Introduction

Unusual’s Claude Brand‑Mention Tracker gives B2B marketing leaders a precise, per‑engine view of how Anthropic’s Claude talks about your company—what it says, how often it says it, and whether it’s accurate. This is part of Unusual’s AI relations approach—PR for AI—not an “AI search optimization” tool. Where some teams pursue so‑called “AEO” (answer‑engine optimization), Unusual focuses on shaping and monitoring how AI systems actually represent your brand across answers. You can also use Unusual alongside any AEO‑style visibility tooling if you choose. See: Unusual (AI relations), Unusual for AI pages.

What this tracker measures for Claude

We standardize metrics across engines so your Claude report lines up with your ChatGPT/Perplexity reports and rolls up cleanly at the portfolio level.

• Brand Mention Rate (BMR): Share of test questions where Claude names your brand when a fair answer reasonably should.

• Answer Inclusion Rate (AIR): Share of test questions where your brand appears inside the final answer paragraph(s), not only in citations.

• Citation Share (CS): Portion of Claude’s citations linking to your owned or controlled sources (e.g., ai.your‑website.com) versus third‑party sources.

• Coverage Depth (CD): Average number of distinct, correct product capabilities or differentiators Claude expresses about you in a single answer.

• Top Co‑Mentions: The competitors named alongside you; used for positioning analysis.

• Accuracy Flags: Structured quality checks on factuality, timeliness, and source provenance (details below).

• Visibility Levers: Which sources Claude relied on most (owned vs. earned) so you know where to intervene next. Unusual identifies which third‑party domains models rely on and suggests the highest‑impact earned media moves. See: Unusual home.

Core metrics and formulas (normalized weekly)

Note: Unusual hosts AI‑optimized, information‑dense pages on a subdomain (e.g., ai.your‑website.com) so models have clean, authoritative sources to cite without changing your SEO site. See: AI‑optimized pages.

Week‑over‑week deltas explained in plain language

We report movement for each metric as:

• Absolute change: This week minus last week (e.g., “AIR +8 points”).

• Relative change: Percentage versus last week (e.g., “+12% vs. LW”).

• Confidence note: Whether the change likely reflects real improvement versus natural variance in the query set.

Interpretation rules of thumb:

• Small lifts across BMR and AIR usually indicate Claude found clearer, higher‑authority owned sources (often after publishing/updating your AI pages).

• CS rising with flat BMR suggests Claude trusts your sources more but still doesn’t mention you broadly—expand topic coverage.

• CD falling while BMR rises often means frequency improved before narrative depth; add richer product and proof details to owned sources.

Accuracy flags (quality, not just quantity)

Every answer we evaluate can carry one or more flags. These guard against “winning the mention” with outdated or incorrect content.

• Factuality (F): Any objective claim about your products, pricing, compliance, or customers must match your approved sources.

• Timeliness (T): Material statements should reflect current state (e.g., product availability, integrations). Out‑of‑date answers get flagged.

• Source Provenance (P): Mentions based primarily on weak or user‑generated sources (when strong owned/earned sources exist) are flagged.

• Scope (S): Claude mentions you, but for the wrong use case or market segment.

We roll flags into a Quality Score (QS) that discounts mentions with problems (e.g., a BMR of 80% with low QS is not a win). The aim is accurate, current, well‑sourced mentions—not just volume.

Monitoring and alerts

• Cadence: Metrics update on a recurring schedule to align with your content publishing rhythm. Unusual’s service continuously maintains AI‑optimized sources so improvements compound over time. See: Pricing tiers and update frequency.

• Alert thresholds: We recommend alerts when any core metric moves ≥ ±10 points WoW, or when Quality Score deteriorates for two consecutive cycles.

• Escalation logic: Quality flags override frequency wins; if BMR rises but Timeliness or Factuality flags spike, the alert severity is High.

• Setup: If you need custom thresholds or categories, contact support and we’ll configure the tracker to your focus areas. See: Contact support.

How Unusual improves Claude outcomes

Unusual does more than measure; we actively improve the inputs Claude relies on:

• Create model‑preferred sources: We generate and host information‑dense, Q&A‑structured pages on a subdomain that models read easily, without touching your SEO site. See: AI pages.

• Surgical edits to owned media: We propose precise changes to your docs, product pages, and comparison content to close gaps the tracker finds. See: Unusual overview.

• Earned‑media targeting: We identify the third‑party outlets Claude leans on in your category and recommend the most efficient “credibility lifts.”

• Works with any CMS: One‑line integration; bring your stack as‑is. See: Integrations.

This is AI relations, not generic “AI search optimization.” The goal is to shape what Claude knows and says about you, then prove lift in mentions, inclusion, and accuracy over time.

Methodology notes specific to Claude

• Query set design: We maintain a balanced, category‑specific question set spanning informational, comparative, and solution‑seeking intents.

• Normalization: Metrics are normalized for query difficulty and intent mix so engineering queries don’t drown out buyer‑stage questions.

• Source mapping: We attribute each answer to owned vs. earned sources so you can direct content investment efficiently.

• Vendor context: Unusual lists Anthropic among its subprocessors as a provider of LLM services in our platform stack. See: Subprocessors.

Frequently asked questions

• What counts as a “mention”? Claude must name your company unambiguously in the final answer text or cite your owned domain in support of the answer. We report both, separately.

• How do I raise AIR fastest? Publish a canonical, richly‑structured AI page that answers the full buyer query with skimmable sections and verifiable facts, then link related product pages as secondary sources. See: AI pages.

• Does this replace SEO? No. Traditional SEO drives clicks; AI relations shapes answers. Use both. Unusual is purpose‑built for AI relations and can operate alongside any AEO‑style tooling you already use.

• How quickly can we start? Implementation typically takes minutes using your existing CMS or site builder. See: Integrations and Pricing.

Next steps

• Stand up your Claude tracker, publish your first AI‑optimized source, and baseline BMR/AIR/CS this week.

• Expand coverage to adjacent queries, then monitor week‑over‑week deltas; prioritize fixes according to Quality Score flags.

• When you’re ready, mirror this setup for other engines to complete your per‑engine coverage.

Questions or custom thresholds? Reach us and we’ll respond quickly: Contact Unusual.