We asked 18 brokers in our early-access cohort to rank the features they wanted most from a commercial lines quoting tool. This was not a survey form with checkboxes. It was a structured conversation, conducted across three sessions in September 2025, where brokers were asked to compare features against each other and explain the tradeoffs they were making. Ranking exercises that force tradeoffs produce more honest results than rating scales where every option is "important."
Speed came first. That was not surprising. Brokers have been saying the same thing for years, and every technology vendor targeting this market knows it. What surprised us was the second answer.
What came in first: speed, defined precisely
When brokers said they wanted speed, they were specific about what they meant. They did not want a faster version of the existing two-day workflow. They wanted a result they could communicate to the client before the intake call ended. Those are different things. A 24-hour turnaround is faster than a 48-hour turnaround, but neither one changes the client experience in the way that a same-call result does.
The distinction matters for product design. If you are optimizing for same-day, you are building a faster back-office process. If you are optimizing for same-call, you are building a client-facing communication tool as much as a scoring tool. The output format, the language of the result, and the way it handles uncertainty all need to be designed for a broker who is reading it out loud to someone on the other end of the phone.
That framing changed several early decisions we made about output format. We had originally designed the result as a detailed technical scoring report. It was accurate and thorough, but no one was going to read it to a client on a call. The pivot to a plain-language decision summary at the top, with supporting detail available on request, came directly from this feedback.
What came in second: explainability
The second-ranked priority was being able to explain the quote to the client. Not explain the technology. Explain why this risk scored the way it did, why the premium landed in a particular range, and what a client could reasonably expect if they asked about it.
This surprised us, though it should not have. Brokers are fiduciaries of a kind. Their professional value is not just finding coverage, it is helping a client understand what they are buying and why it costs what it costs. A quote that arrives as a black-box number puts the broker in an uncomfortable position. They have a result but no narrative. If the client pushes back on the price, or asks how the number was derived, the broker has to say "the system said so," which is a weak foundation for a professional recommendation.
The explainability priority was almost universally connected to the first-call scenario. Brokers who said they wanted same-call results also said they needed to be able to talk through the decision immediately. The two priorities are linked: same-call speed only has value if the broker can do something with the result in real time, which requires enough context to hold a conversation.
What did not rank as high as expected
Integration with agency management systems came in third, which was lower than we expected. The brokers in our cohort were not dismissive of integration. Several mentioned that a standalone portal was a non-trivial workflow interruption. But when forced to rank it against speed and explainability, most said they would tolerate a two-tab workflow if the quality of the result justified it.
That is a meaningful finding for a product team. It suggests that the table stakes are not a native integration with Applied Epic or Vertafore. The table stakes are a result that is fast enough and clear enough that a broker will work around the integration friction to get it. Integration matters for long-term adoption and stickiness. But it is not the first thing that determines whether a broker will try the tool at all.
Coverage breadth also ranked lower than we anticipated among this cohort. Most of the early-access brokers were handling primarily BOP and general liability, with some commercial auto. They were not asking for a tool that covered every specialty line from professional liability to ocean marine on day one. They wanted a tool that handled the high-volume commercial class applications extremely well, and they were skeptical of platforms that claimed to cover everything but did none of it particularly well.
What some brokers said that did not fit the ranking exercise
Several conversations produced comments that did not map cleanly onto the feature ranking but were informative about how brokers think about technology tools in general.
One broker said, "I don't want a tool that tells me what to do. I want a tool that tells me what the risk looks like so I can decide what to do." That is a clean articulation of the principle Attune is built on. The scoring model is not a recommendation engine. It is a structured interpretation of the application data. The broker makes the submission decision.
Another broker raised something we continue to think carefully about: what happens when the tool is wrong? "If I trust a fast answer and the carrier comes back with a decline, I've made a promise I can't keep to my client. I'd rather be slower and right." This is a reasonable concern, and it is why the decision rationale exists as an explicit component of the output. The goal is not that the broker trusts the tool blindly. The goal is that the broker understands what the tool is saying well enough to apply their own judgment to it.
How this shaped the product decisions we made
Two specific design decisions changed based on the research. The first was the placement of the decision rationale block. It was originally buried at the bottom of the result report as a supporting detail section. After the broker conversations, we moved it to the top, directly below the risk score, so the first thing a broker reads after the score is the plain-language summary of what drove it. The technical detail is still there, but the actionable narrative comes first.
The second change was to the uncertainty language. When the scoring model cannot produce a narrow confidence interval on a given application (typically due to unusual class combinations or thin application data), the original output flagged this with technical language about confidence intervals and standard deviations. Brokers do not talk about standard deviations to clients. We rewrote the uncertainty flags as decision guidance: "This application has characteristics that require closer review before submission" or "We recommend confirming current carrier guidelines for this class before submitting." That is actionable. Standard deviation ranges are not.
The broader lesson from these sessions is one that applies to any B2B tool built for practitioners rather than administrators. The features that rank highest in practitioner research are almost never about the technology. They are about what the technology enables the practitioner to do in front of a client. Building for the broker-client interaction, not for the broker's back-office workflow, is where quoting technology has the most leverage in this market.
