with sources (26 of 36) did not cite a single brand's own website.
- A third of the websites the AI assistant cited had been registered in the previous 12 months, according to public domain records.
- 66 distinct websites were cited across 40 questions, meaning AI answers draw from a narrow, rotating set of sources rather than a broad, stable index.
- In a separate test, a US software company on the first page of Google for its main keyword was named in only 2 of 10 answers from major AI assistants and AI search overviews.
- Answers in Spanish linked to a source half as often as answers to the same questions in English, a gap that matters for any brand selling in multiple languages.
Why Traditional SEO Does Not Guarantee AI Citations
The mechanism behind this gap is structural, not incidental. An AI assistant does not return a list of links. It writes one answer from a small set of sources and names the brands those sources support. Tony G., founder of Sourcelift, put it plainly: "A company can lead its category in search and still be absent from that answer."
Traditional SEO optimizes for a ranking system that evaluates pages against queries and returns ten blue links. AI answer engines operate differently. They retrieve information through query fan-out, synthesize a single response, and cite the sources that informed that response. A page can rank first organically and never enter the model's retrieval path for a given question. The AI Shelf Index quantifies this decoupling with real data: 72% of answers did not cite a brand website, and a US software company with first-page Google rankings appeared in only 20% of AI answers.
The finding that a third of cited websites were registered in the previous 12 months adds another dimension. AI assistants are not simply consulting an established hierarchy of authoritative domains. They are surfacing newer, less established sites when those sites answer the specific question more directly. This means the barrier to AI visibility is not necessarily domain age or backlink volume. It is whether the content answers the buying question in a way the model retrieves and cites.
The Broader Context: Citations Have Decoupled from Rankings
The Sourcelift index is not an isolated finding. Research across 2026 has consistently shown that AI citations have decoupled from organic search rankings. Ahrefs reported in March 2026 that only 37.9% of URLs cited in Google AI Overviews rank in the organic top 10, down from approximately 76% in July 2025. BrightEdge placed the overlap even lower, near 17%.
A survey of 45 GEO studies published in July 2026 reached a blunt conclusion: no reviewed technique shows a stable, longitudinal, cross-platform causal effect on AI discoverability. The on-page tactics promoted since 2023, including adding quotations, statistics, and fluency rewrites, have failed replication on modern engine families.
This does not mean SEO is dead. It means SEO alone is insufficient. Sourcelift's founder noted that "much of that work still matters," but the gap between search authority and AI visibility is real, measurable, and growing.
How Sourcelift Measures AI Visibility
Sourcelift's methodology treats AI visibility as something to measure rather than estimate. Each engagement begins with a locked set of 20 to 40 real buying questions for the brand's category, run across major AI assistants and AI search overviews and repeated every 30 days. Raw answers are delivered alongside the results so brands can see exactly what the AI said, not just a score.
Between measurements, Sourcelift works on four areas:
| Area | Focus |
|---|---|
| Brand identification | How clearly can AI engines identify the brand when answering category questions? |
| Site crawlability | Can AI crawlers read and parse the brand's site effectively? |
| Answer pages | Does the brand have pages that directly answer specific buying questions? |
| Third-party sources | Which sources do AI engines already trust in the category, and can the brand earn placement there? |
The firm does not promise a first position in any engine, because no agency controls what a model outputs. First citations typically appear between day 60 and day 90 of an engagement.
The Spanish-Language Gap
One finding from the AI Shelf Index deserves particular attention for brands operating in multilingual markets. Answers in Spanish linked to a source half as often as answers to the same questions in English. For a Montreal-based company serving both English and French Canadian markets, this pattern has a direct parallel. AI visibility in one language does not guarantee visibility in another. Brands that treat translated content as an afterthought are likely missing from AI answers in their secondary markets entirely.
What This Means for Your Brand
The AI Shelf Index reinforces what GEO research has been showing throughout 2026: AI visibility is a separate discipline from traditional SEO. Brands that invest only in keyword rankings, backlinks, and technical SEO are building half a strategy. The other half requires understanding which buying questions AI assistants are answering in your category, which sources they cite, and what it would take for your brand to become one of them.
The practical starting point is measurement. Before optimizing, a brand needs to know where it currently appears in AI answers, which competitors appear instead, and which questions it can realistically win. Sourcelift offers a complimentary AI visibility diagnosis that runs a starter set of buying questions across five major AI assistants and AI search overviews.
For brands that want a deeper assessment tied to their commercial goals, Vivid builds GEO strategies that start with this same principle: measure first, then optimize.
Read Next
- The 2026 AI Trust Report: 3.9 Million Citations Reveal Which Brands AI Actually Trusts
- The Great Citation Decoupling: What Q4 2026 Research Reveals About AI Search
- 79% Running, 2% Secured: SailPoint's 40x Identity Gap for AI Agents
- Shadow AI Agents: The Unregistered Threat Already Inside Your Workflows
- How QuickBooks Earned a 223% Citation Lift in AI Search: The GEO Playbook
