The Bordereau Problem: What Running an MGA Taught Me About Why Reporting Stays Broken

The End of Every Month: Bordereaux Time

There's a rhythm to running an MGA that you learn quickly. The last few days of the month and the first few days of the next one look nothing like the rest of the calendar.

That's bordereaux time.

Every month, we had to collect information from different systems, pull it into one sheet, and send it to our carrier and reinsurer partners. Each partner had different requirements, different formats, different fields, different definitions for what counted as what. So the same underlying data needed to be packaged differently depending on where it was going.

And once it went out, it came back. With questions. Missing rows to explain. Discrepancies between what we reported and what they had on their end. Some items that needed corrections, others that just needed context: here's why this renewal looks different, here's the fee breakdown for this carrier, here's what this cancellation refers to.

Then we did it all over again next month.

The Bordereau Problem Was Never the Report Itself

The bordereaux isn't actually a complicated document. It's a structured view of what happened in a portfolio over a period of time: policies, premiums, claims, endorsements, exposures.

The hard part was everything that needed to happen before the document was right.

Even when we started automating parts of the process, the manual work never fully went away. Money that was collected manually and didn't sync into the right system cleanly. Renewals that always came in slightly differently than expected. Cancellations that needed a note attached. Fees that varied by carrier or by reinsurer. Individual rows in the report that needed specific context to make sense to whoever was reading them on the other side.

None of that is exotic. It's the normal edge cases of running a real book of business. But every one of those edge cases became a manual task at the end of the month, because the data had to be right before it could go out.

What I kept coming back to, sitting in the middle of that process, was how much of it was just moving information. Not analyzing it. Not making decisions with it. Just taking things from one system, reconciling them with another, making sure both sides agreed. Row by row, partner by partner, month after month.

That is a lot of operational capacity to spend on something that should, in principle, happen automatically.

The Part Nobody Talks About

The time cost is obvious. The relationship cost is less discussed.

When you run an MGA, your reinsurer relationships are genuinely important. These are the people who provide your capacity. The ones who decide whether to give you more of it, keep it flat, or pull back. You want to be easy to work with. You want to be a partner they trust.

And a meaningful part of being easy to work with is sending clean, complete, timely data.

When the bordereaux is late, or comes with errors, or requires multiple rounds of back-and-forth before it's right, that's not neutral. It creates friction. It raises questions, not necessarily about how the portfolio is performing, but about how tightly the operation is run. Are these people on top of their book? Do they have their data in order?

In a market where capacity decisions are made partly on confidence, that friction matters. And the frustrating thing is that it often has nothing to do with the actual performance of the portfolio. A well-run MGA with strong loss ratios can still produce a messy bordereaux if the data infrastructure isn't there. The two things are separate. They just don't always look separate to the person receiving the file.

We called it a necessary evil. The information existed. The relationships were good. The problem was entirely the distance between where the data lived and where it needed to be.

Why the Insurance Industry Hasn't Fixed The Bordereaux Problem

The obvious answer is standardization: agree on what the report should look like and the problem goes away.

It hasn't worked that way.

Standards matter. Consistent formats and definitions make data easier to exchange and compare. But a standard tells you what the data should look like, not how to get it there. An MGA still has to extract information from multiple systems. Someone still has to validate it. Partners still have slightly different requirements. The data still needs to reach the right place in the right form.

In practice, standardization often just moves the manual work rather than eliminating it. Instead of the carrier reformatting what they receive, the MGA reformats before sending. The effort is the same. It just lands somewhere different.

The real issue is that the bordereaux gets treated as a monthly reporting event when the underlying data is being created continuously throughout the month. Policies are issued every day. Claims are updated. Premiums change. Endorsements get added.

Waiting until the end of the month to collect all of that, find the issues, and reconcile everything is a design choice. And it's a design choice that creates the problem.

What a Better Bordereau Process Actually Looks Like

We started as an MGA. We did this work manually for years. That experience is what shaped how we think about fixing it.

The shift we're building toward at Notch is treating bordereaux as a continuous data process rather than a monthly reporting event. Instead of collecting everything at the end of the month and hoping it reconciles, the system connects to the sources where data is already being created, including policy administration platforms, claims systems, and spreadsheets, and structures and validates it as it moves.

That means issues get caught when they happen, not 30 days later. A payment that didn't sync correctly. A renewal that came through with an unexpected field. A cancellation that needs context. The system identifies it, surfaces the right information, and routes it to the right person to resolve, before it becomes part of an end-of-month scramble.

It also means that when data does flow to reinsurers or carriers, it has already been validated against the requirements on both sides. Different partners have different formats and field definitions. That mapping gets built once and applied automatically, rather than managed manually each cycle.

This doesn't eliminate the bordereaux. The report still goes out. But it becomes the output of a process that's already clean, rather than the trigger for a process that's about to get painful.

What Both Sides Gain from this New Approach

On the MGA side, the most obvious change is time. The hours that go into month-end data work get redirected toward the things that actually grow the business: strengthening reinsurer and carrier relationships, underwriting more risks, expanding distribution.

But the less obvious change is what that does for partner relationships. When your data arrives clean and on time, consistently, the conversation changes. You stop fielding questions about individual rows and start talking about the portfolio. How it's performing, where you want to take it, what additional capacity would let you do. That's the conversation that leads somewhere.

On the reinsurer side, the benefit is different but the logic is the same. When data comes in structured and validated, they can actually analyze it. They can see how a program is developing, identify where exposure is concentrated, and build a real picture of their distribution. The oversight that currently requires significant manual work can happen continuously when the data infrastructure is in place.

What both sides are really gaining is the same thing: the ability to spend their time on the work that requires their judgment, rather than on making the information trustworthy enough to use in the first place.

The bordereaux has been a necessary evil for years. Necessary because everyone needs the information. The evil part was never required.

‍