Photo by BoliviaInteligente on Unsplash
The Evidence
Zero citations. That is G2's and Capterra's combined score across 40 B2B software categories when ChatGPT is asked to recommend tools — not a thin sample, but 400 individual queries producing 233 recommendations across 219 distinct products. According to EIN Presswire's June 19, 2026 reporting on a DerivateX study, researchers queried ChatGPT ten times per category and logged every source the model credited when surfacing a software pick. G2 and Capterra — the two largest B2B software review platforms, which G2 consolidated under one roof via acquisition in January 2026 — did not appear once across all 40 categories examined.
The source breakdown is worth sitting with. As of June 2026, according to the DerivateX analysis, vendor self-published content accounted for 51% of ChatGPT's citations, small or anonymous websites for 23%, and traditional authoritative sources — analyst firms, review platforms, and business press combined — for just 16%. Separately, 41% of ChatGPT brand recommendations trace back to authoritative list mentions such as Forbes roundups, industry directories, and Wikipedia entries; 18% stem from awards or accreditations; and 16% from online reviews.
Perhaps the most counterintuitive data point: ChatGPT credits a recommended vendor's own website only 11.6% of the time. The remaining 88.4% of citations point to third-party sources — media articles, competitor comparison blogs, Reddit threads, and smaller web properties with no obvious domain authority signal.
A Conflict in the Data Worth Naming
Not every analyst reads the same numbers. SiteUp.ai's blog claims G2 is actually the fourth-most-cited source on ChatGPT and the only B2B software marketplace in the top ten — a finding that directly contradicts DerivateX's zero-citation result. The methodology divergence likely explains the gap: query phrasing, the number of categories sampled, and how citations are tracked all shape outcomes significantly in LLM (large language model) research. Neither finding should be treated as definitive.
My read: the DerivateX sample — 40 categories, 10 queries each, transparent methodology — is serious enough to act on directionally. But the broader question of what ChatGPT actually cites is early-stage and contested, and any vendor making major budget shifts based on a single study is moving faster than the evidence warrants. The consistent signal across both sources is that traditional review platforms are underperforming in AI recommendation contexts relative to their dominance in conventional search.
What It Means for How Buyers and Vendors Actually Operate
For roughly two decades, the B2B software evaluation stack followed a recognizable pattern: analyst report (Gartner, Forrester) to establish a credible shortlist, review platform (G2, Capterra, TrustRadius) to validate peer sentiment, then vendor demo. The analyst tier cost money and time to influence; so did the review tier. Both required sustained investment to move the needle.
If the DerivateX findings hold at scale, that stack has been partially bypassed for AI-mediated discovery. A buyer who asks ChatGPT "what's the best project management tool for a 15-person remote team" now receives an answer weighted toward vendor blog posts and smaller third-party writeups — content categories that have historically been considered lower-trust, not higher. DerivateX named this the "Authority Inversion": the trusted middle of software research has been hollowed out, with buyers often unable to see what replaced it.
This connects to a broader shift Gartner flagged in a February 2024 press release: the firm predicted traditional search engine volume would drop 25% by 2026 as users migrate queries to AI chatbots and virtual agents. As of June 19, 2026, according to Gartner's own timeline, that inflection point has arrived. Alan Antin, Vice President Analyst at Gartner, framed the content response plainly: "Companies will need to focus on producing unique content that is useful to customers and prospective customers. Content should continue to demonstrate search quality-rater elements such as expertise, experience, authoritativeness and trustworthiness."
One additional data point underscores the recency dimension: as of the DerivateX study, 71% of ChatGPT citations reference content published between 2023 and 2025. Traditional backlink authority and domain age — the bedrock of SEO (search engine optimization) strategy — do not appear to translate into how large language models evaluate and retrieve sources. Freshness does. That changes the calculus for productivity software vendors considerably.
Chart: Share of sources ChatGPT credited when making software recommendations across 40 B2B categories. *Traditional Authorities includes analyst firms, review platforms (G2, Capterra — both received zero direct citations), and business press combined. Source: DerivateX study, as reported by EIN Presswire, June 2026.
This AI-layer-as-primary-interface pattern extends well beyond software research — as AI Tools Newslens reported in its Android 17 Gemini AI coverage, AI is increasingly becoming the discovery and recommendation layer at the OS level too, reshaping how users find and evaluate tools before they ever visit a vendor website.
The Job You're Hiring Your Content Strategy to Do — And the Real Switching Cost
The job is specific: get your product named when a prospective buyer asks an AI chatbot what tool to use for your category. That is distinct from ranking on page one of Google, distinct from earning a G2 badge, and distinct from appearing in a Gartner Magic Quadrant. Each of those still matters for different parts of the buyer journey. But they do not automatically transfer to AI recommendation visibility — and that gap is the switching cost vendors need to understand before reallocating budget.
Years of G2 review accumulation, Gartner relationship investment, and traditional SEO infrastructure do not map onto LLM citation frequency. The content that does get cited — recent, third-party, specific, structured around clear use cases — is structurally different from what review platforms and analyst reports produce. Gartner formalized this in 2026 by publishing its first Market Guide for Answer Engine Visibility Tools, recognizing GEO (Generative Engine Optimization — the practice of structuring content so AI models retrieve and cite it in conversational responses) as a distinct discipline, separate from SEO.
For smaller SaaS vendors and remote teams building productivity software, this is arguably good news: the incumbent advantage that made G2 review count an expensive and slow-to-build moat is less relevant in a GEO world. A well-structured, recent, third-party article about your tool's specific use case may surface in ChatGPT recommendations while a competitor with 800 G2 reviews does not. The moat moved. The question is whether your content strategy moved with it.
How to Act on This
Run the query a real buyer would use — "what's the best [your category] tool for [your target buyer size and use case]" — ten times and log what ChatGPT returns. Note which products appear, which sources get cited, and whether your brand appears at all. This is free, takes twenty minutes, and gives you an actual baseline. Without it, any GEO investment is untargeted.
As of June 2026, 71% of ChatGPT citations reference content published between 2023 and 2025, and anonymous blogs account for 23% of citations — more than analyst firms and review platforms combined. Getting your product mentioned in recent, specific, well-sourced third-party content — comparison posts, niche roundups, community discussions — matters more for AI discovery than accumulating additional G2 reviews. This is not a reason to abandon review platforms entirely; it is a reason not to treat them as your only distribution lever for the AI-mediated buyer journey.
Multiple specialized GEO agencies and tools launched in 2026 specifically targeting AI recommendation visibility. Gartner's Market Guide signals the category is real — but also that it is early enough that methodology varies widely. Before engaging any vendor, ask: how do you measure citation frequency in actual AI responses, what query set do you use, and how often do you retest? "We improve your AI visibility" without specific measurement is a feature list masquerading as a capability. The demo is not the product.
Frequently Asked Questions
How does ChatGPT decide which software to recommend for a specific business need?
Based on the DerivateX study published in June 2026, ChatGPT appears to weight vendor self-published content heavily (51% of citations), followed by small or unknown third-party websites (23%), with traditional analyst firms and review platforms accounting for just 16% combined. Recency is also a significant factor: 71% of citations referenced content from 2023–2025. This suggests the model weights freshness and third-party mention frequency more heavily than traditional domain authority or review platform standing. However, the field is contested — SiteUp.ai's separate analysis reaches different conclusions — so treating any single study as definitive is premature.
Why isn't my SaaS product showing up in ChatGPT software recommendations despite strong G2 reviews?
G2 reviews optimize for traditional search and human-curated discovery, which operate differently from how LLMs retrieve and cite sources. As of June 2026, the DerivateX study found G2 and Capterra received zero citations across 40 B2B software categories in ChatGPT responses — though this finding is disputed by SiteUp.ai's research. What the available evidence suggests: LLMs appear to draw on recent third-party articles, comparison blog posts, and community mentions rather than review platform aggregations. If your content strategy has been primarily review-platform focused, your AI visibility may lag your traditional search visibility significantly.
What is Generative Engine Optimization (GEO) and how does it differ from traditional SEO for business tools?
GEO (Generative Engine Optimization) is the discipline of structuring content so that AI language models like ChatGPT, Perplexity, and Google AI Overviews retrieve and cite your brand in conversational responses. Traditional SEO targets how search engines rank web pages using signals like backlinks, domain authority, and keyword density. GEO targets how LLMs evaluate sources — with apparent emphasis on recency, specificity, and third-party credibility rather than backlink profiles. Gartner published its first Market Guide for Answer Engine Visibility Tools in 2026, formally recognizing GEO as a distinct discipline. Multiple specialized agencies launched that same year to serve the demand.
Are ChatGPT software recommendations reliable enough to inform actual purchasing decisions for small teams?
Industry analysts urge caution. The DerivateX study found that ChatGPT cites a recommended vendor's own website only 11.6% of the time, meaning the majority of sourcing comes from third-party content of widely varying quality — including anonymous blogs accounting for 23% of citations. This does not mean the recommendations are wrong, but it does mean the sourcing infrastructure is less rigorous than a peer-reviewed analyst report or a curated review platform. A reasonable approach: use AI chatbot recommendations as a discovery starting point to build your initial shortlist, then validate through free trials, reference calls, and pricing conversations before committing to any business tool or workflow automation platform.
Bottom line: The DerivateX study documents something the B2B productivity software industry has been sensing but not quantifying: established review platforms and analyst firms carry less weight in AI-mediated discovery than they do in traditional search, and the content types that actually drive LLM citations are structurally different from what those platforms produce. If you are a vendor whose entire discovery strategy runs through G2 badges and Gartner placements, that moat is narrowing for a meaningful slice of buyer journeys. If you are a buyer, treat AI software recommendations the way you would a well-read colleague's opinion — directionally useful, worth a follow-up conversation before you sign anything.
Disclaimer: This article is editorial commentary based on publicly reported facts and third-party research. It does not represent independent product testing. Tool features, pricing, citation behavior, and AI recommendation outputs may change. Always verify current details on official vendor websites. Research based on publicly available sources current as of June 19, 2026.