This article is by Gary Huddleston, founder of DIBNI
Commercial insurance has no shortage of technology. Across the market, firms have invested in portals, workflow systems, artificial intelligence, API connections and automated processing. Yet many of the frustrations experienced by clients, brokers and underwriters remain surprisingly familiar: repeated questions, inconsistent information, missing context and unnecessary delay.
That suggests the underlying problem is not simply a lack of automation.
In many cases, the more fundamental weakness sits earlier in the process. Risk information still arrives in forms that are difficult to compare, reuse or interpret consistently. A submission may contain the right facts, but those facts can be spread across emails, proposal forms, attachments, spreadsheets and free-text explanations.
Before anyone can make a professional decision, somebody often has to reconstruct the story of the business.
Automating that process without improving the information model can simply move the same ambiguity faster.
Better automation starts with better inputs
Commercial insurance depends on context.
Two businesses with the same broad trade description can present very different exposures because of their activities, turnover, locations, contracts, processes, customers or loss history. The quality of the decision therefore depends heavily on the quality and structure of the information received.
If the baseline data is incomplete, inconsistent or poorly categorised, automation has limited value. A faster workflow does not necessarily produce a better underwriting picture. It may simply accelerate the need for clarification later in the journey.
This is especially important where pricing risk is concerned. An automated process can only work reliably with the information it has been given. If key facts are absent, buried in narrative or expressed differently each time, the technology cannot create certainty that was never present in the input.
The practical lesson is straightforward: digitisation should begin with the information architecture, not the automation layer.
Structure does not mean removing judgement
There is sometimes a concern that more structured data means forcing every commercial risk into a rigid template.
That is not the objective.
The purpose of structure is to make the common elements of a proposal clear, reusable and easier to pass between parties. Core information — the identity of the business, its activities, scale, locations and relevant operational characteristics — should not need to be repeatedly rediscovered every time the proposal moves from one professional to another.
Good structure creates a foundation.
Professional judgement then sits on top of it.
Commercial insurance contains nuance that cannot always be reduced to a dropdown menu. Brokers and underwriters still need narrative, explanation and experience to understand what makes a particular business unusual.
The technology should therefore organise the factual base while preserving room for expert interpretation.
The goal is not to replace judgement. It is to stop skilled people spending their time reconstructing basic information.
The handoff is where value is often lost
Over the years, I have seen the same problem appear in different forms: businesses providing the same information more than once, brokers rebuilding submissions manually and underwriters spending valuable time trying to establish what the risk actually looks like.
Technology can improve that process, but only if the information reaching each stage is clearer than what came before.
One of the least discussed parts of the insurance journey is the handoff between stages.
A client may provide detailed information to one system, only to be asked for much of it again later. A broker may refine that information before passing it to an insurer, but the structure may change in the process. An underwriter may then need to interpret a mixture of structured fields, free text and supporting documents before understanding the complete risk.
Every handoff creates an opportunity for context to be lost.
A better digital model should treat the proposal as a reusable body of permissioned information rather than a one-off form.
Where appropriate, the same reliable client-provided facts should be capable of supporting subsequent stages of the journey without unnecessary repetition, subject to appropriate professional review where required.
That does not mean unrestricted data sharing. Permission, purpose and appropriate control remain essential.
Within those boundaries, however, reusing reliable information can reduce friction for the client and improve clarity for the professionals receiving it.
Where automation becomes genuinely useful
Once the information foundation is sound, automation becomes far more valuable.
Routine validation can identify missing fields before a proposal moves forward. Systems can move information through defined workflows more consistently, reduce duplicate entry and highlight where additional context is required.
Structured data can also make it easier for brokers and underwriters to compare what has changed between versions of a submission rather than rereading everything from the beginning.
Artificial intelligence may add another useful layer, particularly in summarising narrative information or identifying gaps for human review.
But the same principle applies: AI should not be treated as a substitute for reliable source information.
The strongest use of automation is therefore not to remove people from commercial insurance.
It is to remove avoidable administrative friction around them.
Digitising the right problem
The industry understandably wants faster journeys. Clients expect modern digital experiences, and insurance professionals want systems that reduce repetitive work.
But speed is only useful if the information moving through the system is good enough to support the decision at the end of it.
Commercial insurance digitisation should therefore be judged less by how many steps have been automated and more by whether the quality of the handoff has improved.
Are clients being asked the right questions once?
Is business context preserved?
Can information be reused responsibly?
Do brokers and underwriters receive a clearer picture rather than simply a faster one?
If the answer is yes, automation can create substantial value.
If not, the industry risks doing little more than digitising the inefficiencies it already has.
Be the first to comment