Coalition is winning the commercial-cyber LLM surface that Chubb owned eight days ago.

TL;DR

• On 2026-04-26, Coalition won 0 of 5 commercial-cyber verticals cleanly cross-LLM. Chubb won at least one LLM in all 5.
• On 2026-05-04, Coalition wins fintech-startup (Perplexity 5/5 + Gemini 3/5), healthcare-SMB (Perplexity 3/5 + Gemini 3/5), and SaaS-startup (Gemini 5/5 alongside Chubb 4/5 on Perplexity).
• Chubb still wins law-firm (4/5 + 5/5) and manufacturer (2/5 + 5/5) — the latter two industries where established carrier relationships and AM-Best ratings dominate the comparison-site editorial.
• The shift happened across both Perplexity and Gemini in 8 days. That's not a single-LLM editorial accident; that's the citation graph re-weighting.
• The round-2 framing needs revising: in commercial cyber, specialty insurtechs are recoverable from "specialist alone doesn't surface" once their editorial surface saturates a vertical. Coalition got there in fintech and healthcare before Chubb's editorial inertia could push back.

What we measured

Phidea’s observation tool re-ran the commercial-cyber probe with the same configuration on 2026-04-26 and 2026-05-04. Five queries, five runs per LLM, two LLMs (Perplexity Sonar Pro + Gemini 2.5 Pro with grounding). The exact prompts:

• "What's the best cyber insurance for a 50-person SaaS startup?"
• "What's the best cyber insurance for a mid-market law firm?"
• "What's the best cyber insurance for a healthcare SMB?"
• "What's the best cyber insurance for a fintech startup?"
• "What's the best cyber insurance for a manufacturer with 200 employees?"

The diff:

Three of five verticals now lean Coalition on at least one LLM with high stability. Eight days. Same probe configuration.

Why this matters

The round-2 framing was a confident-sounding generalisation: the personal-lines pattern (generalists with editorial depth beat specialty insurtechs) generalises to B2B cyber. We went on to write that Coalition, At-Bay, and Resilience needed to invest in editorial presence to recover the surface.

Eight days later, Coalition has done at least part of that work — or had already done it, and the citation graph has caught up. Either reading should change how Phidea readers think about the timeline of LLM-retrieval shifts:

• It happens fast. A vertical that "belonged to Chubb" on April 26 is a contested or Coalition-owned vertical on May 4. Shipping cycles in this space are weeks, not quarters.
• The shift is uneven across verticals. Coalition is winning fintech and healthcare-SMB. Coalition is NOT (yet) winning law-firm or manufacturer. Those are the verticals where Chubb's brand-incumbency, AM Best AAA rating, and broker-channel saturation matter more in the editorial.
• The change reads like it followed a content investment. A look at the citation hosts captured by today's probe shows new sites surfacing in the cyber-vertical answers: insurtechdigital.com, seedpodcyber.com, foundershield.com — these are the long-tail editorial surfaces that Coalition has presence on and Chubb doesn't to the same degree.

What changed in 8 days, hypothesis

We can't prove causation from probe data alone, but here’s the most plausible reading:

1. Coalition's owned content compounded a threshold. Coalition has been publishing cyber-vertical case studies and threat-intel reports for several years. The volume and variety crossed a citation-graph threshold where comparison-aggregation engines (vertexaisearch, comparison sites) started preferring Coalition's coverage over Chubb's general cyber pages.

2. Chubb's general cyber pages got out-flanked on vertical specificity. Chubb's cyber pages are excellent in general — the law-firm and manufacturer wins confirm that. But on fintech and healthcare-SMB, Coalition's vertical-specific pages (with named regulatory hooks like SOX, HIPAA Privacy Rule, NYDFS) read as more directly relevant to the prompt's specificity.

3. New editorial coverage from cyber trade press. Sites like insurtechdigital.com, seedpodcyber.com, foundershield.com are showing up in citation hosts for the first time. They tend to feature insurtechs (Coalition, At-Bay, Resilience) more than traditional carriers. As trade-press citation share rises, Coalition's surface rises with it.

The third hypothesis is the most operationally important: the citation graph has a long tail that is editorially aligned with insurtechs, and as that tail enters LLM grounding more reliably, the insurtechs win more verticals.

What this means for B2B cyber competitors

For the four large insurtech-shaped commercial cyber writers — Coalition, At-Bay, Resilience, Cowbell — and the analytic player CyberCube:

1. Vertical-specific editorial pays back fast in LLM retrieval. Coalition's lead is in two verticals (fintech, healthcare) where it has visible editorial depth. Other vertical-specific surfaces (manufacturing, legal-services, professional-services, hospitality, real estate) are still up for grabs. Whichever insurtech publishes vertical-specific pages first, with named regulatory and threat-model context, can take those.

2. Chubb's brand-incumbency is real in two verticals. Law-firm cyber and manufacturer cyber didn't shift. The pattern there: AM Best ratings, capacity floors, claims-handling track record matter more in the editorial than vertical specificity. Insurtechs targeting these segments will need a different angle than vertical pages alone.

3. Trade-press relationships compound differently. The new citation hosts in this probe (insurtechdigital, seedpodcyber, foundershield) are insurtech-aligned. A Coalition press placement on a cyber-trade-publication is now visibly more LLM-leveraged than the same placement was 6 months ago.

For Chubb specifically:

4. Defending the law-firm and manufacturer verticals matters more than re-fighting fintech. Chubb's editorial surface is strongest where AAA-rating + claims-handling-history + broker-channel-saturation lead the editorial. Doubling down on those characteristics in those verticals is the higher-return investment than trying to out-publish Coalition on fintech-specific content.

The operational takeaway

Phidea's observation tool is useful precisely because it tracks shifts like this. A finding that was round-2 confident eight days ago is now visibly weaker. The recommendation that flowed from the round-2 finding ("specialty insurtechs lose the editorial surface to generalists with depth") was directionally correct then but is empirically wrong now for fintech and healthcare-SMB cyber.

Three things follow:

• Re-test confident findings on a 1-2 week cadence when they're directly informing strategy. The 5-day time-stability retest of the multi-query observations and the 10-day validated-study retest both surfaced material drift; this commercial-cyber drift surfaced in the same window.
• Treat citation-host shifts as a leading indicator of carrier-placement shifts. New trade-press hosts entering the citation graph for a vertical predicts which carriers gain placement next.
• Don't write B2B-LLM-retrieval recommendations as universal. The pattern that holds in personal lines (specialists lose to generalists) does NOT hold uniformly in commercial cyber. It holds in some verticals (law firm, manufacturer) and breaks down in others (fintech, healthcare-SMB).

The full citation hosts captured for each cyber vertical are on the observation tool detail pages. The raw probe data lives at data/probe-commercial-commercial-cyber-by-vertical-2026-05-04.json in the Phidea repository.