How We Track the Commercial Insurance Customer Experience

Every business owner who completes a quote on Get Business Coverage produces some signal about the commercial-insurance buying experience. The most useful signals come from a small set of places: post-quote NPS responses, the points in our 14-step funnel where operators abandon, the support emails and chat conversations our team has with applicants, and the patterns in carrier-decline responses that get fed back to us.

This page documents how Get Business Coverage observes that experience, how we turn it into vertical-by-vertical patterns we publish in our recurring «What we hear from operators» series, and what we deliberately keep out of public reporting.

What we collect

CX signals across four sources:

• Post-quote NPS surveys. One-question Net Promoter Score sent after every completed quote. Optional free-text comment box. Voluntary response.
• Form-abandonment events. When an applicant stops the 14-step quote funnel mid-flow, we record which step + the elapsed time + the vertical + the state. We do NOT record what they typed.
• Support transcripts. Email + chat threads from applicants to our customer-service team. Categorized at intake by a structured taxonomy (eligibility question, premium concern, coverage-mechanism question, policy-document request, escalation, etc.) and by vertical (NAICS).
• Carrier decline signals. When a carrier partner declines to quote, we capture the reason code if provided. These get aggregated into «why operators in vertical X struggle to find quotes in state Y» patterns.

How we aggregate patterns

Raw CX events become publishable patterns through three steps:

1. Vertical bucketing. Every event carries the applicant's NAICS code or vertical category. Patterns are reported within a vertical (or vertical cluster), not across the whole platform.
2. Stage bucketing. Events are tagged by funnel stage: pre-quote (browsing /learn or /compare), in-quote (steps 1–14 of the application), post-quote (NPS + carrier-decline window). Patterns reported at the stage level.
3. Pattern coding. Support transcripts are read by a member of our CS&O team and tagged with a structured code (e.g., «needs clarification on COI requirements», «asks why GL doesn't cover X», «wants payment-plan options»). We publish frequency-of-code patterns once a code has appeared at least 20 times in a quarter.

Every published CX observation reports: (1) the vertical or cluster, (2) the stage, (3) the calendar quarter window, (4) the sample size n. A statement like «In 2026-Q1, 31% of restaurant operators in the post-quote stage asked specifically about liquor liability inclusion (n=47, GBC support transcripts)» is fully bounded by what it claims.

How we protect operator privacy

Operators talking to our team should never end up identifiable in anything we publish. Specifically:

• No applicant name, business name, DBA, email, phone, or location appears in any aggregate or any quoted transcript.
• No verbatim quote from a support transcript is published. Patterns are described in our own words; recurring framings are paraphrased.
• Minimum cluster size for publication is n=20 within a quarter. Patterns below that threshold get rolled up to a broader vertical bucket or held for the next quarter.
• NPS free-text comments are never quoted. Only the score distribution + structured pattern codes get published.
• Geographic granularity stops at state. We do not publish CX patterns by city or ZIP.

What we explicitly do NOT publish

Categories of CX detail kept strictly internal:

• Individual NPS verbatim comments. We track them for internal product improvement; we never quote them externally.
• Identifiable complaints. A pattern like «operators frustrated by X» might appear in a published piece; the specific complaint of any specific operator never does.
• Support-agent identities. We talk about «our team» in aggregate; named CS agents are not surfaced.
• Carrier-identifying decline reasons. Patterns can be reported (e.g., «driving-record-related declines spiked in commercial-auto Q1») but never tied to a specific carrier in a way that could be read as a public review of that carrier's underwriting.
• Coverage advice. Paige is not a licensed insurance agent. Anywhere a CX observation could be read as advising on coverage selection or pricing, the language is changed by editor or removed by the reviewing agent (Jason Wootton, CA P&C #0I94454).

How operators submit feedback

If you completed a quote on Get Business Coverage and want to share a pattern that was not surfaced by the NPS or support intake, contact our CS&O team at hello@getbusinesscoverage.com. Note in the subject line that you'd like the message to inform our CX series; we will not include identifying details, but it helps to know that's how you intend the feedback to be used.

Journalists or trade-association researchers studying small-business commercial-insurance buyer behavior who want access to a specific vertical's aggregated CX pattern beyond what we publish can reach editorial@getbusinesscoverage.com.

How we made this article

• Drafted by Paige Tyrrell, Customer Service & Operations.
• Edited by Justin Marks, Founder & Editor. (Not a licensed insurance agent.)
• Reviewed for regulatory accuracy by Jason Wootton, California-licensed P&C insurance agent (CA #0I94454). Verify license ↑
• Last updated May 30, 2026. We refresh this methodology when collection, coding, or anonymization practices change.

See our editorial methodology for our broader source policy, and the marketplace data methodology (Dave Benson) for how we report on premium and carrier patterns.