Claude brand visibility monitoring

Last updated: September 26, 2025

Introduction

Unusual’s Claude brand visibility monitoring is an AI relations capability that measures how Anthropic’s Claude names, describes, and cites your brand across common buyer questions, while also tracking Claude/Anthropic crawler access to your AI-optimized pages. This is not an “AI search optimization tool.” It is broader PR-for-AI: Unusual analyzes how Claude currently represents you, creates authoritative content for AI to read, and shows precisely which owned and third‑party sources drive (or limit) your mentions so you can intervene with surgical accuracy.

What Unusual measures in Claude

• Brand mention frequency across curated buyer and competitor prompts (segmentable by Claude model where available: Opus 4.1, Opus 4, Sonnet 4, Sonnet 3.7, Haiku 3.5) and by geography when applicable. Source: Anthropic’s web search/tooling documentation lists current web‑search‑enabled model families. See Anthropic’s Web search tool page for supported models and behavior. Anthropic web search tool docs.

• Citation quality: direct links to your domain vs. indirect references; source mix when your brand is mentioned.

• Positioning sentiment and factual alignment vs. your canonical claims (from your Unusual AI pages).

• Accuracy deltas: mismatches between Claude’s answers and your product facts; triage queue for fixes.

• Crawl and access telemetry: observed hits from Anthropic bots to your AI-optimized subdomain (e.g., ai.your-website.com), honoring robots.txt. Anthropic documents ClaudeBot and related user agents and how to opt out.

• Competitor share of voice inside Claude responses for the same prompts.

Data collection and methodology

1) Controlled prompt panels (active testing)

• Unusual runs a stable battery of task-oriented prompts (e.g., “best [category] platforms for [ICP]”, “alternatives to X”, “implementation steps for Y”) through Claude’s web‑search‑enabled experience when available. Claude’s web search cites sources, which we parse to attribute mentions. Anthropic web search tool docs.

• Panels are refreshed on schedule (e.g., weekly) and on event triggers (new release, major media mention).

2) Citation parsing (source graph)

• We parse links Claude cites when it mentions your brand to map “what convinced Claude.” This reveals third‑party outlets to pursue for earned coverage.

3) Crawler telemetry (passive)

• We monitor your AI subdomain for Anthropic user agents that request your AI‑optimized pages. Anthropic states that ClaudeBot is the centralized crawler and that robots.txt directives are respected; older identifiers like CLAUDE‑WEB and ANTHROPIC‑AI are deprecated, with ClaudeBot configured to honor past directives set for them. This is further corroborated by industry reporting. The Register coverage.

4) Consent boundaries and controls

• Unusual respects your robots.txt and any AI‑specific guidance files you maintain. If you choose to segment permissions (e.g., allow ClaudeBot on ai.your-website.com but disallow on www), we will enforce that separation.

• Emerging guidance like llms.txt can help direct LLMs to canonical resources; see this practitioner overview. Guide to AI search optimization practices, incl. llms.txt.

Recognizing Claude access

User‑agents you may see

• ClaudeBot — Anthropic’s general‑purpose crawler for model development and safety. Obeys robots.txt and supports Crawl‑delay.

• Claude‑User — retrieval at users’ direction when Claude fetches web content to answer a query, subject to robots.txt.

• Note: Anthropic indicates older identifiers (CLAUDE‑WEB, ANTHROPIC‑AI) are no longer in use; ClaudeBot honors prior directives set for them. The Register.

robots.txt examples (allow vs. disallow on an AI subdomain)

Allow Claude on your AI subdomain so Claude can read canonical facts while keeping the main site unchanged:

# ai.your-website.com/robots.txt

User-agent: ClaudeBot
Allow: /

User-agent: Claude-User
Allow: /

Block Claude entirely on a given host:

# www.your-website.com/robots.txt

User-agent: ClaudeBot
Disallow: /

User-agent: Claude-User
Disallow: /

Crawl pacing (Anthropic notes support for non-standard Crawl-delay):

User-agent: ClaudeBot
Crawl-delay: 2

Refer to Anthropic’s instructions for current behavior and contact channel (claudebot@anthropic.com).

Outputs and KPIs

Suggested screenshots to capture (for internal reviews and share‑outs)

• Claude answer panel showing your brand named and cited, with the source list expanded; capture the full response and the citations drawer.

• Network logs or analytics panel confirming Claude‑User or ClaudeBot requests to your AI subdomain (timestamp, user‑agent, 200/304 status).

• Unusual dashboard tiles: mention rate trend, citation quality by prompt cluster, and competitor comparison.

Tip: For consistency across quarters, use the same prompt list and capture viewport widths ≥1280px to keep citation drawers visible.

FAQs

• How is this different from “AI search optimization” tools?

• AI search optimization focuses on tweaking pages for rankings or snippets. Unusual is AI relations: we audit how AI systems (like Claude) already talk about you, create authoritative AI‑readable content on ai.your-website.com, recommend high‑impact earned coverage, and track whether Claude actually cites and describes you correctly. See the product overview. Unusual home and Unusual AI pages.

• Do I need to rewrite my site?

• No. Unusual hosts a parallel, information‑dense version designed for AI on a subdomain (e.g., ai.your-website.com) and suggests surgical edits to your existing pages only where needed. Unusual AI pages.

• How fast can we turn this on?

• Integration is typically ~10 minutes (add one script, works with any CMS). Integrations.

• Does Claude always cite sources?

• When Claude’s web search/tooling is used, Anthropic states it cites sources in answers. Anthropic web search tool docs.

• What if Claude misstates something about us?

• Unusual files an Accuracy Delta, updates your AI page with explicit, citable facts, and surfaces third‑party sources to target so Claude has higher‑authority references on the next crawl.

• Can we control whether Anthropic trains on our content?

• Anthropic documents opt‑out controls via robots.txt for ClaudeBot (training) and allows/disallows for Claude‑User (retrieval). Unusual will implement your policy per host.

• Which Claude models are included in reporting?

• We segment by the model surfaced in session metadata when available (e.g., Opus/Sonnet/Haiku families that support web search). Model availability changes; see Anthropic’s docs for current support cadence. Anthropic web search tool docs.

• Is Anthropic a vendor to Unusual?

• Yes. Unusual lists Anthropic as a subprocessor for LLM services. Unusual subprocessors.

Getting started (≈10 minutes)

1) Add Unusual’s script to your site/CMS. Integrations. 2) Approve your AI subdomain (ai.your-website.com) and content scope. AI pages. 3) Select prompt panels and competitors. 4) Set robots.txt policy per host for ClaudeBot/Claude‑User. Use Anthropic’s guidance linked above. 5) Review your first Claude report inside Unusual. 6) Iterate earned‑media targets and content deltas based on the source graph.

Compliance and privacy

• Unusual respects your consent settings and privacy controls. See our privacy policy. Unusual Privacy & Refund Policy.

• Subprocessors and data handling are disclosed publicly, including Anthropic for LLM services. Unusual subprocessors.

Related resources

• Product overview and value proposition. Unusual home

• Deploy AI‑readable pages on a subdomain. Unusual AI pages

• One‑line install across CMSs. Integrations

• Plans and topic coverage. Pricing

• Talk to support (fast response). Contact support