The MGA systems readiness checklist.

This is the exercise we run at the start of every mapping session, published so you can run it yourself. Use it before you talk to any vendor, including us. Get the two or three people who actually run your workflows in a room, work through the four parts in order, and you'll land in one of two honest places: a clear first fix, or a clear decision to wait. Both are good outcomes.

Everything below works on this page. If you'd rather mark it up on paper, the printable version is linked at the end, no email gate.

1. Inventory your stack

Start with what exists, not the org-chart version, the real one, including the spreadsheet that quietly became a system of record. If work happens in it, it goes on the list.

• List every system you run. The AMS, the rater, every carrier portal, the spreadsheets that act like systems, and the email workflows holding them together. Expect the list to be longer than you thought. That's normal.
• Put one name next to each. The person who actually fixes it when it misbehaves, not the person whose title says they should.
• Trace what feeds it. Where does each system's data come from? Every arrow that turns out to be a person re-keying is worth marking in red. You'll count those in part two.
• Run the vacation test. For each system, ask: what breaks if its owner is out for two weeks? If the answer is "quoting" or "the bordereaux," that system just moved to the top of your list.

2. Spot the friction signals

These five patterns show up in almost every stack we map. Count how many you recognize, honestly, not aspirationally.

• The same data is typed into more than one system. Every retype costs time, and every retype is a chance to send a carrier the wrong number.
• Month-end reporting is assembled by hand. Exports, copy-paste, a late night, and a spreadsheet nobody wants to open again until the next month-end.
• Quotes wait on a person to be free. The submission is complete; the bottleneck is a human queue. Your producers can feel this even when they don't say it.
• Renewal season requires overtime just to stand still. Same book, same team, more hours. That's a systems ceiling, not a people problem, and it gets worse with every point of growth.
• There's one file "only Dana understands." Whatever her real name is, you know the file. So does everyone else, and so does part one's vacation test.

Zero or one of these and your stack is probably fine for now. Three or more, keep going, the next part decides whether the frictions are urgent or merely annoying.

What the five have in common is worth naming before you move on: mundane, redundant work (re-keying, chasing, month-end assembly) that repeats and never improves on its own. That is the work an AI Operational Workforce takes over, with feedback loops that correct it as it goes, while your people keep every decision that touches rating or binding.

3. Check your trigger

Friction alone rarely justifies a project. A trigger does, something external, with a date or a consequence attached. Check whether one of these is live today:

• A carrier or fronting partner is demanding more data on tighter schedules, reporting requirements your current month-end scramble can't reliably meet.
• A program launch has a date on it, new program, new capacity, and a go-live your current stack wasn't sized for.
• Producers are drifting to whoever quotes faster. Not complaining, quietly sending the next submission somewhere else.
• An error reached a carrier. A re-keyed number, a missed field, a report with the wrong total, and the conversation that followed it.

If none of these is live, modernization can probably wait, and that's a fine answer. Write down the date you ran this checklist, put a note in the calendar for next quarter, and spend the money on the book instead. Momentum you don't need is just cost.

4. Readiness questions

A live trigger says the work is worth doing. These four questions say whether you can actually do it now.

• Can the founder or COO make this decision? Phased work moves at the speed of the person who can say yes. If that isn't you, get them in the room before anyone talks to vendors.
• Can your team give 3–5 focused hours a week? That's the real ask of the people who know your workflows: confirm how things actually work, review the build against reality, approve each gate. Not more, but not zero.
• Is there budget authority for a fixed-fee first phase? Not a platform budget, one scoped phase you can approve on its own merits, with the next phase earning its own decision later.
• Do you know what you'd fix first? If yes, that puts you ahead. If not, that's exactly what an Assessment is for: a complete map of your systems, your three costliest frictions, and a phased plan with a Phase 1 quote, yours to keep either way.

If you worked through this and found a live trigger, a clear first fix (or a stack you can't quite untangle on your own) bring the completed checklist to a mapping session. It's a free 60–90 minute working session, and the one-page map of your flow is yours the same day, whichever way you decide to go. And if the checklist keeps pointing at the core itself, read whether you actually have to replace your AMS before anyone quotes you a migration.