Small publishers lost 60% of search referrals. Google beat a publishers antitrust suit. For brands, the game is becoming a source inside the answer layer.
We ran the same set of buying queries across ChatGPT, Gemini, Perplexity, Claude, and Copilot to see which brands get recommended, how consistently, and what patterns emerge. The results tell you a lot about how AI search actually works.
Here's a question every brand should be asking: when a potential customer asks an AI tool to recommend a product in your category, does your name come up?
We decided to find out. We took one of the most competitive software categories in the world, customer relationship management, and ran a structured test across five major AI platforms to see which CRM brands get recommended, how the recommendations differ between platforms, and what patterns might explain why some brands dominate AI-generated answers while others are invisible.
This is not a review of CRM software. It's a study of how AI tools form and deliver product recommendations, using CRM as the test case.
The methodology
We developed five queries designed to simulate how a real buyer might ask an AI tool for help choosing a CRM. The queries range from broad to highly specific:
"What's the best CRM for a small business?"
"Recommend a CRM that integrates with Gmail and costs less than $50 per month"
"What CRM should a 10-person sales team use?"
"Compare the top CRM platforms for startups"
"I need a CRM for managing customer relationships at a growing e-commerce company"
We ran each query across five AI platforms: ChatGPT, Google Gemini, Perplexity, Claude, and Microsoft Copilot. Each query was run in a fresh session with no prior context. We recorded every brand mentioned, its position in the response (first mentioned, listed among several, or noted as an alternative), and the sentiment of how it was described.
That's 25 total query-platform combinations. Here's what we found.
The dominant brands
Three brands appeared with striking consistency across nearly every query and every platform: HubSpot, Salesforce, and Zoho CRM.
HubSpot was the most frequently recommended CRM overall. It appeared in the response to every single query on at least four of the five platforms, and was the first brand mentioned more often than any other. The AI tools consistently positioned HubSpot as the default recommendation for small businesses and startups, typically citing its free tier, ease of use, and scalability.
Salesforce appeared in nearly every response but was positioned differently depending on the query. For broad queries like "best CRM for a small business," it was mentioned but often with a qualifier about complexity or price. For queries about sales teams or growing companies, it moved to the top of recommendations. The AIs consistently treated Salesforce as the enterprise-grade option, powerful but potentially more than a small team needs.
Zoho CRM rounded out the top three, appearing consistently across all platforms as the value-oriented choice. It was especially prominent in responses to the price-constrained query about Gmail integration under $50 per month.
The second tier
A clear second tier of brands appeared frequently but less consistently: Pipedrive, Freshsales, Monday CRM, and Copper.
Pipedrive showed up most often in responses about sales teams, positioned as a pipeline-focused alternative to HubSpot. Freshsales appeared primarily in budget-conscious queries. Monday CRM surfaced in responses about small teams, typically described as a more visual or project-management-oriented option. Copper appeared almost exclusively in responses to the Gmail integration query, which makes sense given its positioning as a Google Workspace-native CRM.
The interesting finding here is how sensitive these second-tier recommendations are to query specificity. A broad query like "best CRM" rarely surfaced Copper. But the moment the query mentioned Gmail, Copper appeared on multiple platforms. This suggests that brands with clear, well-established positioning in a specific niche can punch above their weight in AI recommendations when the query matches their niche.
Platform-level differences
The five AI platforms did not produce identical results, and the differences reveal something about how each model forms its recommendations.
ChatGPT produced the most structured responses, typically listing four to six options with clear explanations of why each was recommended. Its recommendations skewed toward market leaders and well-established brands. ChatGPT was also the most likely to include a caveat like "the best choice depends on your specific needs" before listing options.
Gemini delivered shorter, more opinionated responses. It was the most likely to lead with a single strong recommendation rather than a list. In several cases, Gemini named HubSpot as its top pick and then briefly mentioned alternatives. This suggests that brands with dominant market positioning may benefit disproportionately from Gemini's more decisive response style.
Perplexity provided the most detailed responses with explicit source citations. Its recommendations closely tracked recent review articles and comparison pages from sites like G2, Capterra, and industry blogs. Perplexity was the only platform where we could trace exactly why a specific brand was recommended: because a specific source said so. This makes Perplexity the most transparent platform for understanding the mechanics of AI recommendations.
Claude produced nuanced, balanced responses that frequently included context about why different brands suit different situations. It was the least likely to name a single "best" option and the most likely to frame its answer around trade-offs. Claude also surfaced some less common recommendations that the other platforms didn't mention, suggesting it may draw from a somewhat different set of source signals.
Copilot delivered the most concise responses and was the most influenced by Microsoft ecosystem considerations. It was the only platform to consistently recommend Microsoft Dynamics 365, which barely appeared on the other four platforms. This is a clear example of platform bias shaping AI recommendations.
What determines who gets recommended
Looking across all 25 query-platform combinations, several patterns emerge about what drives AI recommendations in a competitive category.
Web presence breadth matters more than any single ranking. The brands that appeared most consistently across all platforms are also the brands with the most extensive web presence: thousands of reviews on G2 and Capterra, hundreds of comparison articles, extensive documentation, active community forums, and frequent mentions in industry publications. This corroborates what GEO practitioners describe as the corroboration principle: AI models recommend brands they encounter repeatedly across diverse, credible sources.
Positioning clarity drives niche recommendations. Copper's appearance specifically in Gmail-related queries demonstrates that clear, consistent positioning in a specific niche translates directly into AI recommendations for that niche. Brands that try to be everything to everyone may dominate broad queries but lose targeted ones to specialists.
Recent content influences Perplexity and Gemini more than ChatGPT and Claude. Perplexity's source citations showed a strong recency bias, frequently drawing from articles published in the last few months. Gemini, with its Google Search integration, showed a similar pattern. ChatGPT and Claude appeared to rely more heavily on accumulated authority over time. This suggests that content freshness matters differently depending on the platform.
Review volume and sentiment are a major signal. Every platform referenced review data in some form when explaining its recommendations. HubSpot's consistent top position tracks with its dominance on review platforms: it has more reviews on G2 than any other CRM, and the aggregate sentiment is overwhelmingly positive. For brands looking to improve their AI visibility, investing in review generation on major platforms appears to be one of the highest-leverage activities.
Structured data likely plays a supporting role. We couldn't directly verify this from the outputs, but the brands with the most complete and consistent structured data across their web properties were also the most consistently recommended. This correlation is worth noting even if we can't prove causation from this test alone.
What this means for brands
The CRM category is instructive because it's one of the most competitive and well-documented software markets in the world. The dynamics we observed here likely apply, in varying degrees, to every product and service category that consumers and businesses research through AI tools.
The practical takeaways are clear. If you're absent from AI recommendations in your category, the solution is not to optimize a single page or run a single campaign. It's to build the kind of broad, consistent, well-corroborated web presence that makes AI models confident enough to recommend you.
That means getting reviewed on the platforms that AI tools cite. It means appearing in the comparison articles and industry publications that serve as source material. It means maintaining consistent, accurate information about your brand across every platform where you have a presence. And it means creating content that directly addresses the kinds of conversational queries people are asking AI tools.
The brands that dominate AI recommendations in 2026 didn't optimize for AI. They built the kind of comprehensive, credible web presence that AI tools naturally draw from. That's the real lesson.
A note on methodology
This test has limitations worth acknowledging. AI responses are non-deterministic, meaning the same query can produce slightly different results each time it's run. We ran each query once per platform in a clean session, which provides a snapshot but not a statistically significant sample. AI models also update their knowledge bases and retrieval systems regularly, so results may shift over time.
We plan to repeat this test quarterly across different product categories to track how AI recommendations evolve. If you'd like to see a specific category tested, let us know.
James Calder is the editor of The Search Signal, covering AI-powered search, generative engine optimization, and the future of brand discovery.
