Article's Content
Article's Content
Brex ranks #1 on Google for “business credit cards for startups.” Their Spend Trends library has nearly doubled in traffic value to $536,000.
The AI Overview sitting above that #1 result draws on a Reddit thread and cites Nav instead. This happens because AI engines treat SERP ranking and being an authoritative source as two different qualifications, and Google’s #1 result doesn’t automatically meet the second one. Let’s look into the numbers.
The AI Overview reads its own trusted sources rather than rankings
The Foundation lab’s test makes this concrete. For “business credit cards” (54K searches per month), Brex ranks #2 and Ramp ranks #7, which is five spots lower. However, the AI Overview cited Ramp, Amex, Visa, Citi, Wells Fargo, and NerdWallet. Not Brex, despite outranking Ramp by five positions.
So we checked the query where Brex actually holds #1: “business credit cards for startups” and got the same outcome. The AI Overview led with Reddit consensus and cited Nav, naming Brex as one of several options, not the pick. Rank didn’t move the citation in either direction.
So that’s the mechanism. An AI Overview builds its answer from whichever sources it treats as authoritative for that question: comparison pages, aggregators, forums it trusts, and a brand’s own #1 page isn’t on that list just because Google’s ranking algorithm put it there.
Two entirely separate systems are making two separate judgments. Google evaluates traditional SEO signals: backlinks, on-page relevance, and site authority, while AI Overview relies on trustworthy sources for that topic.
Being named and being cited are different numbers
Brex’s 51.7% appearance rate is the second-highest in the category, behind Ramp at 60.6%. But brex.com supplied only 6.5% of citations, behind Nav (9.78%) and NerdWallet (8.63%). Take a look at Brex’s citation share across four major AI engines:
- ChatGPT: 13.8%
- Google Gemini: 10.0%
- Google AI Overviews: 8.1%
- Perplexity: 4.6%
Even on ChatGPT, Brex’s best engine, fewer than one in seven citations point to their own pages.
Competitors and earned media supply 46% of the source material
Nearly a quarter of the citations (24.6%) came from competitor sites. Another 21% came from earned media. Combined, competitors and earned media account for 45.6% of the content feeding these AI answers, compared with just 6.5% from Brex’s own pages.
Foundation’s Reddit specialist, Troi Leemuel Lamboon, put it simply: “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.”
That’s the gap Brex and other B2B brands need to close, providing real presence in the conversations the AI already trusts.
3 action items to improve your AI visibility
- Track citations and recommendations alongside rankings: A #1 ranking position tells nothing about which company the AI answer above it actually favors.
- Map the source gap: For each query where a competitor wins, identify which domains fed that answer, then find out whether those sources mention you.
- Earn a place in the cited sources: For example, in one of Foundation’s GEO engagements, a financial services client went from 12 of 100 priority Reddit threads cited by LLMs to 73, outranking competitors in 53 of them. Our work was getting into the cited sources.
Read the full study → for the full breakdown, including citation data comparison from different engines, and answer screenshots.
Ranking gets you seen. Sources get you chosen.
📊 Even more data, for you
35% of companies with fewer than 250 employees already use AI search for product research, compared with 47% who use Google, per Foundation’s buyer research. That gap is closing fast enough that a rankings-only view of your search performance is already missing a third of the room.
⚡ Quick hits
- Foundation’s broader research found that nearly 90% of AI citations for B2B SaaS come from sources outside the cited brand’s website. With this, Brex’s 6.5% owned-citation share is close to the category norm.
- Reddit outranks every major B2B vendor across keywords, representing roughly 957,000 monthly searches. If your competitor’s Reddit presence is stronger than yours, that shows up in the AI answer before your own site gets a chance to.
- Listicle citations in AI engine responses have dropped to 7.8%, down more than 50% since earlier this year, while pages structured to answer one specific question directly are holding their citation share, according to Peec AI data. Brex’s Spend Trends library is largely a listicle-and-roundup format, worth watching whether that citation erosion shows up in their numbers next quarter.
- B2B buying committees have grown from five to 16 members on average, according to Demand Gen Report’s 2026 B2B trends research. When 16 people are each running their own AI-assisted research, a citation gap can cost credibility among a much larger set of stakeholders evaluating the same purchase.
🔗 From the lab
- How Buyers Are Actually Using AI to Shortlist Vendors
- Recommended, Not Just Mentioned: What We Learned Running One Buying Prompt 11 Times
- The Reddit Cloud 100: Ranking B2B SaaS Subreddits by AI Visibility
- Why Nearly 90% of AI Citations Point Off-Site
Ranking #1 and still losing AI recommendation? Reply, we’ll take a look.
