Why an MGA's submission triage is the highest-leverage process it owns

The inbox is not the problem you think it is

Your underwriters are buried. Broker submissions arrive as emails with six attachments, half of them incomplete, and the team spends its mornings sorting instead of underwriting. Every vendor who has called this quarter has the same answer: extract the data, classify the submission, route it, and your underwriters get their time back.

Everyone agrees triage matters. Almost everyone then sells throughput.

We think that misreads what triage is for. At a managing general agent, triage is not an intake step that happens before underwriting. It is the first underwriting decision, and for most submissions it is the only one that counts. The leverage in the process is not how many submissions you can move. It is how many you decline, how early, and how consistently.

Triage is where appetite lives or dies

Every MGA has an appetite document. Very few have an appetite that is enforced the same way on a Tuesday in March as on the last Friday of the quarter.

Triage is where that enforcement happens, or quietly stops happening. Indico, a vendor in this space, defines triage as reviewing submissions, checking completeness, assessing appetite fit, prioritizing the best opportunities, and only then routing (Indico). Look at the order. Appetite and priority come before routing. That is a concession, from the sell side, that triage is an underwriting control. The same piece names the failure plainly: appetite checks applied inconsistently across teams, and priority set by "whoever follows up loudest."

That second failure is worth sitting with. When priority follows the loudest broker, your appetite is being set by your distribution relationships rather than by your underwriting guidelines. Nobody decided that. It emerged, one urgent follow-up at a time.

What happens when you only speed it up

Here is the pattern we see. An MGA has more submissions than its underwriters can review. It buys extraction and routing. The underwriters now see more submissions per day, quoted faster. Premium grows. The capacity problem is solved.

What has actually happened is that the filter got wider. A team that used to decline by neglect, because marginal submissions aged out of the inbox, now quotes them. Some bind. Those accounts were marginal for a reason, and the reason shows up in the loss ratio a year or two later, on a delay long enough that nobody connects it to the software purchase.

Speed up triage without tightening appetite and you convert an underwriter capacity problem into a loss ratio problem. The first is visible and uncomfortable. The second is invisible until it is expensive.

The vendor metrics make this easy to miss, because they measure the wrong thing well. SortSpoke's MGA page promises 45 percent more submissions processed and a 75 percent improvement in quote turnaround (SortSpoke). Those may be accurate. They are also both measures of volume moving through a pipe. Neither tells you whether the right submissions moved.

The number that matters: quote-to-bind by broker

If throughput is the wrong metric, what is the right one?

Start with quote-to-bind ratio, and cut it by broker. Not in aggregate. An aggregate bind ratio blends the brokers who send you tightly fitted business with the ones who send you everything and see what sticks. Split them apart and two things become visible.

The first is where your underwriting hours go. A broker whose submissions bind at a low rate is consuming quote effort that produces almost nothing. Faster triage makes that broker cheaper to serve, which is the opposite of the signal you want to send.

The second is what the bound business looks like. A broker who binds at a very high rate with you may be sending you exactly what fits. Or you may be the easiest market on their list for accounts others have declined. The ratio alone does not tell you which. The ratio next to loss experience by broker does.

Submissions processed per underwriter tells you how hard the team is working. Quote-to-bind by broker, read against losses, tells you whether the work is pointed anywhere. Only one of those belongs on a capacity provider's dashboard.

You do not own the risk, which is exactly the point

An MGA underwrites on someone else's paper. The carrier or the syndicate owns the risk, grants the authority, and watches the results. That arrangement is why triage matters more at an MGA than at a carrier, not less.

Your capacity provider is watching this process whether or not you are. AM Best recently revised its outlook for the delegated authority segment from positive to stable, citing tighter renewal economics and increased scrutiny of delegated partners, and carriers are described as increasingly focused on "underwriting consistency, loss ratio stability, and portfolio discipline" (Carrier Management). Hyperexponential, which sells pricing software to this market, observes that rapid growth in delegated underwriting "has outpaced the development of sophisticated oversight practices" (hyperexponential).

Read those together. The people who give you capacity are tightening their view of how consistently you underwrite. Consistency is decided at triage. An MGA that can show, broker by broker, which submissions it declined and why, is having a different renewal conversation from one that can only show growth.

There is an uncomfortable asymmetry here. A faster triage process that widens your filter will look excellent on your own metrics for about a year. Your capacity provider's metrics will catch it later, and they will not ask how many submissions you processed.

Why declination is hard to automate, and should be

Declining a submission is a judgement that the business does not fit, made on incomplete information, often against the wishes of a broker you want to keep. It is the least automatable decision in the process, and it is the one most automation projects are designed around rather than through.

Extraction tools are genuinely useful. Pulling the schedule of values out of a spreadsheet and the loss runs out of a PDF is not where underwriting judgement lives, and nobody should be paid to retype it. The problem is not the tool. It is what gets measured after the tool is live. If the success metric is submissions cleared, the easiest way to clear a submission is to quote it.

A well-run triage process does three things before it does anything fast:

• It states the appetite in terms specific enough that two underwriters would decline the same submission.
• It makes declination a recorded decision with a reason, not a submission that aged out.
• It reports outcomes by broker, so the pattern of what arrives is as visible as the volume.

Those are policy decisions, not software features. Once they exist, automation makes them cheaper to apply. Before they exist, automation makes their absence cheaper to scale.

We have made a version of this argument about claims before: automating intake before fixing what you capture produces bad claims faster, with a timestamp. Triage is the underwriting cousin of the same mistake. The process you automate becomes the process you are stuck with.

Where the vendor story and the MGA story diverge

Vendors are not wrong that underwriters spend too much time sorting. The disagreement is about what the recovered time is for.

The vendor story says: recovered time becomes more quotes, more quotes become more premium. The MGA story, the one its capacity provider cares about, says: recovered time becomes better decisions on fewer, better-fitting submissions. Those lead to very different configurations of the same software, and very different renewal conversations two years later.

The same shape shows up elsewhere in insurance. In our piece on manual certificate issuance at independent agencies, the finding was that most agencies do not have a volume problem so much as a handful of accounts whose terms were never negotiated. MGA triage has the same shape. A small number of brokers usually account for a large share of the submissions that should never have been quoted. Finding them is a reporting question before it is a technology one.

Put another way, the goal of triage is not to get to yes faster. It is to get to the right answer sooner, and a good share of the right answers are no. We would call that a decline in volume, but it is really an increase in standards.

Questions to answer before you buy anything

We are not going to give you a threshold or a shortlist here, because both depend on your programs, your capacity agreements, and your broker mix. But these are the questions that decide whether triage automation helps or hurts:

• What is your quote-to-bind ratio by broker, and do you know it without running a special report?
• When a submission is declined, is the reason recorded, or does it just stop moving?
• Would two of your underwriters decline the same marginal submission?
• What would your capacity provider say your appetite is, and does it match what you bound last year?

If those answers are clear, automation will sharpen a process that already works. If they are not, it will speed up the one that is drifting.

The conversation to have first

If your underwriters are drowning in submissions and a vendor is offering to drain the pool, the question to answer first is which submissions should never have been in it. That is a question about your appetite, your brokers, and what your capacity provider is going to see at renewal. It is also one we have answered before.

If that is where you are, talk to us before you sign the triage contract. We will help you work out whether you have a capacity problem, an appetite problem, or both, and which one the software should be pointed at.