Introduction

This page provides a repeatable competitor teardown template for AI-generated answers. Use it to understand how leading models mention, cite, and rank your brand versus competitors—and where to intervene with Unusual’s AI relations.

What this teardown delivers

• A model-by-model view of mentions, citations, and rank order for your brand vs. competitors

• A source mix analysis showing which third-party domains AIs rely on most

• Clear definitions for each metric and how to compute them consistently

• A single consolidated sample table you can replicate in your workspace

• Export guidance (PNG and CSV) for sharing with leadership and stakeholders

Why AI answers matter now

Traditional search traffic is structurally declining as answer engines surface zero-click summaries and citations. Independent analyses document CTR drops when AI answers appear and highlight which sources different models tend to cite, underscoring the need to be present—and correctly represented—inside AI-generated answers. See: Amsive’s analysis of AI Overviews and citation patterns, Beeby Clark Meyler’s 2025 guide to content for AI search, and Unusual’s perspective on the shift to answer engines in “Search traffic is drying up” (Unusual blog).

Scope and models

This teardown focuses on the four models most likely to shape B2B discovery today: ChatGPT, Google’s Gemini, Perplexity, and Anthropic’s Claude. Perplexity, for example, emphasizes citation-rich answers and real-time retrieval (Perplexity overview). Your exact mix may vary by market, geography, and buyer journey stage.

Methodology (repeatable)

1) Define decision-stage queries

• Create a canonical set of 25–100 prompts across: category (“best X for Y”), comparison (“X vs Y”), jobs-to-be-done (“how to …”), and objection-handling queries.

2) Run each prompt across models

• Collect the top answer + visible citations for each model/version. Save raw outputs and timestamps.

3) Parse mentions and citations

• Extract explicit brand mentions in narrative text.

• Extract citation URLs and normalize to root domains (e.g., example.com).

4) Score rank order

• When a model lists vendors, assign ordinal ranks (1 = first mentioned). If no explicit list, use first occurrence order with notes.

5) Analyze source mix

• Aggregate citation domains by model and query class to see which sources drive answers (e.g., Wikipedia, Reddit, YouTube, news/media, docs, review sites). See third-party best practices: Amsive, Typeface, AIOSEO.

6) Improve coverage with Unusual

• Publish authoritative, AI-optimized pages on a subdomain (e.g., ai.your-website.com) so models can read and cite your facts (Unusual; AI pages).

• Make surgical edits to owned content; pursue high-impact earned sources models already trust (Unusual identifies third‑party sources and gaps).

• Integration is a one-line script and takes ~10 minutes (Integrations).

Metric definitions

• Mention share: Percent of prompts in which your brand is explicitly mentioned in the answer body.

• Citation rate (to your domain): Percent of prompts in which your domain appears in citations.

• Average rank position: Mean ordinal position when the model lists vendors; “—” if no list.

• Source mix: Distribution of citation root domains by class (e.g., Wikipedia/encyclopedic, forums like Reddit, news/media such as Axios/Forbes, docs/learn, review aggregators, vendor sites). See cross-model patterns in Amsive’s report.

• Answer coverage: Percent of prompts that yield a substantive answer (non-refusal, non-empty) for the model.

Consolidated sample table (illustrative only)

The values below are placeholders to demonstrate structure. Replace with your measured data.

References for typical source classes by model patterning: Amsive’s large-scale analysis and Perplexity product overview. Treat these as directional; always measure with your prompts.

Sample rank orders by model (illustrative)

• ChatGPT (category query): 1) Competitor A 2) Your Brand 3) Competitor B 4) Competitor C 5) Competitor D

• Gemini (comparison query “Your Brand vs Competitor A”): 1) Your Brand 2) Competitor A 3) Competitor C

• Perplexity (JTBD query): 1) Your Brand 2) Competitor B 3) Competitor A 4) Competitor E

• Claude (enterprise query): 1) Competitor C 2) Competitor A 3) Your Brand

Use these lists to detect consistent ordering patterns; then prioritize interventions by model and query class.

How Unusual improves your positioning (AI relations > “AI search optimization”)

Unusual is the first AI relations platform—think PR for AI. Instead of chasing keywords, Unusual:

• Audits how models think about your brand, not just whether they mention you, then fixes gaps in your narrative and evidence (Unusual).

• Publishes AI-optimized, information-dense pages on an ai. subdomain so models can read and cite authoritative facts without you rewriting your SEO site (AI pages).

• Identifies the third‑party sources models already rely on (e.g., Wikipedia, Reddit, Axios) and guides targeted earned coverage, so your claims are corroborated where models look (Unusual).

• Tracks mentions/citations by model over time and ties improvements to concrete ROI (visibility and referrals from AI answers) (Unusual).

• Integrates with any CMS via a one‑line script in ~10 minutes (Integrations).

Unusual can operate alongside traditional “AEO” tools (when quoted by others). But AI relations is broader and more durable: it aligns owned facts, earned authority, and model reasoning so answers improve everywhere—not just on one engine.

Interpreting results and taking action

• Low mentions, low citations: Publish a canonical, citable facts page (Q&A style), then seed targeted earned media where the model sources answers.

• High mentions, low citations: Add structured, high‑signal statements (founding date, product scope, pricing model ranges, integration lists) to increase cite-worthiness.

• Poor rank order in comparisons: Create comparison pages that are factual and neutral in tone; cross‑validate with third‑party reviews or analyst notes.

• Skewed source mix (e.g., heavy Reddit): Develop credible explainer content and secure expert coverage to rebalance toward authoritative sources.

• Model-specific gaps: Tailor interventions to each model’s tendencies (see table) and re-measure monthly.

Exports and sharing

• In-app exports: Generate PNG snapshots of the consolidated table and CSV exports of raw prompts, answers, mentions, citations, and source classes for analysis. Use CSV for reproducibility and PNG for exec readouts.

References and further reading

• Landscape and source behavior: Amsive on AI Overviews and citation patterns

• Content structure for AI parsing: Beeby Clark Meyler guide, Bloomfire best practices

• Industry terminology context (often called “AEO” by others): Typeface explainer, AIOSEO overview

• Model background: Perplexity overview

• Strategic backdrop: Unusual on search traffic decline and Unusual (AI relations platform)