Neptune Insurance

Neptune's public disclosures make it possible to separate three measures that are often blurred in discussion of digital MGAs: premium placed, revenue earned by the intermediary and underwriting results delivered to capacity providers. In the second quarter of 2026, the company reported $126.9 million of written premium and $55.9 million of revenue. Its adjusted EBITDA margin was approximately 62%, compared with a GAAP net-income margin of 28%. The adjusted measure describes Neptune's operating economics; it is not an insurer's underwriting margin or a substitute for GAAP earnings.

The company says it does not take insurance balance-sheet risk. That structure does not make its business independent of catastrophe performance. Our assessment is that losses can affect an MGA indirectly through the terms, availability and economics of the capacity on which it depends. The relevant test is therefore not only whether the platform remains profitable in a quarter, but whether the insurers and reinsurers supporting the business continue to find the portfolios attractive after adverse experience.

Neptune reported 45 capacity providers across eight programs at June 2026 and said its two largest programs had renewed on improved economic terms. Provider count is useful context, but it cannot establish how independently those providers make decisions or how much exposure sits with each. A meaningful capacity review would examine common issuing carriers, geographic accumulations, renewal timing and dependencies on similar hazard assumptions. Those are diligence questions, not findings that Neptune's panel is inadequately diversified.

Retention also requires attention to its denominator. Neptune defines policy retention as acceptance among policyholders who receive renewal offers. That is not necessarily the proportion of the entire expiring book that remains insured, because non-offered renewals are outside that calculation. In our view, readers should pair the reported measure with information on nonrenewals and changes in price or coverage before drawing conclusions about customer loyalty or portfolio stability. The distinction matters particularly when an automated underwriting system changes eligibility.

The company-reported lifetime written loss ratio of 19.5% provides historical context, not a forecast for the next storm season. An analytical review would examine individual event years, geographic mix and loss development rather than extrapolate a single cumulative result. The same discipline applies to technology claims. Faster quoting and frequent model releases are operational capabilities; demonstrating durable selection requires evidence that model changes improve outcomes without introducing unrecognized concentrations. Neptune's disclosures offer a useful starting point, but the decisive questions remain how capacity, customer retention and underwriting performance respond together when conditions are less favorable.

Neptune’s March 12, 2026 conversational-quoting announcement illustrates a separate distribution boundary. The company described preliminary quotes within ChatGPT, followed by a transition to Neptune’s website to complete the purchase. Its disclaimer states that the conversational quote is neither a binder nor an offer or guarantee of coverage; eligibility, underwriting review and policy conditions still apply. It also states that ChatGPT is not Neptune’s insurance agent, broker or representative. This is the company’s announced workflow, not an independent end-to-end test.

MGA Index analysis: an additional interface can expand access without transferring underwriting authority to that interface. The useful operating measures are the accuracy of information carried into the purchase workflow, differences between preliminary and final terms, and how clearly the customer can identify when coverage actually takes effect. A fast estimate and an issued policy are different outcomes.