It was 12:30 AM on March 3, 2020. An EF-3 tornado was on the ground in Nashville, cutting a path straight through the neighborhoods around our office. By the time the sirens stopped, houses were gone. Roofs, cars, fences, trees, all of it rearranged by a storm that most of our clients slept through and woke up to.
We didn't wait for the phones to start ringing. We pulled up our claim map, overlaid the damage path, and inside an hour we had a list of every client whose home sat inside the impact zone. We started filing claims that night. By 5 AM the carriers had adjusters rolling to addresses we had already flagged. Other agencies in town didn't even open their doors until 8 or 9.
Our clients didn't have to find us. We found them. That's the entire thesis of a claim map, and it's the single highest-leverage thing a small agency can build.
What a claim map actually is
A claim map is a private map of every client household your agency insures, plotted by address. Layered on top of that base map, you pull in live data feeds, NWS storm tracks, wildfire perimeters, hail swaths, flood zones, power-outage reports, whatever risk applies to your geography. When something happens, the intersection tells you exactly who to call.
It's not a CRM report. A CRM report tells you who lives in a zip code. A claim map tells you which of those houses had an EF-2 pass within 300 yards at 12:43 AM. Those are different answers, and in the middle of a disaster they lead to very different mornings.
What our 2020 map actually did
March 3, 2020 · Nashville EF-3
That was one storm. The same map handled the December 2023 tornado, the Hurricane Helene remnants in 2024, and every hailstorm that has rolled through since. Each event, the same pattern: we call the clients before they call us, claims get in the queue before the queue fills up, and retention numbers on that book of business sit about 8 points above the agency average the following renewal cycle.
Why this matters more than it used to
Carriers are stretched thin on adjuster capacity during CAT events. An F-3 tornado in 2010 meant an adjuster at your client's door in two or three days. In 2026 that same event can mean a week, sometimes ten days, because carriers are routing adjusters across four states at once. Whoever's claim is in the queue first gets worked first. Whoever calls at hour 48 waits behind a backlog that's already three days deep.
Your client doesn't see the backlog. They see that their neighbor's adjuster showed up and theirs didn't. The neighbor's agent had a claim map. You didn't.
What to put on the map
The first version doesn't need to be fancy. Here's the list we started with in 2019 and still run on today, give or take a few layers:
- Every client household with a homeowner's, dwelling fire, condo, or renters policy, pinned by address
- Auto-only clients as a secondary layer (useful for hail, flood, and wildfire events)
- Client name, policy number, carrier, primary phone, primary email, and last-contact date in the pin tooltip
- NWS severe weather feed (tornado tracks, severe thunderstorm warnings, flash flood warnings)
- Hail swath data from a service like HailTrace or CoreLogic (your carrier reps can often get you access)
- Wildfire perimeters from InciWeb or your state fire service
- FEMA disaster declaration zones for longer-tail events
- A filter that lets you export any selection as a call list straight into your AMS or CRM
That's it. If you're tempted to add ten more layers before you launch, don't. Ship the v1, use it on the next event, and let reality tell you what's missing.
Get started with Cowork
If you already have Claude's Cowork mode with access to your agency folder, you can have a working v1 on your screen in under an hour. Paste the prompt below, point it at a CSV export of your clients with addresses, and let it stand up a single-file HTML map you can open in any browser on any device in your office.
Cowork Prompt
Copy & Paste this into Claude
I want to build a "claim map" for my independent insurance agency. The goal is that when a disaster happens, I can pull up a map showing every client household we insure, overlay the damage zone, and quickly pull a call list of affected clients so we can proactively file claims and reach out.
Here's what I need you to do:
1. Read the client export CSV I've attached. It has these columns: client_name, policy_number, carrier, address, city, state, zip, primary_phone, email, line_of_business.
2. Geocode every address. Use a free method (OpenStreetMap / Nominatim, or a local geocoder) unless I tell you otherwise. Cache the results in a local file so we don't re-geocode every time.
3. Build a single-file interactive HTML map using Leaflet that: - Plots every household as a pin, colored by line_of_business (home, auto, commercial, other) - Shows client_name, policy_number, carrier, phone, email in the pin popup - Has a search box to filter by name, policy number, or zip - Has a "draw a polygon" tool so I can trace a damage path on the map - Produces a button that exports every pin inside the drawn polygon as a CSV call list (name, phone, email, address, policy_number, carrier)
4. Add a second layer for severe weather: - Pull the latest NWS active alerts feed (https://api.weather.gov/alerts/active) and display current tornado, severe thunderstorm, and flash flood warnings as polygons on the map. - Refresh the layer every 5 minutes.
5. Save the final HTML file and the geocoded client data to my selected folder so I can open it on any browser. Don't rely on external services I'd have to pay for.
6. When it's working, give me three things: - A one-paragraph README explaining how to refresh the map when I update the client CSV - A short checklist for the team on what to do the moment an event hits - A list of the next 5 layers I should add (hail, wildfire, flood, power outage, FEMA declarations) and where I'd source each one
A few notes before you run it. The NWS feed is free and public; no API key is needed. Leaflet is open source. Geocoding with OpenStreetMap is free for small volumes but slow, so run your client file overnight the first time and cache the results. Don't put client PII on a hosted map; keep it local, on a laptop or internal server. If you have a carrier that offers hail swath data, the layer addition is usually a one-line change once the base map is up.
Once it's built, use it on quiet days too
A claim map isn't just a disaster tool. Rotate through it quarterly. Filter by zip code and identify the streets where you have density, the ones where you don't, and the gaps where a producer should be knocking doors. Overlay carrier appetite maps and you'll see which neighborhoods are underserved by your current carrier roster. Pull it up in staff meetings when a new producer starts, so they see the agency's footprint at a glance.
The map pays for itself the first time you use it in an event. Everything after that is compounding return.
