A study released today by Pranas, an AI search visibility platform headquartered in Singapore, quantifies what many marketers have suspected but could not measure: having a strong website is no longer sufficient to shape how AI systems describe your business to potential customers.
The research, published on September 21, 2026, analysed 20,839 unique prompts that generated Google AI Overviews during August 2026. After deduplication and normalisation, Pranas recorded 225,094 source citations across 13,996 root domains. The findings reveal a structural shift in how information flows from the web to AI-generated answers, and it has significant implications for any business that depends on being found online.
What the Study Found
The most striking finding is that AI systems do not build their answers primarily from company websites. Instead, they draw heavily from third-party platforms.
| Source | Citations | Rank |
|---|---|---|
| YouTube | 26,558 | 1 |
| 16,779 | 2 | |
| Wikipedia | 11,951 | 3 |
| Other third-party domains | ~170,000 | — |
Non-owned platforms accounted for the overwhelming majority of citations. YouTube alone received more than twice as many citations as any single company website could expect. This means that a brand could invest heavily in its own site, publish authoritative content, and optimize every page, yet still find that AI systems cite a competitor's YouTube review or a Wikipedia article when a potential customer asks about its product category.
Brand Mentions Do Not Always Lead to Brand Traffic
A separate Pranas analysis examined 10,227 travel-related prompts to understand the relationship between brand mentions and direct website attribution. The results varied dramatically by brand: between 63% and 82.5% of mentions were accompanied by a direct link to the brand's own website.
That gap is commercially significant. Two businesses could receive a similar number of mentions in AI-generated answers but experience very different outcomes. One brand gets a direct citation that gives users a clear pathway to its website. The other is mentioned using information attributed to a publisher, marketplace, or competing platform, meaning the AI surfaced the brand's name but routed the resulting traffic elsewhere.
This distinction creates two separate measurements that marketing teams need to track: brand visibility (whether the brand appears in an AI-generated answer) and traffic attribution (whether the answer cites or links directly to the brand's owned website).
Market Share Does Not Equal AI Visibility
Pranas also studied 10,000 smartphone and electronics-related prompts to compare AI visibility with established market position. In one finding, the second-most visible brand in AI-generated answers appeared in 21.1% of relevant prompts despite accounting for only a single-digit share of local device shipments.
Physical market share and AI visibility do not necessarily move together. Brands with extensive product documentation, comparison content, creator coverage, and third-party discussion may appear more frequently than larger competitors who sell more products but have a thinner information footprint.
This creates both an opportunity and a risk for small and midsize businesses. Companies that invest in building a stronger information presence across the platforms AI systems actually cite can gain disproportionate exposure relative to their market share. Conversely, market leaders who neglect their broader digital footprint could lose visibility to smaller competitors with better distributed evidence.
Why This Matters for SMBs
For small and midsize businesses, the Pranas study carries three practical lessons that change how marketing budgets should be allocated.
1. Your Website Is One Source Among Many
Traditional SEO treated the company website as the primary asset and everything else as supplementary. In the AI search era, your website is one source among many, and it may not be the most influential one. AI systems cross-reference what you say about yourself against what YouTube creators, Instagram reviewers, Wikipedia editors, and industry publications say about you. If the third-party evidence is thin, inconsistent, or absent, the AI system's confidence in your brand drops.
2. Mentions Without Links Are Worth Less Than You Think
If an AI system names your business but does not link to your website, the customer has no clear pathway to your door. This means you need to track not just whether you are mentioned, but whether you are cited with a direct link. A mention without attribution is a bill without payment.
3. Smaller Brands Can Win AI Visibility
The smartphone study proves that market share does not dictate AI visibility. A business with a strong presence on the right platforms can outperform a larger competitor in AI answers. For SMBs, this means the playing field is more level than traditional search ever allowed, but only for those who understand where to invest.
Three Priorities for Marketing Teams
Pranas recommends three priorities based on the findings, and we agree with each one from our own GEO practice at Vivid Visuals.
Measure citations alongside mentions. A mention confirms that a brand appears in an answer. A citation identifies the source the AI system relied on to build that answer. Tracking both reveals whether your visibility is supported by your own content, independent publishers, creator platforms, or competitors' domains.
Build authority beyond owned channels. Company websites remain important, but they represent one part of the information environment AI systems access. Brands need to strengthen their presence across trusted publications, creator platforms, industry resources, marketplaces, and other channels that AI systems regularly cite. Accurate and consistent third-party coverage helps AI systems understand a company's products, expertise, and market position.
Analyse visibility by market, language, and search intent. AI-generated answers vary between models, countries, languages, and prompts. A global visibility score can conceal weaknesses in individual markets or at critical stages of the customer journey. Brands should assess how they appear for different question types, including awareness, product comparison, evaluation, and purchase-intent prompts.
The Vivid Takeaway
Pranas Co-Founder Arshad Rahman framed the challenge precisely: "AI systems are becoming an intermediary between brands and buyers, but most brands are still unable to understand their visibility on AI assistants deeply or correctly map how to optimize it."
The study, with its sample size of over 20,000 prompts and 225,094 recorded citations, provides the strongest available evidence that the gap between digital presence and AI visibility is real, measurable, and consequential. For SMBs and mid-market companies, the takeaway is clear: investing only in your own website and conventional SEO is no longer a complete strategy. You need to understand which sources AI systems actually cite, where your brand appears in those sources, and where the gaps are.
Read next from the blog : Salesforce Koa and the Agentforce ROI Gap: What CRM Buyers Need to Know · WSO2 Agent Manager: Open-Source Governance for the AI Agent Sprawl Era · Meta's Stilla.ai Acquisition and WhatsApp Business MCP: The Automation Play for SMBs · Google Ads AI Video Dubbing: Why Multilingual Video Just Got Fast
