It’s mid-August, and Watch Duty’s CEO, John Mills, says fire season is only going to get worse. With British Columbia under a province-wide state of emergency and Spokane reeling from the most destructive wildfire in Washington state history, Mills takes us inside the most catastrophic fire season his team has ever faced. Since the L.A. wildfires made Watch Duty a household name, the Santa Rosa, California-based nonprofit has doubled its head count and revenue—growth Mills describes as a little too wild for comfort. He talks about the decision to expand into flood monitoring across all 50 states, and what it takes to scale a life-saving operation without investors, a cap table, or the luxury of getting it wrong.
This is an abridged transcript of an interview from Rapid Response, hosted by the former editor-in-chief of Fast Company, Bob Safian. From the team behind the Masters of Scale podcast, Rapid Response features candid conversations with today’s top business leaders navigating real-time challenges. Subscribe to Rapid Response wherever you get your podcasts to ensure you never miss an episode.
Right now, big wildfires are burning in British Columbia, a province-wide state of emergency, and of course near Spokane, Washington. Fires purposely set apparently that have burned through more than 10,000 acres, destroyed hundreds of homes, forced 65,000 people to evacuate. Can you give us a snapshot of where things sit in the Pacific Northwest and what Watch Duty’s operations look like? Who and how it’s being used in this season?
Yeah, it’s been a harrowing experience up in Spokane, unfortunately. That area is not all that fire prone. It was really a catastrophic event. Once our communities start getting impacted, we really spring into action and sleep in shifts, which gets kind of wild. It gets a little tiring, I think, for the team, but we love what we do. And so it’s great to be of service.
How did you hear about the fires in Washington? Did you know immediately they were a big deal? You must be getting pinged about fires all the time.
Yeah. We know very early. We either know when the firefighters know, or we know before the firefighters know. Let me explain how that happens. The obvious one is when a firehouse gets what’s called toned out, that means the siren goes off and the apparatus bay door automatically opens. The other way we get alerted is oftentimes through things like the wildfire cameras that have AI detections on them. So there are lots of different signals that we are scraping and mining 24 hours a day every minute. And then when that happens, our system gets a ping. It says, “Hey, at this lat and long [latitude and longitude], there was a tone out for a vegetation fire.”
Then we look it up in our system and we say, “Oh, that’s Spokane, Washington. Go find that channel in Slack, ping that channel,” and then everybody springs into action.
Every fire’s a big deal because we don’t know what it is yet. And so, we treat them all like it’s a life-and-death emergency. In this case, we knew very quickly that there were more than one fire, there was wind on it, and it was near structures, and that’s when it starts to become an all-hands-on-deck situation. But this happens in a matter of minutes. It’s very quick.
The way Watch Duty itself is operating, say in Spokane, compared to the way it operated a year, year and a half ago in L.A. during the wildfires, is that still roughly the same? Is your system the same?
Yeah, our system is the same yet more robust. So what happens after every event, similar to the fire service, we’ll do what’s called an AAR, which is an after action report. We’ll go through what we did, what we could have done better, what could have changed. It’s like a retrospective. And so our SOPs, sorry for all the TLAs, three-letter acronyms, our SOPs, our standard operating procedures, they get more hardened and more robust. And so, we learn something every time this happens. So the system is identical. It is the same. It is just better and stronger and more durable.
And I guess faster?
It is, but actionable and accurate is higher than speed. So it’s important. . . . If we’re one minute later and it’s more accurate or more precise, we will do that. And that’s always been part of our mantra. And so, just because we’re speedy doesn’t mean we’re hasty, which is what we had to get over early on with a lot of government agencies. They’re like, “You’re so quick, you’re doing X, Y, and Z, and it’s not correct.”
I’m like, “Look, I can see on the camera the fire’s burning through the town. I’m hearing reports that this is happening. We know this is true.” And so again, speed is a feature, but it is not at the [expense] of being accurate and actionable.
When we talked after the L.A. fires, you were considering ways to expand Watch Duty’s reach beyond fires to other emergencies and natural disasters. And you recently announced flood monitoring across all 50 states, almost exactly a year after the Texas floods that killed more than 130 people. Was that incident what focused you on flooding, that so many residents in that Texas tragedy didn’t receive better warnings?
Well, what triggered it actually was this was always part of the plan. And so ironically enough, at the end of the year of ’24, we were planning on starting our flooding research and development in January of 2025. And then January 7th happened, and then we scrapped the project for a whole year because frankly, we spent two months in the L.A. fires. And so, we really doubled down on focusing on fire.
So the incidents are unrelated. Frankly, I wish we did it earlier so we would’ve been ready for the Texas floods, but we were not, unfortunately. It took us, I don’t know, probably nine months of research and development. We brought on meteorologists to help us. We have a staff meteorologist. He’s one of the 99 incident meteorologists in America who does this. He was actually incident meteorologist on the Palisades and Eaton fires [in Los Angeles County from January 7-31, 2025].
What’s different about flood risk versus fire risk? I’ve heard you say that a lot of people underestimate the danger of water, that the average person doesn’t understand the risk the way they might with fire.
Yeah, it’s interesting. People will ford the river in their car, and yet they will run away from a wildfire. But we have fire in our stoves in our house. You know what I mean? It’s really interesting to look at the sociology and psychology of how these events happen, how people die, how they misjudge these events. And so a great example is, Bob, I think you’re a New Yorker, if I’m correct, right?
I am, yes.
If you remember before Sandy [a 2012 hurricane], there was a, I shouldn’t say a false alarm, but there was a storm that came through New York and they told everybody to evacuate. Well, I have friends who evacuated, and then they got stuck in endless amounts of traffic coming and going from the city. And then when Sandy came, guess what they did?
They didn’t leave.
They didn’t leave. And so, it’s a real problem when you’re not telling the story as properly as you can. And so we try and not sensationalize as much as possible, but floods behave a lot differently. So first of all, there’s more wildfires than floods. And so, there’s so many fires to track. There’s only about 15 to 30 catastrophic floods a year and usually less. And so, the reporting cadence is different. You have more time usually, not always. And there’s a lot more digital data. And what I mean by that are there is Doppler radar, there’s USGS stream gauges, precipitation sensors.
The U.S. Geological Survey.
Yes. And so, we have a little more of a leg up than you do on fire when all of a sudden an ignition happens and wind hits it, and next thing you know, it’s Spokane or Palisades is happening.
And because people emotionally react to flood differently than fire, does that mean the way that you communicate with them on the app is different?
We rely on the great work of NOAA and the NWS [National Oceanic and Atmospheric Administration and National Weather Service]. And so, they’re putting out reports constantly. Sometimes they’re like three pages long. It’s like a very long PDF talking about very nerdy things, and it’s not really consumable by the end user. We’re able to take that information, we rebroadcast it, and then of course we add color to it as well. So we’ll add data from precipitation gauges. This river is flooding north of you. It’s hitting a catastrophic level. It’s hitting a flood plain.
And so when you get a longer storyline of events, that helps you gauge the disaster rather than just the, “Get out of Dodge,” it’s like, “No, this is getting worse and worse and worse.”