Ask most MGA distribution leaders how things are going and you'll get activity. Emails sent, broker calls, meetings this quarter, submissions received, new producer relationships opened. Real numbers, tracked, reported and reviewed.
They're tracked because they're the numbers you can actually get. That's a good enough reason, and it's worth saying before pulling them apart, because the alternative isn't obviously available either.
Why activity metrics get tracked
Every one of them falls out of something that already exists. A calendar gives you meetings. A shared mailbox gives you submissions received. Nobody had to build anything.
The submissions number is usually the weakest of them, incidentally, because at most MGAs it's a count of what arrived in the shared submissions mailbox, and anything a broker sent straight to an underwriter isn't in it. So the denominator moves depending on how your brokers happen to address their mail. That's a rubbish denominator and almost everyone is using it.
The deeper trouble is that activity measures effort. It tells you the team is doing things. It doesn't tell you whether the book is in better shape than it was in March, and those two questions come apart more often than you'd expect, especially in a soft market where there's plenty to be busy with.
What distribution performance means at book level
The question underneath all of it is whether your producer relationships are worth more this quarter than last. Not how many things arrived. Whether the relationships that generate business are generating more of it, and whether the ones that have stopped are ones you meant to stop.
Which means the unit of measurement is the producer firm rather than the week's traffic, and the comparison is against that firm's own history rather than against the rest of the book.
Four things worth counting
Four rather than three because coverage and signal mix catch different failures, and four rather than five because the fifth one everybody asks for is average response time, which we'd leave off. Aggregate response time is dominated by easy mail and it comes out reassuring almost regardless of what's happening underneath.
What none of these will tell you
This is the part usually left out, so it gets its own section.
None of them tells you what anything was worth. A firm that dropped from twelve threads to four may have lost you nothing, if the eight were a single account that went to another market for reasons of appetite rather than service. We can count the drop. We can't price it, and neither can you.
None of them separates a relationship that's mature from one that's dying. Both look like servicing traffic and a steady renewal. We've not found a way round that, and we're sceptical of anyone claiming they have.
None of them survives a short history. A producer you started dealing with in June has no normal to be measured against, so the movement number is meaningless for them and stays meaningless for a couple of months. Anything that gives you a confident trend on six weeks of data is telling you something it doesn't know.
And counting threads counts threads. One bind instruction and one query about a certificate are both a thread. Weighting them differently means classifying them, and classifying them means either reading them or inferring intent from the outside of the message, which is doable and is less accurate than reading.
We produce versions of some of these numbers and they carry exactly the same limits. Worth knowing before anyone puts them in a board pack.
The four counts, arriving weekly instead of costing you an afternoon. Same limits either way: it can tell you a firm has halved, and not what the half was worth.
Doing it by hand this month
You don't need anything to try this. Take fifteen producer firms, ideally a mix of large and small rather than the top fifteen by premium, since the top fifteen are the ones you already know about.
For each firm, search on their email domain across whatever mailboxes you can get at, and count threads for last month and for the month before. Mark each thread commercial or servicing, roughly, without agonising over the borderline ones. That gives you signal mix and movement for fifteen firms.
Then list the firms that produced nothing commercial in either month. That's your coverage number, on a sample.
It takes an afternoon and it's dull work. What you'll get is a rough answer to whether your book is warming or cooling, and a much sharper sense of which producers you had genuinely no view on, which is usually the more useful half.
Doing that every week for the whole book is where it stops being feasible by hand, and doing it from Microsoft 365 email metadata is what we build at BindSignal. The afternoon version will tell you whether the weekly version is worth anyone's money, including ours.