ACORD forms represent one of the genuine achievements in insurance standardization. The ability to submit a commercial auto risk using an ACORD 137 that any carrier's intake system can read, regardless of what policy admin system they're running, is real infrastructure. It reduces friction at scale. I spent three decades writing E&S risks before joining Undwrlyft's advisory team, and I have a complicated relationship with ACORD standardization — because the gaps are as instructive as the successes.
What ACORD solved — and where it was designed to operate
The ACORD library covers standard commercial lines well. The ACORD 125 (Commercial Insurance Application), 126 (Commercial General Liability), 140 (Property), 137 (Commercial Auto), and 130 (Workers' Compensation) form the backbone of most commercial lines submissions and give you a structured information set that supports straight-through processing.
For personal lines — auto, homeowners, renters — ACORD standardization is even more complete, which is one reason personal lines automated faster. The risk categories are well-defined, the required information fields are stable across carriers, and the form populations are consistent enough that template-based processing handles most of the submission intake load.
This is exactly where E&S begins to diverge. E&S, by regulatory definition, is non-admitted coverage written for risks that admitted market carriers have declined to take on standard forms. The premise of E&S is that the risk is non-standard. The information needed to underwrite it reflects that non-standardness.
The supplemental application problem
The most visible ACORD gap for specialty lines is the supplemental application. For a technology E&O submission, an ACORD 126 gets you general liability information that is largely irrelevant to the professional liability assessment. What the underwriter actually needs — technology revenue breakdown by product line, description of the software products and their deployment environments, indemnification and liability cap structure in customer contracts, incident response procedures, prior claims and circumstances — appears in carrier-specific supplemental applications.
Every specialty carrier has their own technology E&O supplemental. Some are three pages. Some are twelve. The field ordering differs. The question framing differs — one carrier asks "describe your quality assurance process," another asks "do you maintain a documented software development lifecycle?" about the same underlying risk characteristic. There is no ACORD standard for these fields because the insurance industry has not converged on a standard risk categorization for technology professional liability.
The same dynamic applies across specialty classes: environmental, professional liability, management liability (D&O, EPL, fiduciary), cyber, medical professional, architects and engineers E&O. For each of these classes, the ACORD form covers the minimum required structure, and then the actual underwriting information lives in supplementals that vary by carrier and sometimes by year.
What the broker actually submits
When a wholesale broker packages a specialty submission, they typically provide: the relevant ACORD form, their own cover letter or submission memorandum, the applicable supplemental applications (which may be the carrier's form, a broker house form, or the insured's own questionnaire responses), prior loss runs, and supporting documents depending on risk class.
The submission memorandum is particularly instructive as an artifact. It's a document the broker writes to package the narrative of the risk — describing the insured's operations, the coverage being sought, why this risk is being presented to this carrier, and any underwriting considerations the broker wants to surface proactively. There is no ACORD form for the submission memo. It's unstructured prose, and it often contains information that is critical to the underwriting decision but does not appear anywhere in the structured form sections.
For an automated intake system, the submission memo is where a lot of the interesting extraction work happens. An underwriter reading a memo will immediately pick up on things like "insured has a pending claim not yet reflected in loss runs" or "coverage was non-renewed by prior carrier due to underwriting concerns." That information has to be extracted and surfaced even though it's not in any ACORD field.
The NAIC code alignment problem
A structural gap that affects appetite rules enforcement is NAIC code alignment between what a broker populates on the ACORD form and what the underwriter needs for risk classification. NAIC codes on ACORD forms are often populated by brokers using the broadest applicable code rather than the most specific one — partly because brokers don't always have detailed NAIC knowledge, and partly because some brokers believe a broader code gives them more flexibility in carrier placement.
A commercial risk submitted under NAIC 23740 (Specialty Trade Contractors) might be a plumbing contractor, an HVAC installer, or an electrical subcontractor — risks with meaningfully different loss profiles under general liability and workers' comp. An underwriter who sees NAIC 23740 knows to look past it to the description of operations. An automated appetite filter that stops at NAIC code will either pass risks it shouldn't or decline risks it should quote.
We're not saying ACORD should fix this — NAIC codes are a taxonomy maintained separately from ACORD, and the granularity limitations are inherent to how NAIC classifies business activities. The point is that ACORD form data alone is insufficient for specialty lines triage, and any automated intake system that treats ACORD as a complete information source will under-perform on the risk classification that actually matters for appetite matching.
ISO class codes and the rating system interface
Related to NAIC codes but distinct is the ISO classification structure — the Commercial Lines Manual (CLM) classification codes that drive rating in commercial general liability and property lines. ISO class codes are more granular than NAIC codes for many risk types, but they're also carrier-specific in their application. An admitted carrier writing standard commercial GL may map their pricing to ISO class codes directly. An E&S carrier writing excess GL may use an entirely proprietary classification system that doesn't map cleanly to ISO.
This matters for automated pricing pre-fill because the field that populates the pricing model class code input isn't always obtainable from the ACORD form alone. The underwriter — or the intake system — has to infer it from the description of operations, the SIC or NAICS code, the prior carrier's classification, and any supplemental information about the insured's activities.
The implication for submission automation
Specialty lines submission automation has to treat ACORD forms as one input among several, not as the primary information source. The forms provide a useful structural starting point — confirming the risk type, capturing the named insured and policy period, establishing basic coverage parameters. But the information that drives the underwriting decision — the loss history detail, the exposure characteristics, the coverage terms being requested, the risk narrative — lives in the supplementals, the loss runs, and the submission memo.
After three decades of watching how specialty submissions actually flow from broker to carrier, what strikes me about modern extraction approaches is how much they benefit from having seen many examples of the same risk class. The variability is real, but it's not infinite. A system that has processed hundreds of technology E&O supplementals from different brokers develops a working model of where the revenue figures tend to appear, even when they're not in a standard field location. That pattern recognition is where the practical value of trained extraction systems over generic OCR comes from.
The ACORD gaps are a feature of specialty lines, not a bug. They reflect the genuine complexity of the risks that don't fit standard forms. Automation that acknowledges those gaps and works around them is more useful than automation that pretends ACORD covers what it doesn't.
