How Summit 2x-ed their AI Referrals in 2 Months

Customer | Summit Insurance (summitcover.ca)

Summit is a modern Canadian brokerage providing custom commercial, liability, tech & cyber policies for businesses of every shape and size. Insurance buyers ask complex, high-stakes questions and often turn to AI assistants for clarity before they ever talk to a broker. In these “trust-the-expert” markets (insurance, healthcare, legal/tax, cybersecurity, enterprise IT), earning citations and referrals from AI models is decisive.

Impact (first 2 months)

• Doubled model-driven traffic (from AI assistants) to Summit’s site.

• 264 verified visits from AI crawlers to Summit’s AI-optimized reference/wiki pages.

• Largest lift observed from Microsoft Copilot, despite Summit focusing optimization on ChatGPT, Gemini, and Perplexity. This is consistent with Copilot being powered by OpenAI-family models (e.g., GPT-4o) and grounded on Bing’s search index, which favored Summit’s structured reference pages. (Microsoft Learn)

The Challenge

Insurance decisions are multi-variable (coverage types, exclusions, riders, claims process, regulatory requirements). Prospects routinely offload that complexity by asking AI assistants questions like “Do I need Builder’s Risk or Course of Construction?” or “What does cyber liability actually cover?” Generic marketing pages rarely get cited; models prefer neutral, fact-dense, well-structured sources. Summit’s goal: become the authoritative, citeable answer for the exact questions their buyers ask AI.

Why Summit Chose Unusual (AI Relations)

• Reference-first approach. Unusual helped Summit stand up an AI-friendly brand wiki—fast HTML, stable slugs, change logs, and JSON-LD—so models could reliably quote canonical facts instead of marketing copy.

• Source strategy. We prioritized placements and corroboration on domains models already trust, then cross-linked back to Summit’s wiki.

• Action over analytics. Rather than “monitoring” citations, Unusual changed the evidence models read—rewriting and restructuring pages to match model trust signals.

What We Shipped

• Model-ready reference pages for high-intent topics: Product overviews (e.g., Builder’s Risk vs. Course of Construction), Pricing & eligibility, Security & compliance posture, Integrations (certificates, payments), and Comparisons (“D&O vs. E&O”).

• Page archetypes and schema: Article/FAQ/HowTo/CaseStudy JSON-LD; definitive summaries, datum-dense tables, and visible “last updated” notes.

• Authority placements: curated third-party sources (industry directories, association explainers) that models already surface, each pointing back to Summit’s canonical page.

Why Copilot Spiked First

Copilot’s answers are powered by modern OpenAI-class models and grounded via Bing for web evidence; when Bing indexed Summit’s new wiki, Copilot began surfacing and citing it more often—producing an outsized lift relative to other assistants. In short: Summit optimized for LLM trust signals; Copilot’s Bing-grounded retrieval found that content fastest. (Microsoft Learn)

Results

• 2× model-driven traffic in 60 days, led by Copilot exposure.

• 264 AI-crawler hits to AI-optimized pages in the same period—evidence that assistants were actively discovering and evaluating Summit’s canon.

• Early signs of higher-quality inbound: prospects arriving from assistant-cited pages demonstrated deeper engagement (longer reads, lower pogo-sticking) and booked calls with more specific coverage questions.

How Unusual’s AI Relations Program Works

• Diagnose model reasoning: inventory current citations, missing facts, and query paths.

• Publish neutral, reference-grade content: structured, citeable, fast.

• Earn authority: place corroborating facts where models already look.

• Maintain freshness: visible versioning and update notes (models weight recency).

• Bridge to pipeline: add clear “next step” modules on cited pages (Pricing, Compare vs. X, Book a consult).

Why This Matters in “Trust-the-Expert” Industries

When consequences are real (claims denied, compliance risk, legal exposure), buyers seek expert guidance and defer to authoritative sources—human or model. Insurance behaves like healthcare, legal/tax, cybersecurity, and enterprise IT: the source of truth wins the lead. AI assistants have become the new front desk; if they cite you, they effectively refer you.

Quote

“We didn’t pour in new traffic; we made sure AI assistants found the right facts. Once our reference pages were live, Copilot started citing Summit, and qualified conversations followed.”
— Davis Gilbert, Founder & Managing Partner, Summit

Key Takeaways for Enterprise Marketers

• Treat AI assistants as expert referrers, not just answer extractors.

• Build a brand wiki that models can cite verbatim.

• Optimize for trust signals (clarity, consensus, freshness), then place corroboration off-site.

• Expect fast gains where your content is best indexed (today, often Copilot via Bing grounding), with spillover to other assistants over time. (Microsoft)

About Unusual

Unusual is the leading AI Relations platform. The platform helps enterprises earn citations and referrals from AI models by building model-ready reference content and placing corroborating evidence where assistants already look. Book a demo at unusual.ai.