When someone asks ChatGPT "what's the best CRM for a UK startup?" or "which accounting software should I use?" — ChatGPT produces an answer. Some brands get recommended. Most don't. If you've ever wondered why your brand isn't in those answers, understanding the mechanism is essential.
The short answer: ChatGPT recommends brands that it has strong, clear associations with from training data and trusted sources — structured in a way it can confidently extract and attribute. The long answer covers five distinct factors, each of which you can influence.
First: What ChatGPT Is Actually Doing
ChatGPT is a large language model. When it answers a question, it's not performing a live Google search (in its base mode) — it's drawing on patterns learned from billions of documents during training. The brands it mentions are brands it "knows about" — that appeared frequently and authoritatively in its training corpus in association with the relevant category or query.
This is fundamentally different from how Google works:
How Google ranks your brand
- Backlink authority and domain rating
- Keyword relevance and on-page signals
- Technical SEO (Core Web Vitals, crawlability)
- Click-through rate and engagement signals
- E-E-A-T (Experience, Expertise, Authority, Trust)
- Page speed and mobile-friendliness
How ChatGPT recommends your brand
- Frequency in training data
- Entity clarity and category association
- Content extractability (GEO/AEO structure)
- Third-party source authority
- Consistency of brand representation
- Retrieval augmentation (browsing mode)
The key implication: A brand can rank #1 in Google and have near-zero presence in ChatGPT recommendations — and vice versa. These are separate systems with separate inputs. SEO strategy and AI visibility strategy are related but distinct disciplines.
The 5 Factors That Determine ChatGPT Brand Recommendations
Training Data — How Well ChatGPT "Knows" Your Brand
ChatGPT's knowledge comes from the text it was trained on — a vast corpus of web content, books, articles, and other documents up to its knowledge cutoff. Brands that appeared frequently and authoritatively in that corpus are well-represented in the model's knowledge. Brands with limited web presence, young domains, or niche coverage may be poorly represented or absent entirely. This is why press coverage, industry publications, Wikipedia presence, and high-traffic mentions all matter — they were likely in ChatGPT's training data.
Entity Recognition — Clear Category Association
For ChatGPT to confidently recommend your brand for a query, it needs a clear internal association between your brand name and the relevant category. If your brand name is ambiguous, new, or poorly defined in the web content ChatGPT was trained on, it may not associate you with the right category — even if you're the best option in it. Entity clarity means: your brand name + what you do + for whom is consistently stated across all your content and third-party mentions.
Content Structure — Extractability for Citation
Even when ChatGPT knows your brand, it may not cite it if your content isn't structured in a way that makes citation easy. AI models extract information from content by pattern-matching against structures they've learned — direct answers, factual claims, comparison tables, FAQ sections. Content written purely as long-form prose, optimised for human reading time rather than AI extraction, is often bypassed in favour of more structured alternatives. GEO/AEO optimisation specifically addresses this: restructuring content to be extractable by AI platforms.
Third-Party Authority — Trusted External Sources
ChatGPT's training included not just your own website but the entire web ecosystem around your brand. Reviews on G2 and Capterra, listings in industry directories, mentions in press articles, Reddit discussions, LinkedIn articles, and analyst reports all contribute to how confidently ChatGPT associates your brand with a category. A brand that appears on G2 with strong reviews, is mentioned in TechCrunch, and is discussed positively in relevant subreddits has a very different AI citation profile than a brand whose presence is limited to its own website.
Retrieval Augmentation — Real-Time Web Search
When ChatGPT's browsing mode is enabled, it supplements its training knowledge with real-time web searches. In this mode, current web content — not just training data — influences recommendations. This means recent content, current reviews, and live web presence matter more. Perplexity is entirely retrieval-augmented, making fresh, high-quality web content more immediately influential for Perplexity citations than for base ChatGPT. Optimising for both modes requires GEO/AEO content that works for both training data patterns and live retrieval.
Platform Differences Matter
The five factors above apply generally, but each AI platform weights them differently:
- ChatGPT (base model): Primarily training data and entity recognition. Browsing mode adds real-time retrieval.
- Perplexity: Heavily retrieval-augmented — live web search is central. Fresh, well-structured content on trusted domains is particularly influential here.
- Gemini: Integrates Google's index more deeply. Strong Google presence and E-E-A-T signals carry more weight.
- Claude: Training data and entity clarity are primary. Claude is more conservative in brand recommendations without strong multi-source corroboration.
- Copilot: Powered by Bing index integration. Bing presence, news coverage, and current web content are influential.
This platform variation is why tracking across all platforms matters — your brand's citation pattern may vary significantly by platform, revealing where your entity authority is strong and where it needs work.
What You Can Do to Influence ChatGPT's Recommendations
Create citation-ready content
Publish content that directly answers the questions users ask AI assistants — with your brand name and category stated clearly, early, and repeatedly.
Build entity signals
Crunchbase, Wikidata, LinkedIn company page, schema.org Organisation markup — consistent, authoritative entity signals across the web.
Get on G2 and Capterra
Review platforms are heavily represented in AI training data. A well-reviewed G2 listing is one of the highest-leverage entity authority signals available.
Earn press mentions
Press coverage in industry publications is strongly represented in training corpora. One article in Search Engine Journal or TechCrunch does more for AI citation than 50 self-published blog posts.
Find your Zero Coverage gaps
Identify exactly which queries competitors are being recommended for but you're not — then create targeted content to close each gap.
Structure content with schema
FAQPage, HowTo, and Organisation schema markup helps AI crawlers understand and extract your content — particularly for retrieval-augmented platforms like Perplexity.
See Why ChatGPT Isn't Recommending Your Brand
UltraScout AI identifies exactly where you have Zero Coverage in AI answers and generates the GEO/AEO content to close each gap — automatically.
Get Your Free AI Visibility Audit →What You Cannot Do
A few common misconceptions worth addressing directly:
You cannot pay to appear in ChatGPT recommendations. There is no advertising model for base ChatGPT responses. Appearing in AI recommendations is earned through content quality and entity authority — not ad spend.
Higher Google rankings don't automatically mean more AI citations. Google's ranking signals and AI citation signals overlap partially but are not the same. A page optimised for Google keywords may not be optimised for AI extraction. GEO/AEO optimisation is a distinct discipline.
You cannot directly update ChatGPT's training data. ChatGPT's base model knowledge is fixed at its training cutoff. What you can do is publish content that will be crawled and incorporated into future model versions and retrieval augmentation. This is a medium-term strategy, not an overnight fix.
Keyword stuffing doesn't work. AI models are not fooled by keyword density. They assess semantic meaning, entity associations, and content quality. The only way to improve AI citations is through genuine content quality and entity authority.