The AI Visibility Paradox: Why High SEO Rankings Don’t Guarantee Recommendations in the Age of Generative Search

By conventional digital marketing metrics, Brex is winning. Its Spend Trends content library has nearly doubled in estimated traffic value to $536,000, and the corporate card and spend management platform comfortably holds the prized #1 organic search position for high-intent queries like “business credit cards for startups.”

Yet, step into the modern world of generative engine optimization (GEO), and a startling disconnect appears. When a potential corporate buyer asks an AI-powered search tool or engine—such as ChatGPT, Google AI Overviews, or Perplexity—for recommendations, Brex’s dominant SEO presence often evaporates. Instead of citing brex.com, AI platforms lean heavily on third-party aggregators, financial publishers, and community discussions on Reddit, frequently positioning competitors like Ramp as the default recommendation.

New research from strategy firm Foundation breaks down this exact phenomenon. Analyzing US market data across 348 distinct AI responses over eight major engines, the findings reveal a hard truth for modern brands: traditional search engine optimization (SEO) and generative engine optimization (GEO) are governed by entirely different ecosystems. Ranking well, appearing in AI answers, and actually being recommended by those answers are three distinct outcomes driven by three separate mechanisms.

Main Facts: The Great Disconnect Between Rankings and AI Citations

The core thesis of the study centers on a fundamental shift in buyer behavior. According to recent buyer research, roughly 35% of companies with fewer than 250 employees now utilize AI search engines for initial product research, creeping up on the 47% who rely on traditional Google search.

However, standard rank tracking reports are blind to this shift. A brand can proudly wave a #1 rank for a target keyword while an AI-generated overview sits directly above it, citing entirely different sources and favoring a completely different competitor.

Key statistical takeaways from the August 2026 data include:

• The Appearance vs. Citation Gap: Brex appeared in 51.7% of evaluated AI engine responses (second only to competitor Ramp at 60.6%). However, brex.com was explicitly cited as a source in a mere 6.5% of those total responses.
• The Aggregator Advantage: Financial aggregators and publishers consistently out-cited actual product vendors. Nav captured 9.78% of total citations and NerdWallet captured 8.63%, easily outpacing brex.com (6.51%) and ramp.com (3.73%).
• The Off-Site Source Weight: Competitor sites (24.6%) and earned media (21%) account for a staggering 45.6% of the citations fueling AI answers in this category, dwarf-sizing the influence of any single brand’s owned website.

Chronology: How Generative Engines Replaced the Traditional Search Funnel

To understand how Brex—and hundreds of other B2B SaaS companies—found themselves in this visibility paradox, it helps to look at how search technology has evolved over the past several years:

• The Traditional SEO Era (Pre-2023): Brands built massive content moats (such as Brex’s Spend Trends library, which expanded from 268 high-performing pages) to capture organic traffic directly. Winning a keyword meant owning the click, directing the user straight to a first-party landing page or blog post.
• The Rise of Zero-Click and Generative Overviews (2023–2025): Search engines began synthesizing answers directly on the search engine results page (SERP). Brands noticed organic click-through rates dropping as users consumed AI summaries rather than browsing traditional blue links.
• The Multi-Engine Generative Era (2026 and Beyond): Buyers increasingly bypass traditional search engines altogether, turning directly to conversational models like ChatGPT, Perplexity, and Gemini to curate vendor shortlists. These models do not build their answers by scanning the highest-ranking blog post; instead, they synthesize consensus data from trusted forums, review aggregators, and financial comparison sites.

This chronology explains why Brex’s traditional SEO gains—such as growing its Spend Trends traffic value from $280,000 in April 2025 to $536,000—failed to translate into guaranteed AI-driven recommendations.

Supporting Data: Engine Breakdown and Query Nuances

The data collected by Profound using US market inputs from August 2026 provides a granular look at how different platforms treat corporate card brands.

Engine-Level Citation Shares for Brex (brex.com)

• ChatGPT: 13.8%
• Google Gemini: 10.0%
• Google AI Overviews: 8.1%
• Perplexity: 4.6%
• Google AI Mode (Partial Data): 2.5%

Even on ChatGPT, where Brex performed best, fewer than one in seven citations pointed back to brex.com. Across every other major engine, the vast majority of underlying data came from external domains.

Query Phrasing Matters

The study revealed that Brex’s visibility fluctuates wildly depending on the specificity of the prompt:

1. Startup Queries: When prompts explicitly targeted startups, venture-backed businesses, or cards without personal guarantees, Brex’s appearance rate held firm. AI models recognized Brex’s strong niche authority.
2. Broad Category Prompts: When queries were generalized—such as simply asking for “business credit cards”—AI tools relegated Brex to an honorable mention while relying heavily on legacy issuers (Amex, Chase, Visa) and aggregators.
3. Founder Recommendation Prompts: When prompted with questions like "What business credit cards do founders recommend?" ChatGPT surfaced Reddit discussions as its primary source material, leading to Ramp, Amex, and Chase being framed as the dominant founder choices over Brex.

Official Responses and Expert Insights: The Value of Community and Context

Why do generative engines bypass slick corporate product pages in favor of forum comments and third-party listicles? Industry experts point to the unique nature of human dialogue versus marketing copy.

Troi Leemuel Lamboon, a Reddit specialist on Foundation’s strategy team, notes that AI models are ultimately programmed to give the most helpful, objective-sounding answer to a user’s prompt:

"A vendor is not going to say ‘don’t buy this, you’re too small for this.’ But Reddit will. The AI tool is just picking the result that actually answers the question, and sometimes it’s the comments."

Buyer discussions contain raw, unfiltered experiences that vendor landing pages omit—such as edge cases, implementation headaches, strict qualification criteria, and transparent comparisons of flaws. When an AI model evaluates which product to recommend to a small business owner, it values this third-party consensus over self-serving corporate claims.

Furthermore, even when a brand’s own comparison page successfully enters the AI’s source set—such as Perplexity citing brex.com/versus/ramp—the synthesized verdict may still favor the competitor. Being a cited source does not guarantee winning the recommendation.

Implications: The Strategic Pivot from SEO to GEO

For B2B marketing leaders, the lessons from Brex’s AI visibility audit are profound. Traditional SEO is no longer enough to secure a brand’s position at the top of the modern buyer’s consideration set.

To adapt to this new paradigm, organizations must implement a multi-layered visibility strategy:

1. Measure Citations and Recommendations, Not Just Rankings

Marketing dashboards must evolve. Alongside traditional rank tracking, teams need to monitor:

• Whether the brand appears in generative answers for core buyer queries.
• Which external domains are cited alongside or instead of the brand’s website.
• Which competing company ultimately receives the AI’s endorsement.

2. Map and Influence the Off-Site Ecosystem

Because nearly 90% of AI citations for B2B SaaS come from outside a brand’s owned domain, marketing budgets must expand to influence third-party touchpoints. This includes building positive presence and accurate positioning across:

• Aggregators and Review Sites: Ensuring platforms like Nav, NerdWallet, G2, and Capterra have up-to-date, favorable descriptions of your product.
• Earned Media and PR: Securing thought-leadership coverage in top-tier industry publications that LLMs treat as authoritative sources.
• Community Platforms: Engaging authentically within niche forums and discussions (such as Reddit or specialized Slack communities) where buyers seek peer validation.

3. Respect Community Platform Rules

While managing off-site sentiment is critical, experts warn against treating communities like traditional marketing channels. Platform moderation and community rules mean brands have far less control over public discourse. As Lamboon highlights, "If Reddit bans your account, your previous posts may stay live even though you can no longer manage them or respond to the conversation around them." The goal must be to genuinely earn trust where buyers already look, rather than trying to forcefully game the system.

Conclusion

Brex’s content engine remains an elite benchmark in traditional SEO, generating massive traffic value and capturing top spots on Google’s search results pages. However, the emergence of generative search has decoupled SEO performance from brand recommendation.

As buyers increasingly form their vendor shortlists inside AI answers, companies can no longer rely solely on their own websites to shape the narrative. To win in the age of GEO, brands must look beyond the traditional blue link, audit the off-site ecosystems feeding the AI, and ensure that when algorithms go looking for answers, the broader market is telling the right story.