CNA AI Market Strategy Report - Professional Liability Insurance

Answer Capsule

CNA holds a visible but under-recommended position in the September 2026 Professional Liability Insurance benchmark. The brand appeared in 13.30% of qualified AI answers but earned valid recommendation coverage of only 6.44%, meaning it was named far more often than it was shortlisted. Its clearest win is a positive framing profile with a net sentiment score of 0.6129 and zero negative mentions. Its clearest weakness is recommendation depth: a 0.43% top-three rate and a 0.00% rank-one rate across 466 qualified observations. The clearest opportunity is converting existing mentions into shortlist placement in the Brand Recommendation cluster, where every qualified observation in the benchmark currently sits.

Who This Report Is For

This report is written for CNA marketing, brand, and distribution leaders, and for professional liability insurance executives who need to understand how their brand is positioned at the moment AI systems form a buyer shortlist.

Report Card

Executive Summary

CNA is present in the Professional Liability Insurance AI answer set but is not being recommended at scale. The September 2026 benchmark recorded 62 mentions across 466 qualified observations, a raw mention presence rate of 13.30%. Valid recommendation coverage was 6.44%, drawn from 30 valid recommendations. The gap between being named and being shortlisted is the central finding for this brand.

The framing around CNA is positive. The benchmark recorded 38 positive mentions, 24 neutral mentions, and zero negative mentions, producing a net sentiment score of 0.6129. That is the lowest net sentiment score among the ten tracked brands, but it reflects a higher neutral share rather than any negative framing. CNA is described accurately and without caution in the answers the benchmark observed.

Recommendation placement is where CNA separates from the category leaders. CNA earned 2 top-three placements in September 2026, a top-three rate of 0.43%, and zero rank-one placements. The Hartford, by comparison, recorded a 53.43% top-three rate and a 34.12% rank-one rate. Next Insurance recorded 40.56% and 15.88%. CNA is not competing in the same placement tier as the category leaders.

The strongest cluster signal for CNA is the Brand Recommendation cluster, which is also the only cluster with qualified observations in the September 2026 public series. All 466 qualified observations fell into this class. The Pricing and Value and Multi-Brand Comparison clusters were defined but empty, so the benchmark cannot yet show how CNA is positioned on cost or in direct comparison prompts.

The strongest platform signal for CNA is ChatGPT, where the brand recorded a 10.71% valid recommendation coverage rate and 6 valid recommendations. The weakest platform signal is Copilot, where CNA recorded 5 valid recommendations at an 8.20% coverage rate and no top-three placements. Gemini produced the lowest coverage at 3.33%.

The clearest gap is recommendation conversion. CNA is mentioned in roughly one in eight qualified answers but appears in a valid shortlist in fewer than one in fifteen. The brand is visible as context, not as a recommendation. Closing that gap is the strategic priority.

What CNA Is Winning

Questions This Section Answers

• Where is CNA's sourcing quality strongest in Professional Liability Insurance?
• Which buyers should treat CNA as a viable Professional Liability Insurance option today?
• What does CNA's citation or authority profile look like in the AI answer set?

CNA's clearest win is framing quality. Across 62 mentions, the benchmark recorded zero negative mentions and a net sentiment score of 0.6129. No tracked brand recorded a negative mention in September 2026, but CNA's profile is notable because it is entirely positive or neutral. There is no cautionary framing to correct.

CNA also recorded a modest improvement in valid recommendation coverage against the July 2026 baseline. Coverage moved from 4.50% in July 2026 to 6.40% in September 2026, an increase of 1.90 percentage points. That is one of only three positive movements in the benchmark's coverage table, alongside the two new tracking entities.

The brand's strongest platform is ChatGPT, where CNA recorded a 10.71% valid recommendation coverage rate, 6 valid recommendations, and a 0.7692 net sentiment score. ChatGPT is also the platform where CNA's positive visibility rate was highest at 17.86%. This is a narrow but real recommendation pocket.

CNA's presence rate also rose within the benchmark's normal range. The brand is being named in AI answers more often than it was at the July baseline, even as its shortlist conversion remains low.

Where CNA Has the Clearest AI Visibility Gaps

Questions This Section Answers

• How large is the gap between CNA's mention presence and its valid recommendation coverage?
• On which AI platforms is CNA's recommendation coverage weakest?
• Where do competitors take the shortlist placement when CNA is mentioned but not recommended?

CNA's largest gap is the distance between presence and recommendation. The brand appeared in 13.30% of qualified observations but earned valid recommendation coverage of only 6.44%. That means roughly half of the answers that mention CNA do not place it in a recommendation shortlist. The brand is being used as reference material rather than as a candidate.

The placement gap is sharper still. CNA recorded 2 top-three placements across 466 qualified observations, a top-three rate of 0.43%. The Hartford recorded 249 top-three placements. Next Insurance recorded 189. Hiscox Usa, in its first month of tracking, recorded 102. CNA is not appearing in the first three positions of AI recommendation lists at any meaningful rate.

Rank-one placement is effectively absent. CNA recorded zero rank-one placements in September 2026, unchanged from the July 2026 baseline. The Hartford recorded 159, Next Insurance recorded 74, and even Embroker, which recorded only 3 valid recommendations in total, did not register a rank-one placement. CNA is in a group of brands that are named but never placed first.

Platform coverage is uneven. CNA's strongest platform, ChatGPT, produced a 10.71% coverage rate. Copilot produced 8.20%, Perplexity 6.25%, Google AI Overviews 5.83%, Google AI Mode 5.79%, and Gemini 3.33%. The brand is not absent from any tracked platform, but it is under-represented on the surfaces with the largest observation counts. Google AI Mode carried 121 qualified observations and Google AI Overviews carried 120, yet CNA's coverage on those surfaces was 5.79% and 5.83% respectively.

The competitive picture is concentrated. The Hartford, Next Insurance, and Hiscox Usa hold the recommendation-stage strength in this category. CNA sits in the lower tier alongside Travelers, Chubb, and Biberk Business Insurance, all of which show presence without proportional recommendation conversion.

Biggest Opportunity

CNA's single clearest opportunity is converting existing mentions into valid shortlist placement in the Brand Recommendation cluster. The brand is already named in 13.30% of qualified answers, which means the retrieval layer is working. The recommendation layer is not. The gap between 13.30% presence and 6.44% valid recommendation coverage represents the most direct path to improved position, and it does not require new visibility, only better recommendation conversion from answers that already reference the brand.

Competitive Landscape

Questions This Section Answers

• Who leads Professional Liability Insurance AI recommendations, and by how much?
• How does CNA's top-three and rank-one rate compare to the rest of the tracked set?

The Hartford holds dominant recommendation power in Professional Liability Insurance, with Next Insurance as the strongest challenger and Hiscox Usa as the strongest new entrant. CNA sits in the lower tier of the tracked set, with presence that is not converting into shortlist placement.

Average recommended rank covers rank-eligible recommendations only.

CNA ranks ninth of ten on top-three rate and ninth on rank-one rate, ahead of Embroker only. Its average recommended rank of 5.23 is the second-lowest in the set, meaning that when CNA does receive rank credit, it appears near the bottom of the list. The table shows a brand with accurate framing and weak placement.

Prompt Evidence

Questions This Section Answers

• Which prompts produced a valid CNA recommendation, and which produced only a mention?
• What did CNA's strongest platform result look like at the prompt level?

ChatGPT / Brand Recommendation Prompt: "best small business insurance" Result: CNA was mentioned and received valid recommendation credit, contributing to its strongest platform coverage rate of 10.71%.

Google AI Mode / Brand Recommendation Prompt: "What are the top 20 insurance companies in the US?" Result: CNA appeared in the answer set but did not register a top-three placement, consistent with its 5.79% coverage rate on this surface.

Google AI Overviews / Brand Recommendation Prompt: "business liability insurance" Result: CNA was named as a reference point without shortlist placement, reflecting the pattern of presence without recommendation conversion.

Perplexity / Brand Recommendation Prompt: "Who has the cheapest commercial insurance?" Result: CNA received valid recommendation credit on Perplexity, where it recorded a 6.25% coverage rate and 3 valid recommendations.

What CiteWorks Studio Would Do Next

Questions This Section Answers

• Which prompts and surfaces should CNA prioritize to convert mentions into shortlist placement?
• What does the recommendation readiness plan focus on for CNA?

Phase 1: AI Market Discovery Audit Map every prompt where CNA is mentioned but not shortlisted, and identify which competitors take the placement instead.

Phase 2: Recommendation Readiness Plan Prioritize the Brand Recommendation cluster and the Google AI Mode and Google AI Overviews surfaces, where observation volume is highest and CNA coverage is lowest.

Phase 3: Owned Answer Layer Buildout Strengthen the pages and content assets that answer the specific prompts where CNA is named as context rather than recommended.

Phase 4: Citation and Authority Layer Development Build the public evidence layer that AI systems retrieve from, so that CNA's credentials, coverage strengths, and category fit are easy to find and synthesize.

Phase 5: Monthly AI Visibility and Recommendation Tracking Track valid recommendation coverage, top-three rate, and rank-one rate month over month to confirm whether presence is converting into placement.

Why This Matters

AI systems are forming the buyer shortlist before a buyer ever visits a carrier site. CNA is being named in those answers, which means the brand is not invisible. But being named is not the same as being recommended. A buyer who asks an AI system which professional liability insurer to consider will see a shortlist, and CNA is currently appearing in that shortlist in fewer than one in fifteen qualified answers.

The next move is not more visibility. It is targeted correction of the prompt, page, and citation layers so that the answers already mentioning CNA place it in the recommendation set. That is a narrower and more achievable problem than building awareness from zero, and it is the difference between being referenced and being chosen.

Core Metrics

Sentiment Score

Sentiment Score = (positive mentions × 1 + neutral mentions × 0 + negative mentions × -1) / total mentions

For CNA in September 2026: (38 × 1 + 24 × 0 + 0 × -1) / 62 = 0.6129.

This matters because unclassified mention counts are misleading. A brand that appears in 62 answers sounds healthy until the mentions are separated into positive recommendations, neutral references, cautionary notes, and competitor-displaced placements. CNA's 62 mentions break down into 38 positive and 24 neutral, with no negative framing. That is a clean profile, but it is also a profile with a large neutral share, which typically indicates the brand is being listed as context rather than endorsed as a choice.

Share of voice is a diagnostic metric, not a business KPI. A positive recommendation, a neutral reference, a cautionary mention, and a mention that only appears because a competitor was recommended are not equal events. Counting all mentions as wins is bad measurement. Classified sentiment is required before interpreting AI visibility, because it is the difference between being talked about and being chosen.

Sentiment by Platform

Methodology

1. This report is a benchmark-based analysis of CNA's position in the September 2026 Professional Liability Insurance AI Market Discovery Index. It is not a client result and does not describe work performed by CiteWorks Studio.
2. The reporting window is September 2026. The benchmark also carries July 2026 and August 2026 measurements for comparison.
3. Six AI and search surface families were tracked: ChatGPT, Copilot, Gemini, Perplexity, Google AI Overviews, and Google AI Mode.
4. The September 2026 collection began with 800 source prompt-surface observations and produced 466 qualified observations after qualification. Brand-level percentages use the 466 qualified observations as the public denominator.
5. Ten brands were tracked in the September 2026 set: The Hartford, Next Insurance, Hiscox Usa, Thimble, Progressive Commercial, Chubb, Travelers, Biberk Business Insurance, CNA, and Embroker.
6. All 466 qualified observations fell into the Brand Recommendation buyer-intent class. The Pricing and Value and Multi-Brand Comparison classes were defined but contained no qualified observations in this measurement.
7. Stage 0 extraction retained the query, AI or search surface, answer, brand outcome, recommendation placement, sentiment, and, where exposed, citations or attributable evidence sources.
8. A mention is counted when a tracked brand appears anywhere in a qualified AI answer, regardless of placement or framing.
9. A valid recommendation is counted when a brand appears in a recommendation shortlist with positive framing and rank eligibility between 1 and 10. Neutral, cautionary, comparison-anchor, and listed-only mentions are not counted as valid recommendations.
10. The benchmark does not measure market share, sales, revenue, organic search ranking, social mention volume, or causality. Month-over-month movement identifies changes worth investigating and does not establish the cause of those changes.
11. A naming shift affected two tracked entities in September 2026. Hiscox and biBERK, tracked in July and August, did not appear in the September tracking universe, while Hiscox Usa and Biberk Business Insurance appeared in their place. This is consistent with a change in how AI systems name these providers rather than a change in underlying visibility.
12. CNA's September 2026 rates are drawn from a qualified observation base of 466. Small-count movements should be read as directional signals rather than settled trends.

See Where AI Is Recommending Your Brand

The public benchmark shows where CNA stands in the category. A company-level AI visibility audit shows which specific prompts CNA is winning, which competitors take the recommendation when CNA loses, what attributes AI systems associate with the brand, and which external sources shape those answers. That is the layer beneath the percentages, and it is where the next move gets decided.