Exclusive Interview with WeatherSource
In this interview, WeatherSource Vice President Craig Stelmach and our Managing Director George Cambanis discuss weather data usage and WeatherSource Listings on the marketplace.
Enlarged Screen-sharing segments:
Hi, this is George Cambanis and we're here with Craig from WeatherSource for another data provider showcase. Craig and WeatherSource are very strong data partner that cover environmental and weather data in the Mobito data marketplace. And today we have a great opportunity to speak with a true expert on this field, understand better the market, the industry, and most importantly for today, deep dive into some of the use cases that we have prepared.
And then Craig can share about the usage of weather data in different industries. So Craig, thanks for being here and maybe kicking off the interview with a word on the broader environmental data market. This is of course, a data category that has existed for for quite some years. It's one of the more mature as we see whether these are categories.
has grown to be a, you know, a $14 billion market with a lot of different companies in different industries consuming this data for their own products and services. But today we do see many big trends and many changes into the market that are seeming to pose an even bigger market and generating new use cases of this type of weather and environmental data.
Can you can you walk us through some of the most important such industry trends that you're seeing?
Good to be here and excited about this conversation obviously environment and weather data are a big part of of what's going on in the globe. I think we can look around at anywhere whether it be the excessive heat waves, whether it be the anomalies, weather, the wildfires that we're seeing across the globe.
More and more companies are now starting to pay attention to weather. Everybody has loosely always had a solid understanding of how weather impacts their business. But now that we're starting to see anomalistic weather or departures from normal companies are starting to need to pay even more careful attention to what's going on with the weather and have extremely actionable, and accurate hyperlocal weather information.
So I only see the the need for weather data increasing as we see more and more of these record type weather type events. And then on top of that, add in all of the sustainability initiatives in the ESG mandates. So now you're starting to see actual, you know, businesses whose lending and or success is actually inextricably intertwined with weather.
So a business that is going to build, say, a new development is now needing to convince their lenders that there is not a weather risk for that particular location. And then again, adding on top of that also mandates. So companies saying how are you going to achieve zero carbon output by 2025 or 2030? So we're only seeing weather become more and more prevalent in basically all lines of business, and that could be anywhere from a small mom and pop store that that sees weather really impacting footfall traffic all the way up to the large enterprise clients who are making decisions on a $500 million build for a new manufacturing facility.
All of those are weather are now becoming weather centric versus weather independent. So my my expectation or what I'm seeing in the market is there is a massive increase in demand for weather information to really understand how weather is changing and evolving and what's the evolution of that risk related to weather, particularly as it impacts a business or an asset location?
Yeah, Yeah. No, that's fascinating. And it's, you know, probably one of the unfortunate developments that weather is gaining in importance and it's becoming ever more anomalous and ever more predictive. But indeed, we're seeing that more and more companies can't but pay very close attention to this and incorporate it in their core business models. And great. So, you know, with this changing landscape, it would be great if you can introduce us to the company itself, Weather source.
Talk to us about perhaps its origin and how it's positioning with its data products in this changing landscape.
So weather source is a marketable different weather company than what many people are used to. Many companies will tell you what the weather is at the beach, but we're going to tell you how many widgets you're going to sell as a beach as to the as a result of the weather that consumers are experiencing.
We also like to say that not all weather data providers are created equal. So what weather source did was our many years of working with weather data. Literally, some of our chief scientist have been working with weather information for 30 years. When we built weather source, we really took all of the rabbit holes and all of the issues with historical weather
And we we use that to basically build a company that provides analytics grade global weather information. And what I mean by that is we bring a continuum of weather information with deep histories that flow seamlessly into the present and seamlessly into the forecast. And all of that data is globally uniform and temporally spatially complete, because our view has always been how can you predict what your business is going to do with forecast alone if you don't have historical weather data to quantify the impact in a historical and historical analysis to be able to then predict the future.
So what weather forecast does is we have global weather information covering every landmass in the world in up to 200 miles offshore. But unlike other providers that rely on singular focal points or weather sensing instruments or weather sensing observations, WeatherSource is ingesting all of the best weather information available, including satellite radar, airport reporting stations, cell tower information, and a tremendous of other weather sensing inputs.
We're bringing those all on to a globally uniform, homogenized global grid. And then from those grid points, very high resolution grid, we're talking five kilometers. We basically then map from these grid points and deliver that data directly down to precise latitude longitude. Now, why that's different and why that's markedly different is historically, many weather companies have relied on singular inputs that could be many miles away from your location of interest with the weather source technology, all of your location of interest are bounded by what we would call virtual weather stations or our end point grid points, which is our core data, our core gridded product.
And again, all of this data is is temporally and spatially, globally complete and with this information is readily served through marketplaces such as mobile, auto, readily accessed and easy to ingest and easy to work with. I can grab many years of historical information, perform a regression analysis, quantify the impact of weather, and then take the output of that regression analysis forward into my predictive model that is now being driven by forecast data, all working from a single source of truth.
So we've done all of the heavy lifting for you to really assemble a petabyte scale database that covers every landmass in the world in up to 200 miles offshore.
Thanks for sharing. Maybe a follow up, you know, a little bit and almost basic question to to what you just described. You know, I've noted the coverage you mentioned.
I've noted the accuracy that you guys provide. I know the precision by means of which clients can request hyperlocal data and the way in which you get there, Could you could you maybe help us understand which are the the those parameters that help you basically, or rather that make your users choose you? Like there may maybe I don't know if it varies by different company sizes and different company and different industries, but do you see any one of these or a combination of these factors playing a bigger role in the selection of weather data from companies?
Absolutely. When we first started this company, we started this in 2015 and in two years we were literally powering a majority of Fortune companies, many of which had done a pretty exhaustive analysis of benchmarking the accuracy of our data. And what they found is that our data, particularly in areas with high topography or geography or remote locations, was much more actionable than what historically was available.
And a good example of of why our data is is much better is we're a Boston based company. Many providers would be pulling data exclusively from Boston Logan Airport recordings Station. Now, the interesting thing about that station is it's at zero sea level, so it's basically flat to the surface. You've got a massive onshore ocean breeze and then you've got a massive urban heat sink 175 yards away from that airport reporting station.
So I was the weather information at that singular input, actionable for a brick and mortar retailer or a logistics company that might be 20 or 50 miles away. It just isn't. And historically, what providers would do is they would take that information at Boston Logan and use a simple interpolation method to extend it to your location of interest.
And they would call it hyper local. But that interpolation method in many instances does not account for things weather influencing factors such as topography and or geography. So the further you get away from that singular weather sensing input, the higher the degradation of signal quality and accuracy of that data. So a lot of our customers in a good example was we had a customer who came and said, you know, I have have I always have these massive outdoor customer appreciation events and all my customers are taking a look at their app on their phone and it's saying that it's raining.
Well, guess what? At my location, it's actually not raining because they're all drawing from, again, this singular weather sensing input that is located next to an airport is many times located next to a large body of water and or a large city, all of which, again, could have a significant influence on weather. So having our graded data and being surrounded by virtual stations ensures that not only your businesses but your consumers are also having access to data that is properly co-located for that location of interest.
So as we grew this company, we had more and more people who are starting to say, okay, this data is X, it's gap free. There's a deep continuum that from the past to the present. Now let's look at the data quality. So we're doing a tremendous amount of benchmarking of the accuracy of our graded data. And what we're finding is that our data is performing considerably better than most providers who are relying on singular weather sensing inputs.
So it has been so that is really why people are choosing us. They're also recognizing that, you know, many large customers are like, again, how can I just work with forecast data that is not going to tell me it's not going to put contacts behind what is the impact of that weather in the forecast on my consumers without having a deep historical database to quantify the impact in a historical time series and then carry that forward where I can now put context behind what these values in the forecast mean related to my business performance.
So there's a lot of factors that go into why weather source has really grown from a small company to now one of the leading weather providers in the in the world and also, you know, again, we started this in 2015. We grew this company so quickly that in 2019, Pelmorex, which also owns the Weather Network and Altium, owned a lot of other assets, purchased a majority interest in WeatherSource.
Now this is a business that had been around for 30 plus years in Canada. It's one of the most respected brands in Canada and it's got a tremendous longevity in its brand. But they said, who is this company that is serving up this extremely actionable, hyper local and analytics grade weather information? So them making that purchase of weather of a majority interest in weather service was really an accolade to our process in our methodology that is producing that this data that is resonating so well with our business and enterprise customers.
Yeah, great. No, and definitely the partial acquisition is a great, great testament to what you guys have built and what you have achieved until then. And from our side, we can we can definitely empathize with the multilayer approach that you described. We're seeing a lot of the more successful companies today that are based fundamentally on having a robust data product need a lot of these different data layers such as yourself, including the topography, including the altitude, seeing how all of these different factors can affect weather ultimately and the different weather attributes.
And so, you know, of course, from our side, we we have mostly our lens focused on the mobility space and a lot of the mobility behavior and the mobility activities are, of course, very much influenced by mobility, by by weather. I mean, and it's something that you can see a lot looking at traffic, looking at footfall, looking at even the way in which companies are delivering their services, last mile delivery and so forth.
Weather is, of course, one of the one of the constants that they need to to face and to incorporate and the way they do their business planning. And we are super glad to have had you in the marketplace and to be able to offer to our users and the clients your data sets and the different data products. And can you can you briefly walk us through this collaboration, how you guys are part of the marketplace and what type of data datasets are they, the products someone can access through them on the platform?
Absolutely. And I always like to say that weather data and I've actually got to use case, which we'll be talking about later on, peanut butter sales. But I like to say that weather data in mobility, mobility, meaning the movement of goods, people and or services, logistics, etc., are like peanut butter and jelly. They really go hand in hand because weather is one of the most significant influencers of the movement of people in goods and services.
Obviously, what we just saw in the United States was a good example. With Southwest. We had a lot of storms that were occurring across the United States. There was tremendous travel disruption over the holiday break that just occurred in 2022. So that's just a prime example. But it also relates to things like logistics for supplies. If I have service level commitments where I have to deliver supplies with a certain schedule and I have weather disruption, more than likely I'm not going to meet those service level commitments.
Add in the supply chain disruption. Now I have stores that are not in stock. Now I have distributors who don't have products. Now I have consumers who are not able to find products, sell it. So logistics, whether it be the movement of people, footfall, traffic, etc., or whether plays a part in a role in all of this. And as we said, as we see the increase in anomalies, tick weather increasing not only in frequency duration but severity, you're going to see an even greater impact on the movement and mobility concerns.
And again, across the board, any industry that's focused on the movement of goods, people and or services is going to be impacted by this weather, and it's only going to be amplified as we have increasing degrees of departure from normal. So with respect to the Mobito Data Marketplace, we were super happy to make this a strategic partnership.
But I've been working with George and the team for several years now, really starting to figure out what we wanted to present within the Mobito marketplace. So what we're offering right now is we offer basically broad access to all of our services from the mobile marketplace. You can access up to 20 years of historical weather information in both hourly and daily format.
You can access our forecast products, including both the GFS forecast, which is put out by NOAA and NCEP, or ECMWF which is the best forecast product in the world. We've got all of that available on the Mobito Marketplace. We also have climatology data, which is the mean and standard deviations of weather, very, very powerful tool for really putting context behind whether it's a baseline tool also allows you to identify and quantify the impact of departures from normal.
We also have flooding information. That flooding information is literally at a ten meter resolution. So extremely high fidelity and actionable flooding information. We also air quality and UV data, so we cover the gamut of weather parameters, but then we also have other type of assets like air quality, UV, flooding, etc., that are a little bit out of the norm of what you would consider normal weather.
And when I talk about normal weather, I'm talking about air temp, wind speed, wind direction. So we've got you covered all the way from, you know, air temperature all the way up to wildfires and everything in between. All of that data is available in the Mobito marketplace. We encourage consumers to try before they buy. We stand behind our data, so we offer very, very robust samples.
We're very, very generous testing of that data. And we also provide information and assistance. So if a user was within the Mobito marketplace, wanted to talk to one of our meteorologist, we always will. No problem. Hop on a phone, put you on with a very skilled meteorologist who has dual training. As a data scientist, we can really come in and help you get started with that data and give you some robust samples so you can really start to see what sets weather source apart from other providers.
Super Yes. And I can also testify to that. And you know, even though a lot of your data products have robust APIs that they'll be self served and indeed you guys are very hands on and super helpful in educating new users and in guiding them through your assumptions, your trial phase and how they can best leverage your data.
And with that, very happy to move to the next phase. My favourite phase and seeing how this data really is put into action with a couple of use cases that you have prepared. So yeah, let's dive right into it. Starting with the use case of how snowfall can impact traffic and the movement of vehicle assets. Super keen to see what you can say around that front.
Perfect. So I've got two use cases are going to cover today. One is the impact of snowfall on traffic patterns up in Canada where our parent organization is. And the second one is the impact of weather data on CPG sales in Florida on a hurricane forecast. Both of these are compelling and I know they're one is obviously directly tied to mobility.
The one is also tied not only to product sales down to the skill level, but it's also talking about footfall traffic because that is an impact on when consumers are coming into the stores in advance of one of these big weather events. So I'll share my screen and I'll go right into that too. It's hope I can do this one sec.
I'm just going to put my slideshow. It's coming up super if you want a full screen and as well. There we go. So this is the first one use case I'm going to talk about. And this is weather's impact on mobility, including the obviously the movement, this use cases around traffic. But as I said, it also can be used for obviously people in retail sales.
I just want to set the stage on this. And as you can see, George, what we had just talked about, what the impact of weather is, I've just got some statistics here where you really talk about speed reductions from 10 to 25% on wet pavement, 30 to 40% with snow, where your slushy movement and as you can see down the line, all of these are pretty significant drivers, particularly when you're talking about very we were still seeing the front page.
Do you see this one now? Yeah. Perfect for marketing. And so what you're going to see here is I've just got some general statistics on the impact of snow on and the movement of traffic, essentially. But what's important to note is that this is amplified.
So these you might see things like a 3% decrease or increase in congestion, etc.. But imagine you're in a metro area like New York or Boston or Toronto or Athens. Once you add in 3% and you're talking millions of people trying to go into one location, that's a pretty big significant impact. And you start to have compounding issues.
So if you have everybody trying to drive in a Boston in the morning and there's a 3% speed reduction that is going to flow down the line to basically everybody trying to get into Boston. Now, somebody who is expecting to be at their office by 9 a.m. is now at their office at 940. So it's a tremendous amount of of of impact.
Basically what you're looking at is a latitude, longitude. And again, all of our data is actionable down to the latitude longitude. We actually do a lot of work with the 407 ETR, which is the private roadway up in Toronto. But as you can see, this is basically what you're looking at on these charts are a Time series from January 12th to January 31st.
There's not a legend on the bottom, but you can imagine January 12th, this is the first date, and January 31st is the last. The traffic patterns that you're looking at, basically the red is going to recognize hour of day. The green is going to recognize basic traffic flow. So normal, normal traffic patterns going through the day, you aren't seeing a decrease in an increase in congestion or a decrease in average speeds.
But now if you look on January 17th, there was a large snowstorm that was there was a lot of accumulated snow. And right at that precise moment where you saw the increase of the snow, you saw a perfectly correlated decrease in the average speed. So you're seeing a pretty significant drop in again, when you see this type of activity, you have to envision that you've got millions of people on the roadway.
So now you're in a situation where you have traffic jams, you have gridlock, you have an increase for accidents. So this dynamic, although it's focused on congestion, changes the entire entire character of a commute. So it's just a simple example of taking our data and combining it with traffic patterns. And as you can see, when you have an increase in snow, particularly snow accumulation, you have a significant accumulation of significant impact of the average speed going down considerably.
And Craig, is this a highways? Is it the motorways? Do we know what type of streets are being monitored here?
Yeah. So this is basically this is the flow into the northeast side of Toronto. So this is really focused on, again, all of those roads, whether it be highways and say, okay, moving into Toronto. And on this particular day, you saw traffic delays of upwards of 2 to 3 hours.
Now, if I'm somebody who's delivering medical supplies or some other type of very, very sensitive product, I'm now working at a 3 to 4 hour delay to get into Toronto and deliver my goods. So very, very substantial in impact. And again, this could easily be provided to a logistics company who may make a decision, okay, when this snow starting, let me pull over and let me rest my truck or let me rest my driver and wait for that accumulation and versus sitting in gridlock for 4 hours, which as we all know, the more gridlock there is, there is an increase for accident potential also.
So a tremendous amount of information you can glean. And again, this can be applied to whether it be consumers commuting in, whether it be a logistics or a supply company or a delivery such as, you know, home delivery or groceries or Amazon deliveries, etc.. So try to make of sense. And there you can definitely add those to the the traffic planners and that side of things as well.
Looking at the traffic management, looking at the road maintenance companies, looking at the winter maintenance that needs to be done at roads. All of these guys are definitely very, very much affected and dependent on a good understanding of those correlations between weather and traffic and other stuff. Here is a very simple analysis like, you know, we have wind speed and wind direction.
You know, we even provide alerts and notifications when wind speed exceeds 40 miles per hour, which can roll over trucks. So there's a lot of weather information you can do around this with mobility. And this is no different with footfall traffic. So if I had footfall traffic in this particular instance, I could wear this weather information over and I guarantee I would see similar correlations to the movement of people versus the movement of traffic.
Yeah, makes a lot of sense. Would that be a good time to move on to the second use case that does have the footfall angle to it?
So like I said, weather and mobility goes together like peanut butter and jelly. So this particular use case is around weather and peanut butter sales. It was tied to footfall traffic also.
But I strip that out because it's a little bit busy on the particular screen. But this is basically we have live data coming from some stores in Florida that sell peanut butter. Okay. And what we always understand is that there's a big run up for products and services when you have a hurricane forecast. So I actually have a live demo, which I'm going to show you now as to how consumers respond when there's a forecast of of a hurricane.
You know, up here in the northeast, we like to say when there's a snowstorm or a nor'easter, we run out of toilet paper water in a lot of other consumables. It's across the board. And it's interesting how consumers respond with these big weather events. So I've got a little a little digit here, also a little dialog box that talks about retail sales and weather analytics and really, you know, businesses, weather, applying weather to retail analytics.
We know baseline that it improves sales by up to 2% and grocery stores see even more because people go out to buy consumables in advance of the storm to be able to have enough resources to to last through that storm period, particularly as logistics companies need to go in and restock shelves, etc.. Home improvement chains actually see a gain of 20% before these big weather events.
People are going out to purchase generators, people are going out to purchase other type of consumables, flashlights, batteries, etc. But I'm going to focus this one just on peanut butter sales. This was from 2017. And it's really going to show you how buying patterns change, which again, is also tied to footfall traffic, but how those patterns change as that forecast, as that forecast in hurricane on false.
And maybe before zooming into that case trend, understand the increase of top line sales that you just mentioned, how companies can use the weather analytics tool to bring forward this effect. And how is that how does that come to place? Is it mostly due to better inventory management predicting the orders that will come in, taking care of the delivery of those orders? What are the the key drivers of this top line increase?
So again, going back to the, you know, the importance of having historical weather information, what I can do with that is I can correlate weather information, not only normal weather, but also the departures from normal and extreme and severe weather. And I can understand what products sell at what volume during these big weather events.
So being able to say, okay, during a hurricane, my peanut butter sales increased 2,000%, that's going to obviously impact some supply in inventory and also product placement. So, you know, if I know that consumers are coming in to purchase road salt or batteries, I'm going to make sure that that is prominently displayed in the store. But I'm also going to try to take advantage of product placement for other products they would purchase in addition.
So if somebody came in and purchased batteries and I found out that 80% of people also purchased a flashlight, I'm going to put those flash right right next to that battery. So it's really to the battery. So generally also optimizing placement in my in my on my shelves of this product, but it's also inventory. And we work with one consumer actually where they just know when there's a big snow event in the south, particularly in the Atlanta area, they literally just drop off pallets of cat litter basically everywhere because cat litter can be used to basically as a to prevent slip and falls and act like an ice melt.
So and also to weight down trucks to put it in the back to add additional weight for rear for rear driven vehicles. So all of this information in an historical sense can really get a solid understanding. It's also product placement in in in your