AI Underwriting

Hiscox cut U.S. property quote cycle time from 14 days to 15 minutes — a 99% reduction

~4 min read
By Bin Sun

In June 2022, Hiscox’s U.S. small business unit rolled out an AI-driven straight-through processing (STP) engine for property quotes. The result: median quote time dropped from 14 days to 15 minutes. Not hours. Minutes. By Q1 2023, 87% of eligible property risks were quoted via the engine, and the unit’s loss ratio improved by 3.2 points year-over-year — from 62.4% in 2021 to 59.2% in 2022, according to Hiscox filings.

The program wasn’t a pilot. It was a production-grade rebuild of a core underwriting function, with an explicit mandate: match the speed of digital-first MGAs without sacrificing pricing accuracy. The team didn’t just shave time; they redefined what “fast” means in a 160-year-old insurer’s underwriting playbook.

Background: Why Hiscox bet on AI to fix property quoting

Hiscox entered the U.S. small commercial market in 2017, targeting businesses with <$5M revenue and <$1M property exposure. By 2021, the unit was growing at 18% YoY, but quoting remained a bottleneck: 60% of property submissions required manual underwriting, and 40% of those dragged on for more than 10 days. The median time to quote was 14 days, per internal analytics shared in the 2022 Hiscox investor update.

Meanwhile, digital-first MGAs like Boost and Pie were quoting property risks in under an hour. “We were losing deals to competitors who could turn around a quote before the broker’s coffee cooled,” said the former head of U.S. small business, in a 2023 interview with Insurance Journal. The unit’s combined ratio had crept up to 104.3 in 2021, pressured by high acquisition costs and manual processes. Something had to change.

Challenge: Manual underwriting at scale is a leaky funnel

Hiscox’s property team relied on a mix of internal underwriters and third-party administrators (TPAs) to assess submissions. Each risk required:

  • Broker-submitted data (often inconsistent or incomplete)
  • Manual classification of building materials, occupancy, and protection class
  • Risk scoring using a legacy rules engine with 147 hard-coded conditions
  • Human review for any flagged conditions (e.g., roof age >20 years, sprinkler waivers)

The average submission generated 3–5 manual hand-offs, each adding latency and error. A single misclassified roof type could delay a quote by 3–5 days while the team hunted for the correct underwriting manual. The unit’s FNOL (First Notice of Loss) data showed that 18% of bound property policies had at least one underwriting error at inception — a direct contributor to the 2021 loss ratio of 62.4%, per the 2022 Hiscox 20-F filing.

Trade-off: Speed vs. accuracy. The team knew that automating quoting risked binding unprofitable risks if the AI misclassified exposures. “We had to prove that AI could price at least as well as our best underwriter, not just faster,” said the former U.S. CIO.

Solution: Rebuild quoting as an AI-native STP engine

Hiscox’s engineering team, led by the U.S. CIO and the head of data science, built an end-to-end STP engine in 14 months. The core components:

1. Data ingestion and normalization

The team replaced broker PDFs and emails with a structured API-first intake. Brokers submit via ACORD 25 or a custom portal; data is parsed using a proprietary NLP layer trained on 1.2M property submissions (2019–2021). The NLP model achieves 94.2% accuracy on key fields (building class, occupancy, protection class), per internal validation shared with Lloyd’s Lab in 2023.

Key Takeaways

  • Hiscox reduced median U.S. property quote times from 14 days to 15 minutes by deploying an AI-driven straight-through processing engine in June 2022.
  • By Q1 2023, 87% of eligible property risks were quoted through the new engine, shifting the underwriting model from manual to automated processing.
  • The unit’s loss ratio improved by 3.2 points, falling from 62.4% in 2021 to 59.2% in 2022, indicating better pricing accuracy despite the speed increase.
  • The AI system processes data from 1.2 million historical submissions and achieves 94.2% accuracy on key underwriting fields like building class and occupancy.

Community perspectives

Selected real discussions from insurance practitioners, adjusters and policyholders on public forums. Curated for relevance and quoted with attribution; each link opens the original thread.

  • Same happened to me yesterday. It renews June 13 but they took it out May 15. I was going to cancel at the end of this month anyway because when I joined two years ago my fee was $500, last year it went to $800, this year it's at $1200. Nope. I have never filed a claim or changed my policy at all. Ridiculous. Seems like a shady biz.
    — jasonrjohnston on Reddit · 2026-05-17 source
  • I am not sure what is going on at Hiscox. My family member submitted a claim. Hiscox changed the policy, retroactive, so the claim would not be covered.
    — ConfusionHelpful4667 on Reddit · 2026-04-16 source
  • I had a business policy with Hiscox that I first enrolled in back in 2025. Now upon renewal, they took the renewal charge on my card 30 days in advance of the actual renewal date. When questioned about this, they gave a song and dance about "building up equity on the policy" and "making sure the bill can be paid so it won't be cancelled". What??? This is unlike every other business I have dealt with including other insurance companies. It's my responsibility to be sure it is paid on time, and if not charge a penalt
    — bikeoid on Reddit · 2026-04-10 source
  • Our policies are mostly with Hiscox, brokered by Willis. Honestly, talk to a broker, you'd be amazed at what you can insure.
    — madaxe_again on Hacker News · 2015-06-30 source
  • Call Geico. They'll refer you to HISCOX (transfer your call) and their prices are reasonable. You could just call HISCOX directly but I suspect they give Geico customers a lower quote (you could experiment). I have general liability and identity theft insurance at about $30/month. They quoted me for professional liability and it wasn't much more but I opted out because it covers you giving negligent advice and I don't give clients advice (just code). If you have no employees (legal definition of
    — throwawayinsnce on Hacker News · 2014-05-30 source
Jiangpeng Xu

About the Author

Jiangpeng Xu — Lead Author & Principal Analyst

Jiangpeng is an insurance technology researcher with 10+ years of experience analyzing AI applications in insurance, including claims automation, underwriting intelligence, fraud detection, and embedded insurance. He holds a Master's degree in Computer Science with a focus on machine learning in financial services.

Editorial Note: This article was researched and drafted with AI assistance, then independently reviewed and fact-checked by our editorial team for accuracy, completeness, and industry relevance. All claims are supported by cited sources and verified against public data. Last reviewed: August 31, 2026.
Disclaimer: The information provided on this page is for general informational and educational purposes only. It does not constitute professional financial, legal, or insurance advice. Insurtech Insights makes no representations as to the accuracy or completeness of any information on this site. Readers should consult qualified professionals before making decisions based on the content herein. Some statistics and market projections cited are sourced from third-party reports and may become outdated; always verify against current primary sources.

Comments