The Challenge
Telecommunications is one of the highest-intent categories in consumer AI search. When a customer asks ChatGPT "which mobile plan is best for international calls" or queries Google AI Overviews for "fastest home fibre in [city]," the AI answers with specific brands, plans, and recommendations. Operators that appear in those answers capture consideration at the moment of highest purchase intent, before the customer has visited any website. Those that are invisible lose customers they never knew they were losing.
This telecommunications provider had a strong traditional search position, ranking on page one for hundreds of commercial terms, but an AI search audit revealed near-total absence from AI-generated answers. Across 320 product and service queries tracked on Google AI Overviews, ChatGPT, Perplexity, and Google Gemini, the brand was cited in fewer than 4% of responses. Competitors with weaker conventional SEO profiles were appearing consistently in AI answers because their content was structured, authoritative, and easy for AI models to extract and synthesise.
The root cause was structural: the brand's content was written for keyword rankings, not for AI comprehension. Pages answered search algorithms rather than real questions. Entity relationships were inconsistent. The brand had no structured FAQ or comparison architecture. Schema was absent from most service pages. And the content that answered the questions customers were actually asking AI was scattered, duplicated, or simply missing.
"Traditional SEO wins rankings. GEO wins the conversation, and in a high-stakes category like telecom, the conversation is where decisions are made."
The Approach
AI Search Landscape Audit
A comprehensive baseline was established across 320 queries spanning five category clusters: mobile plans, home broadband, business connectivity, roaming and international, and device bundles. Each query was tested across Google AI Overviews, ChatGPT web search, Perplexity, and Google Gemini, recording citation rates, competitor presence, and the content sources AI platforms were drawing from. The audit identified that three content competitors, not telecom operators, were dominating AI responses by publishing structured comparison content that AI models could easily cite. This became the strategic template.
Entity Architecture & Brand Signal Consolidation
The brand's entity footprint was audited across Wikipedia, Wikidata, major review platforms, news archives, and third-party directories. Inconsistencies in brand name formatting, address details, service descriptions, and associated entities were identified and corrected across 80+ external sources. A structured entity knowledge base was designed, documenting the canonical name, founding date, service areas, product categories, and known associations, and deployed via Organisation and Service schema on all primary pages. This gave AI models a clear, consistent, trustworthy signal about who the brand was and what it offered.
Answer-Layer Content Architecture
Fifty-eight existing service and product pages were restructured using an answer-layer content model: each page was rebuilt around a primary question the customer would ask AI, followed by a clear, extractable answer in the first 120 words, a structured comparison table, FAQ blocks (each with its own schema), and a plain-language summary. Fourteen new pages were created to fill gaps in the AI coverage map, covering query clusters where the brand had no addressable content. Every piece of content was written to be citable: clear authorship, transparent sourcing, and a distinct point of view rather than generic commercial copy.
Schema & Structured Data Implementation
A full schema implementation programme covered all 72 restructured and new pages. Schema types deployed included Product, Service, FAQPage, HowTo, BreadcrumbList, and a custom telecom Service extension for plan and tariff data. Speakable schema was applied to the first two paragraphs of each page to flag extractable content directly to AI crawlers. All schema was validated against Google's Rich Results Test and Schema.org validators before deployment. The implementation reduced AI ambiguity about what each page represented, a critical factor in whether AI models chose to cite the page in generated answers.
AI Crawler Access & llms.txt Configuration
A technical crawl audit identified that several AI crawlers, including GPTBot, ClaudeBot, and PerplexityBot, were being inadvertently blocked by legacy robots.txt rules written before GEO became a strategic consideration. These blocks were removed and replaced with explicit allow rules for all major AI crawlers. An llms.txt file was configured at the domain root, declaring the brand's key service pages, entity relationships, and authoritative sources to assist AI models in understanding site structure and content priorities. Crawl monitoring was set up to track AI bot activity week-on-week as GEO changes rolled out.
Citation Monitoring & Iteration Framework
A 320-query citation tracking system was established, monitoring AI citation rates weekly across all four platforms, broken down by product category and query type. The monitoring framework identified which content updates were driving citation improvements, which categories needed further restructuring, and which competitor content sources were being consistently preferred by AI models. Monthly optimisation sprints used this data to iterate content, update schema, and refine entity signals, compounding the GEO gains throughout the engagement.
Results
60%
Google AI Overview Coverage
4 Mo
Time to Result
72
Pages Rebuilt
320
Queries Monitored
AI Citation Rate by Product Category, Before vs After (% of tracked queries)
AI Platform Citation Distribution
Google AI Overview Citation Rate, 16-Week Growth
“We had invested heavily in SEO for years and ranked well by every conventional metric, yet when customers asked AI which provider to choose, we simply did not appear. Rima identified exactly why and rebuilt our presence from the inside out. Within four months, our brand was being recommended by AI in categories that drive our highest-value acquisitions.”
Key Deliverables
AI search landscape audit, 320 queries, 4 platforms
Entity architecture & 80+ external source corrections
58 pages restructured + 14 new pages created
Full schema implementation across 72 pages
AI crawler access audit & robots.txt remediation
llms.txt configuration & speakable schema
Weekly citation monitoring framework (ongoing)
Monthly GEO optimisation sprints & iteration reports
GEO Advisory
Is your brand visible in AI search?
Most established brands with strong traditional SEO remain invisible to AI platforms. A GEO audit is the first step to understanding your current position, and what it takes to change it.
Related Case Studies