What brokers actually do, beneath the pitch decks and client presentations, is harder to name than it should be. I've come to see it as three distinct but deeply connected acts: representation, translation, and defense. None of them are transactional. All are structural. And when one breaks down, the entire architecture starts drifting in ways that don't announce themselves until pressure arrives.
Representation
We tell ourselves we represent the client. It's written into the mandate, formalized in the engagement terms. But that's not what I mean when I talk about representation as a function. Real representation is interpretive work. It's absorbing not just what the client articulates in meetings but what they can't fully express or haven't realized matters. The CFO's underlying anxiety about retention levels that never gets voiced directly. The operational reality that contradicts what's written in the business continuity plan. The political tension between local offices and headquarters that shapes every decision but never appears in formal communications.
A good broker doesn't just collect this information and file it away. They inhabit the client's frame of reference well enough to anticipate friction before it surfaces, to shape options in language that resonates with how the organization actually makes decisions under pressure, not how they describe their decision-making in the abstract.
This work resists automation because it operates in the space of what isn't said. The risk manager who mentions supply chain concerns but doesn't mention the board's private anxiety about a key supplier in political turmoil. The CFO who approves the renewal but whose body language signals doubt about whether local subsidiaries will comply. You can't checkbox your way into understanding what keeps someone's leadership team awake at 3 a.m., or what past incidents have shaped their risk appetite in ways they haven't articulated.
When representation fails, it fails quietly at first. The client says yes to a structure they don't fully understand, or worse, that doesn't quite match what they thought they were agreeing to. Six months later, at claim time, that gap becomes visible and consequential. By then it's too late to recover the alignment that should have been built from the beginning.
Translation
Here's where the function becomes more complex and less visible from the outside. Clients don't speak insurance. Markets don't speak business operations. The broker stands between these two worlds, translating in both directions, and this is genuine translation in the transformative sense, not simple transmission.
A client says something like "we're worried about supply chain disruption" and that statement, while meaningful to them, isn't something a market can price or structure coverage around. The broker needs to transform that anxiety into something that can be underwritten: specific exposure scenarios, concentration risk analysis, dependency mapping, contractual arrangements with suppliers, mitigation measures already in place.
Then the market comes back with their response, which might be, "We'll write it, but we need a 72-hour notification clause and a sub-limit on critical suppliers." Now the broker has to translate that back into consequences the client can evaluate in operational terms. What does 72 hours mean in the context of their actual supply chain? If the client's procurement operates on quarterly cycles and their logistics team is in a different time zone with no weekend coverage, that 72-hour clause isn't a technical requirement. It's a structural impossibility that will surface as a coverage gap during a claim. What constitutes a critical supplier in the policy language versus in their business model? What happens when those definitions don't align?
This is knowledge work that creates value precisely by bridging incommensurable ways of understanding the same underlying reality. The broker needs to be fluent in multiple languages simultaneously: the operational language of the client's business, the technical language of insurance contracts, the commercial language of market negotiation, and the legal language of claims defense. More critically, they need to know how to move among these languages without losing essential meaning in the translation.
Platform automation can make failures in this translation work particularly invisible. Information passes smoothly between systems without requiring interpretation, and misalignments get embedded from the start, invisible in the flow of data that appears to be working correctly.
Defense
The first two functions matter enormously, but defense is where they actually get proven. When a claim is contested, when coverage becomes ambiguous under specific circumstances, when a market pushes back hard on their obligations, this is the crucible that tests whether the entire structure holds.
No platform can argue a claim under contested wordings. No template can untangle jurisdictional complexity when a multinational client has an incident that touches three countries with different policy triggers and legal frameworks. No automation can stand between a client and a reluctant market when things get genuinely difficult and relationships are strained.
Major claims regularly hang in suspension because the broker who placed the program has moved on, and no one remaining can explain why certain structural choices were made. The documentation exists, perfectly filed. But the reasoning behind a split trigger structure across jurisdictions, or why specific sub-limits were set at particular thresholds, has left with the person who built it. The market sees ambiguity where there should be clarity, and coverage that seemed solid becomes contested.
Defense is existential work. It's the moment when the broker's function becomes most visible, when all the careful work of representation and translation either compounds into coherence or reveals itself as insufficient.
Defense reveals the quality of representation and translation that preceded it. Good representation creates a foundation of mutual understanding that makes defense easier when pressure arrives. Good translation produces clear documentation that withstands scrutiny and doesn't create new ambiguities under stress. When defense succeeds, it typically does so because representation and translation were performed well from the beginning. When defense fails, you can almost always trace the failure back to a gap in representation or a distortion in translation that went unnoticed during the calm periods. The claim becomes the moment when hidden misalignments become visible and consequential.
Why They Work as a System
These three functions don't operate in sequence. They work as a feedback loop. Every claim you defend teaches you what actually matters in representation. Every translation that fails under scrutiny reveals gaps in your initial understanding of the client's operations. The system learns, but only if the same intelligence holds all three functions.
This is why traditional cross-functional team structures don't work well for broking, despite their popularity in other contexts. Cross-functionality distributes partial competencies across specialists who each own a piece of the process. But effective broking requires holding all three functions simultaneously in integrated view. You need to see how choices made during representation will affect translation work later, and how both create the conditions for eventual defense. When these functions get distributed across different people or teams who can't see the full arc, the feedback loop breaks and the system loses its ability to learn and adapt.
Nobody else in the insurance ecosystem is structurally positioned to hold all three at once. That's what makes the broker's role irreducible to its component parts.
What Happens When It Breaks Down
What concerns me is that as platforms become more sophisticated and persuasive in their interfaces, these three functions risk getting simulated rather than genuinely performed. Representation becomes templated intake forms that capture standard data points. Translation becomes automated data mirroring that passes information between systems without transformation. Defense becomes an escalation protocol that routes problems through predetermined channels.
The organizational structure looks the same on paper. The dashboards show activity and completion metrics. But the system stops thinking in any meaningful sense. It stops adapting to what it learns. It loses the capacity for judgment that accumulates through experience.
This pattern appears when operational metrics look excellent while the capacity for genuine broking work degrades. A sophisticated platform migration maps every data field perfectly, preserves every document template, automates every workflow. Six months later, the team can't explain why a D&O program has separate retentions in different jurisdictions, or why certain sub-limits were structured the way they were. The logic lived in institutional memory. The platform captured the output but not the reasoning. When the people who built the structure move on, the program becomes an artifact no one can interpret.
This pattern appears when operational metrics look excellent while the capacity for genuine broking work degrades. Coverage structures get ported through system migrations, and at each step the program logic gets simplified to fit new data architectures. When claims arrive that test the structure, teams can't explain why certain elements exist or what scenarios they were meant to handle. Fluency in the platform's language replaces fluency in the client's business. The interpretive memory of why structures were built gets lost. The simulation of broking replaces its substance, and the difference only becomes visible when genuine pressure arrives.
What This Means Going Forward
I'm not arguing against platforms or digital transformation. The efficiency gains matter, and clients have come to expect sophisticated digital interfaces as table stakes. But we need much greater clarity about what technology can and cannot do.
Technology can accelerate transactions and make process visible. It cannot interpret complexity or maintain judgment across time. It can structure workflows and standardize inputs. It cannot hold the kind of accumulated understanding that gets better with experience rather than worse with scale. It can simulate the appearance of alignment. It cannot defend that alignment when it comes under genuine stress.
The brokers who will remain relevant aren't the ones who adopt new technology most quickly or enthusiastically. They're the ones who understand precisely which parts of their function can be automated without loss and which parts require human judgment that deepens rather than degrades over time. They're the ones who can articulate the difference between performing these three functions and merely simulating their performance through sophisticated interfaces.
Representation, translation, defense. Three functions that work as an integrated system. Easy enough to describe in the abstract. Much harder to automate without losing what makes them valuable. Essential to preserve if broking is going to remain a strategic function rather than devolving into process management.
The question for every broker is whether they're actually performing all three or whether they've gradually allowed platforms to simulate them while the substance quietly erodes. Your clients won't know the difference until the moment they need you most. By then the damage isn't just a failed claim. It's a program structure no one can defend because no one remembers why it was built that way. The broker who can't explain why becomes the broker who can't argue when. And that's when representation, translation, and defense collapse from professional functions into administrative tasks no platform can rescue.
