Most ChatGPT optimisation services treat the problem as a content and backlinks exercise. We start one layer deeper, with how ChatGPT actually constructs an answer before a single word is written.
When a user submits a query, ChatGPT fans the prompt out into an average of 2.1 sub-queries, each probing a different angle: best options in the category, named brand comparisons, review signals from third-party sites, features lists, and date-qualified searches for fresh content. It then runs Reciprocal Rank Fusion across those sub-queries, meaning brands that surface across multiple fan-outs are systematically ranked higher than brands that only appear for one.
This is why "10 best" listicles, comparison and review sites dominate AI search.
Our agency GEO approach maps the specific fan-out patterns firing on the queries your buyers use, identifies which of those sub-query surfaces you appear on and which represent the strongest growth opportunity, and builds a programme that closes those gaps in order of citation impact.
Request an AI Visibility AuditWe work across every layer of the ChatGPT citation stack, from mapping the hidden sub-queries your category triggers to building the content, authority and technical foundations that make your brand the answer across all of them.
Before optimising anything, we need to know exactly what ChatGPT is searching for when your buyers ask about your category. We analyse the fan-out queries firing on your priority prompts, the "best", "top", "reviews", "[competitor] vs [you]", and year-qualified searches ChatGPT runs behind the scenes, and map which of those surfaces you currently appear on.
This gives us a ranked gap analysis: which fan-out angles represent the highest citation upside and should be addressed first. Every engagement begins here, because optimising with full knowledge of the fan-outs produces a programme precisely targeted to the searches that drive citations in your category.
Best" is the single most injected word in ChatGPT's fan-out queries, added even when the original prompt contained no evaluative language at all. Analysis of 5 million fan-outs shows it appears in 24.3% of advice-style prompts, and it directly explains why listicle content consistently dominates AI citations.
We audit which existing "best of" and comparison resources in your category are being cited, identify the ones with the highest fan-out coverage, and build a strategy to earn inclusion in them through digital PR, original data, and content that positions your brand as the definitive answer to "best [your service] for [specific need]".
"Reviews" is the third most injected word in ChatGPT's hidden sub-queries. ChatGPT actively searches for how your brand is rated across third-party platforms, Glassdoor, G2, Trustpilot, Gartner, and niche industry directories, even when the user's original question contained no request for reviews. This means your citation quality is directly shaped by platforms you may not be actively managing.
We audit which review sites are being pulled into your category's fan-outs, assess how your brand is described across them, identify any outlier signals, and build a management strategy that ensures ChatGPT's review-driven sub-queries return signals that support your citation prospects.
A single well-written page rarely surfaces across multiple fan-out angles. To benefit from Reciprocal Rank Fusion, ChatGPT's algorithm for combining scores across sub-queries, your content needs to answer several fan-out dimensions simultaneously: what your service is, why it is the best option, how it compares to named alternatives, what users say about it, and what makes it the current leading choice.
We restructure and develop content around answer-first formatting, comparison architecture, FAQ blocks with FAQPage schema, and passage-extractable summaries specifically mapped to the fan-out queries firing in your category, so a single authoritative page can surface across multiple hidden searches simultaneously.
ChatGPT injects the current year into 5.44% of its fan-out queries, actively seeking out recently updated content to surface alongside established sources. Content that is clearly current gains a material advantage in these date-qualified sub-queries.
We identify which pages on your site are already being used as ChatGPT sources, prioritise those for structured freshness updates, new data points, updated comparisons, current year references, schema date signals, and maintain a cadence that keeps your highest-value content ahead of the freshness filter without requiring a full site overhaul.
ChatGPT reasons in entities before it reasons in content. It needs to know unambiguously who your brand is, what category you operate in, which topics you are authoritative on, and how your entity connects to the named brands, services and people it already understands.
We build entity clarity through consistent structured data, brand disambiguation across the web, knowledge graph connections, and topical association patterns, so ChatGPT can confidently place your brand in the right category and cite it when relevant prompts are searched.
ChatGPT's fan-outs frequently target specific trusted source types: industry publications, authoritative comparison sites, community platforms, and credible roundups. Being present in these sources is a structural requirement for multi-fan-out citation.
We map the third-party domains appearing most frequently in your category's fan-out patterns, identify which ones your competitors are cited by that you are positioned to target, and build a targeted authority programme through digital PR, data-led content, and strategic placements that earns your brand coverage in the exact sources ChatGPT is already searching when it fans out your buyers' queries.
ChatGPT's web retrieval can only cite content it can access and parse. Many sites limit AI crawler access through robots.txt rules, firewall and CDN configurations, or content rendered in client-side JavaScript that GPTBot cannot execute.
We audit and resolve every technical barrier: reviewing server rules against known AI agents, implementing llms.txt to direct crawlers toward your most authoritative pages, ensuring core content is served in raw crawlable HTML, and deploying the schema types, FAQPage, Organization, Article, Service, Person, that give ChatGPT explicit, unambiguous signals about what each page is and why it should be cited. A technically accessible, well-structured site is the prerequisite for sustained citation.
Buyers were researching providers through AI tools like Google AI Overviews, ChatGPT and Perplexity and Six Degrees wasn't being recommended.
Our three-pillar GEO strategy fixed that, earning the #1 AI Overview citation for "penetration testing services" and lifting AI Share of Voice from 0% to 20% for core security products in seven days. Technically we analysed log files to see exactly what ChatGPT's retrieval was crawling and restructured content into the clear title, answer, CTA format ChatGPT favours for passage retrieval and citation.
The result: +200% penetration testing conversions and +500% across managed security services.
Read our GEO case study
When someone submits a query to ChatGPT, the model generates an average of 2.1 hidden sub-queries, injecting words like "best", "top", "reviews", the current year, and named competitors, then searches for each separately before blending the results into a single answer. This process is called query fan-out. It matters because ChatGPT uses Reciprocal Rank Fusion to score results: brands that appear across multiple fan-out searches are systematically ranked more highly than brands that only surface for one. Understanding and optimising for those hidden sub-queries is the core of effective ChatGPT SEO.
ChatGPT is trained to produce recommendations that are useful and trustworthy. When a user asks for advice or a comparison in any category, ChatGPT interprets that as an evaluation task and enriches its search accordingly, adding "best" to find ranked listicle content, adding "reviews" to find third-party sentiment signals, and adding the current year to find fresh information. Analysis of 5 million fan-out queries found "best" is the most frequently injected word overall, appearing in nearly one in four advice-style prompts even when the original question contained no evaluative language. This is why "10 Best" listicles and review platform profiles have an outsized influence on ChatGPT citations relative to their traditional SEO value.
GEO, or Generative Engine Optimisation, is the broader discipline of optimising across all AI search platforms: ChatGPT, Perplexity, Google Gemini, Claude, and others. ChatGPT optimisation is specifically concerned with how OpenAI's model gathers, retrieves and cites information, which has distinct characteristics. ChatGPT averages 2.1 fan-out queries per prompt, draws on Bing-indexed content for its live retrieval, applies Reciprocal Rank Fusion to score sources across sub-queries, and has particular patterns in which source types it trusts for reviews, comparisons and "best of" searches. If you already have a GEO programme covering AI search broadly, ChatGPT GEO layers on the platform-specific optimisation, fan-out mapping, Bing visibility, review platform management, that makes your brand the consistently cited answer within ChatGPT specifically.
Usually because the content depends on JavaScript that AI crawlers do not execute. Most AI crawlers read only the raw HTML your server returns, so anything loaded client-side, hidden behind tabs or interactions, or placed behind a login can be invisible to them even when Googlebot indexes it perfectly. Server-side rendering, accessible HTML and unblocked crawlers are what close that gap.
We approach ChatGPT GEO from the mechanism up, starting with the query fan-out patterns firing in your category, grounded in how ChatGPT actually retrieves and scores sources. We have been resolving the foundational signals that drive AI citation for over fifteen years through technical search work, and we apply that depth to the specific retrieval mechanics ChatGPT uses: fan-out coverage, Reciprocal Rank Fusion, review platform signals, freshness requirements, and entity clarity. We diagnose your specific gap profile, build a programme against it, and measure progress in terms that connect directly to citation outcomes and commercial pipeline.
"Most ChatGPT optimisation stops at content and links. We go deeper, mapping the exact sub-queries ChatGPT fires when someone asks for a recommendation in your category, then building the content, authority and technical signals that get you cited across all of them. It's the same retrieval-and-ranking depth we've applied to search for over fifteen years, now positioned at how ChatGPT actually builds an answer."James FooteSEO Director
