SeenandCited

The SeenAndCited Methodology in Practice

Master the sequential workflow for improving AI engine visibility using the SeenAndCited framework: from initial discovery through to iterative measurement and refinement.

15 min read
Foundations

Introduction

The SeenAndCited methodology moves beyond theoretical understanding of AI engines to provide a structured, repeatable framework for practitioners. Unlike traditional SEO, which often focuses on keyword rankings and backlinks, AI Visibility Practitioner (AVP) work requires a focus on semantic relationships, conversational relevance, and citation accuracy across Large Language Models (LLMs). This lesson breaks down our six-stage proprietary workflow: Discover, Monitor, Analyse, Recommend, Execute, and Measure. By following this sequence, practitioners can ensure their clients’ brands are not only mentioned but accurately cited and recommended by engines like Perplexity, ChatGPT (Search), and Gemini.

Phase 1: Discover – Identifying the LLM Footprint

Discovery is the foundational audit phase. It involves determining how an AI model currently perceives a brand, its products, and its competitive set.

Practical Steps:

  1. Baseline Prompting: Use a structured set of prompts (Informational, Navigational, Transactional, and Comparison) across the three major LLMs.
  2. Competitor Mapping: Identify which competitors are consistently cited for your primary service categories.
  3. Entity Health Check: Check common knowledge bases (Wikipedia, Crunchbase, LinkedIn, Industry-specific wikis) to see how the 'entity' of the brand is defined.

Example: If you are working for a 'Sustainable Packaging SaaS', you would prompt: "Who are the leaders in plastic-free supply chain software for European SMEs?" If your client is absent, discovery confirms a visibility gap.

Phase 2: Monitor – Tracking the Volatility

AI responses are non-deterministic; they change frequently based on model updates and new training data. Monitoring involves setting up a cadence for checking visibility performance.

Practical Steps:

  1. Sentiment Tracking: Is the AI describing the brand positively, neutrally, or negatively?
  2. Citation Frequency: How many times is the brand’s domain cited as a primary source compared to secondary news sources?
  3. Snippet Attribution: Is the AI using your content as the direct answer (the 'Answer Box' equivalent for LLMs)?

Phase 3: Analyse – The 'Why' Behind the Citation

Analysis is the most critical technical step. You must deconstruct the source material the AI uses to generate its response. Most modern AI search engines (GEO/AEO) use Retrieval-Augmented Generation (RAG). They fetch top search results and then summarise them.

Key Metrics to Analyse:

  • Source Authority: Are the sources cited by the AI high-authority industry journals or low-quality scraped sites?
  • Semantic Proximity: How closely related is your brand to the user’s intent keywords in the training data?
  • Fact Accuracy: Is the AI hallucinating facts about your price points or features? If so, your structured data or 'About' page may be unclear.

Phase 4: Recommend – Strategic Planning

Once you know the gaps, you create a prioritised recommendation roadmap. In AI Visibility, recommendations often fall into three buckets: Technical (Schema), Content (Pragmatic/Semantic), and External (Digital PR/Citations).

Typical Recommendations:

  • Expand Schema.org Markup: Move beyond basic Organization schema to include Product, Review, FAQPage, and SameAs links to verified profiles.
  • Update Fact-Heavy Content: Create dedicated 'Comparison' pages that provide the AI with easy-to-parse data tables.
  • Third-Party Calibration: Target industry-specific directories that the AI consistently uses as sources for your competitors.

Phase 5: Execute – Implementing Changes

Execution involves coordinating with developers and content teams to move the needle.

Worked Example: 'The FinTech Startup'

  • Problem: Perplexity cites a competitor for 'Best Neobank for Freelancers' because the competitor has a specific landing page with a comparison table.
  • Execution: We implement a 'Comparison Hub' on our client’s site. We use structured data to define the 'Service' entity and clear, non-ambiguous headers (H2s and H3s). We then update the brand's LinkedIn and Crunchbase profiles to mirror this specific positioning.

Phase 6: Measure – Quantifying Success

Measuring AI visibility is different from measuring organic traffic. While traffic is a secondary benefit, the primary goal is 'Share of Model' (SoM).

Measurement Metrics:

  • Citation Share: Your brand's percentage of total citations for a specific category prompt.
  • Sentiment Shift: Moving from 'neutral' mentions to 'recommended' mentions.
  • Conversion via AI Referral: Tracking UTM-tagged links that originate from AI engines (visible in Google Search Console as referral traffic from domains like chatgpt.com or perplexity.ai).

Putting it into Practice

To apply the SeenAndCited methodology effectively, start small. Choose one high-value product or service and run through the six phases over a 30-day period.

  1. Week 1 (Discover & Monitor): Map the current landscape and set a baseline.
  2. Week 2 (Analyse): Identify exactly why the top 3 cited sources are being preferred by the AI.
  3. Week 3 (Recommend & Execute): Update your technical schema and on-page content to provide 'perfect' answers for those RAG pipelines.
  4. Week 4 (Measure): Re-prompt and check for shifts in citation behavior and sentiment.

Remember: AI engines are looking for the path of least resistance to the most accurate information. Your job is to make your brand the most legible, credible answer available.

Visual diagram

[ diagram placeholder ]

A circular process diagram showing the six phases of the SeenAndCited methodology with arrows indicating a continuous feedback loop and 'Data-Driven Insight' at the center.

Exercise

Select a client or your own brand and ask Perplexity: 'What are the pros and cons of [Brand Name] compared to [Top Competitor]?'. Identify three specific sources the AI cites for its information. Determine if these sources are owned (your site) or earned (third-party sites) and list one update you could make to improve the accuracy of its answer.

Key takeaways

  • The SeenAndCited methodology is a six-stage circular workflow: Discover, Monitor, Analyse, Recommend, Execute, Measure.
  • Discovery requires testing across various prompt types: Informational, Navigational, Transactional, and Comparative.
  • AI responses are non-deterministic, making regular Monitoring essential for capturing performance volatility.
  • Analysis focuses on RAG (Retrieval-Augmented Generation) sources to understand 'why' an AI chooses specific citations.
  • Technical recommendations often involve deepening Schema.org implementation beyond basic organisational tags.
  • Content recommendations should focus on high legibility, including comparison tables and structured headers.
  • Execution requires a multi-channel approach, updating both owned assets and third-party entity sources.
  • Measuring success involves tracking 'Share of Model' (SoM) and citation frequency rather than just traditional rankings.
  • Hallucinations in AI responses often stem from conflicting or unclear information on the brand’s own digital assets.
  • Practical application works best when focused on a single high-value product category to allow for iterative testing.

Lesson Quiz

Pass at 70%.

1. Which phase of the SeenAndCited methodology involves identifying the current 'LLM Footprint'?
2. What is 'Share of Model' (SoM) used for in the methodology?
3. In the Analyse phase, why is understanding RAG (Retrieval-Augmented Generation) important?
4. Which of these is a 'Technical' recommendation for improving AI visibility?
5. What does it mean that LLM responses are 'non-deterministic'?
6. How can you track referral traffic from AI engines in Google Search Console?
7. What should you do if an AI 'hallucinates' incorrect facts about your brand?
8. What is a 'Comparison Hub' in the context of the Execute phase?
9. Which of these is NOT one of the six stages in the SeenAndCited methodology?
10. During the Recommend phase, why target third-party industry directories?
Create a free account to save progress and earn a certificate.
From Concepts to PracticeBack to moduleSetting Up Your Practitioner Toolkit