GCI - The Last Great Data Race - Inspire 2017

One thousand miles of the roughest, most beautiful terrain on the planet. A musher with a sled-dog team in below-zero temperatures on a frozen trail dotted with wildlife, bordered by mountain ranges, and peppered with treacherous river crossings. This is the Iditarod: "The Last Great Race on Earth," and analyzing data for it has some unique challenges created by cultural and environmental factors. When your customer's address is "The Last Grey House in Noatak" or "SW Face White Mountain," GIS mapping can be a problem. In this session, you'll learn how GCI, an Alaska-based technology company, has brought technology to the Iditarod. You'll also hear how self-service analytics has played a major role in GCI's corporate strategy. Cute puppy pictures guaranteed!

Video Transcription

Matt Childs:
And I thank you all for coming today. So, to get us started, we're going to talk a little bit about who GCI is. So as Ramen said, we are the largest telecommunications provider in Alaska. Which I like to joke around and say is a little bit like being the tallest midget at the circus. It's a thing. I want to talk about what the Iditarod is. Just interestingly, by a show of hands, how many of you had heard of the Iditarod? Know a little bit about it. Those of you have children have probably watched Balto a million times.

Then we're gonna look at the traffic analyst study we did in conjunction with the race. Talk a little bit about the workflow that we built to handle the data, our analysis, our next steps. And then I'm gonna leave a little time at the end so if you all have questions I can hopefully answer them. There a couple of short videos because rather than talk about Alaska and talk about the Iditarod, it's a little bit easier to show you what they are.

Around here, we don't wait for others to pave the way. We forge our own paths. Overcome our own challenges. GCI is driven by this spirit, fulfilling customer's needs when no one else dares. Investing in people and communities, and delivering the best technology to the farthest corners of the Earth. All to connect you to the business you depend on, the people you care about, and the life you love. We are GCI.

Matt Childs:
So, as you can see from that we cover a wide range. Alaska is a big place, about twice the size of Texas. We have the largest network throughout the state. We serve our customers with local telephone service, long distance, which is where the company got its start. We are a wireless provider. We have a high speed broadband. In fact, in our urban areas we have one gigabyte service. And we are also a cable television provider in many of our locations.

I have been with GCI 10 years, two stints. I started with them in 2000, left in 2005. Went out and did some consulting work for a while and came back in 2011. It's a good place, has interesting problems to solve. So, started BI work in 1998. I went to work for Entergy, did a little bit of data warehouse work for them, micro strategy and Oracle. Really old stuff. I helped them build our first warehouse at GCI, which we're gonna put down next year. So, I'm pretty excited about that as well. Not only did I get to help build it, I'm gonna help take it apart.

I started this self-service business intelligence initiative back in 2013 when I realized we needed to move away from that siloed operation. We had been a traditional report provider, mostly descriptive analytics. Wasn't really working for us anymore. We weren't meeting our users' demand, so it was time to move away. So, we went looking for some tools and we found Alteryx. We started with four licenses, we have about 60 now. We're on track to have about 80 users by the end of the year, which for our company is a lot of folks. We have good penetration.

And we are currently working on first big data implementation. So, our Hadoop platform just went in last week. We're excited about what capabilities that's gonna bring for us. We generate an enormous amount of usage data probably over a million records per day per usage type and we several usage types. So, we're generating quite a bit of data.

And that lovely picture is the only picture you will see of a dog that is not a sled dog. That is Wyatt, my dog he is not smart enough to pull a sled.

So, we talk about the Iditarod. It's a pretty big challenge. The race is roughly about 1,049 miles and you can see on this map there are actually two routes. There's a Northern route and a Southern route, and they alternate each year. And in fact, when we look at this year's data you'll see that there is a third route. Those mountains right there near Rainy Pass and Finger Lake, the Alaska Range, this year did not have enough snow on them. So, the Race Committee could not take the race through that mountain pass. So, the race was started in Fairbanks, which is not on this map, but a little but northeast of Anchorage. About 500 miles.

So, this year the race was won in eight days, three hours, forty minutes, and that shaved 19 hours off the record time, which had been in 2002. And keep in mind that that eight day run, that is with a mandatory 24 and a mandatory 12-hour in them. So, these guys are really moving. The race has been around since 1973. It is still younger than I am, makes me angry. And the slowest race ever won was in 20 days back when they first started running it. So, technology has significantly changed the nature of the race like its changed everything else. And these pictures you can see were taken by Henry Aaron who is back there in the audience, who went down. The official race start is held in Anchorage no matter where the race starts from. So, they always do a run down our main street, 4th Ave. And so Henry was down there and took these great pictures.

Thru beards thick with ice, toes numbed with cold, and eyes heavy with sleep, you and your team push on. Forging your own path and overcoming every obstacle. You've worked together since the beginning and have never backed down when things got tough because you know the biggest challenges bring the biggest rewards. GCI lives by the same spirit, always working together to find newer, better ways to pull Alaska forward. Fulfilling customer's needs when no one else dares, bringing the best technology in the world to the farthest corners of the Earth. No matter the length of the journey.

GCI is powered by a team of Alaskans who like you, find innovative and resourceful ways to connect our neighbors to the life they love and the people they care about. From Anchorage to Fairbanks to the Burled Arches of Nome, our team is proud to be a part of yours.

Matt Childs:
So, we sponsor the race and that was a video our marketing group put together for the Musher's Banquet, which happens the night before these 60 crazy people attach themselves to a dogsled and begin a 1,000 mile trek across No Man's Land. For which I have ultimate respect for, but they are crazy. And for those of you who run, I'm runner. Just thought that was funny. All right, can't have a little bit of snark.

So, traffic analysis. So, one of the interesting things about the race route is it's fixed. Well, relatively fixed when weather is not screwing with us, it's fixed. And our wireless network, basically now, runs along that route. It didn't use to. We used to have to drop temporary equipment out for the race to support communication for the race, but now we have fixed towers out in those locations.

One of the challenging things about Alaska is that we are twice the size of Texas. We are half the population of Rhode Island and 350,000 of those 700,000 people live in Anchorage. And then there about 80,000 people in the valley around Anchorage, 20,000 down in Juno, and roughly another 80,000 up in Fairbanks. The rest of those 300,000 people that are left over live in the rest of the state. So there are some villages where there are 2,000 people and there are some villages where there are 15 people. In the last decade, we have spent about 300 million dollars to build a network that services 45,000 customers. So, for us it's a little bit more than just the bottom line commitment. Living out in those rural areas presents a challenge and we believe it is important to deliver services to the folks who live and work in those areas.

So, when you're managing a network that big and that diverse one of the issues you can have is how to deal with capacity and how to deal with surge management. So, that's what led us to looking at this traffic analysis. One thing about the Iditarod is, it has a start date, and it essentially has a stop date. You don't know what that stop date always is gonna be, but you know there will be one and that is typically when the first musher crosses the finish line in Nome. The rest of the mushers typically follow in two weeks, if they're not done, any mushers that are still left on the course about after two weeks are typically picked up then. Because it's starting to get a little dangerous out there for them.

So, you have this fixed moment and then you have this fixed set of events that go along with that. As the mushers begin to go, you're going to have teams. So, support teams, volunteers, moving along with them. So, we knew that we would have this ability to potentially track the traffic along the network in conjunction with the race. And the race data is provided by the Iditarod. So, they have a website. They update the race daily, so that the people who are supporting the teams can know. There's a lot of logistics that go on with getting dog food and hay and supplies and replacement sleds and replacement dogs out on to the trails. So, there's a lot of airplanes moving, lots of folks, lots of traffic, lots of things happening, but again it's predictable.

And what this provided for us was a little bit like a real world lab-like environment because it is a fixed setting. So, we thought this would be interesting place to do a study and then we can baseline with known data. And we started this last year, was a our first year doing this. And were gonna continue doing this each and every year. And it'll be interesting as we build up data what kind of analysis can come from that.

So, when we're looking at the data there are two key pieces of data. There's the race data, which comes from the Iditarod. We scrape that off their website. We use an Import.io to do that. It's not the prettiest data, but we work with it and it's very, very detailed. I can remember 20 years ago I helped a musher in a support effort and I would go over there to the Iditarod headquarters and you would have to go over there physically and walk in and someone would have an Excel spreed sheet and have this giant thing printed on the wall. And they were getting phone calls from the villages as mushers were going through giving them updates. Now, they are able to GPS, essentially GPS track the mushers, which is much safer for everybody. Because occasionally a musher gets snow blinded, wanders off, and as you can imagine if you get far too off the trail out there it becomes a life or death situation. And it becomes a life or death situation for the people who are looking to help them as well.

So, technology has really helped improved that, but what its done is it has provided this really rich data set on the Iditarod website. It's very, very, in-depth data. So, we chose to work it at the hourly level because we wanted to find a manageable level to pull that data in. The other piece of the data is our wireless usage data. So, like I said we have towers in all these locations now, which is great so we can pull that data off of them. And we'll show this a little bit, we're gonna use Alteryx's spacial capabilities to help us plot where those towers are in conjunction with the race path.

So, this is not a special data set. This isn't a data set that's curated specifically for this study. This is just our wireless usage data. It gets generated every day on all these towers. Every time someone picks up their phone and makes a phone call with it or sends a text message or you know looks up something on the web then it generates a usage record or typically multiple usage records. Those usage records come in. And that was important to us too because we didn't want to be looking at special data. We wanted to be using the real usage data because we want this to be the beginning of a larger foundation for making you know better models throughout the company and not just specifically tied to this event. And then we narrow the data down to match the days of the race so that we have a manageable dat set. So, we pull a little bit before the race, a little bit after the race, so you can see the trends, but you're typically looking at data that's matched up with the race.

So, start day it's pretty exciting. They start very early in the morning down on 4th Avenue, they are lacing up the dogs. I can tell you those dogs just want to run. They are barking and howling and ready to go. And as you can see in this bottom picture here, if you're lucky they do a drawing you can sponsor. There are riders typically in the baskets as they ride through town so you can come up to Anchorage you can look for that oppurtunity, though bundle up it will be cold.

So, the race gets started. This year the race started off in Anchorage. They run down through what we call the Campbell Tract, it's probably four or five miles through town on a set path, and that's the sort of the commemorative start. They got up, they pack them up. Normally, they would go 50 miles up the road to Willow and they would start the real race the next day. This year they got on airplanes, they flew to Fairbanks, and they started the race the next day from Fairbanks.

So, lets go look at the workflow and Alteryx. Get a little more interesting. So, this is the, give you, roughly the whole thing there and I'm gonna zoom on some pieces as we talk about it. But basically what you're seeing is this is the process of pulling in both the Iditarod data and the usage data and creating... For us we're a Tablo Shop, so creating multiple TDE's for the visual analysis that comes later, but I'm gonna call out a few places where this analysis probably isn't possible without several of Alteryx's tools in place. Bear with me.

So, these first two are Iditarod data. We are basically pulling in the race logs. Pulling those in, cleaning them up a little bit, and generating an initial set of TDE's. And you'll noticed for those of you, there's some errors here, obviously I'm not connected to our network. So, it can't find home so it's complaining to me. And then the checkpoint locations. So, we had to come in, we're doing a little bit of regex clean up in this data set as well. So, a lot of clean up happening sort of standard stuff, right? Then we're also gonna parse out the times because no one in the world has a standard time format. I don't know if anyone seen that XKCD Comic where they talk about standards and decide they're gonna go create a new standard for everything and yet another standard is created. And then this piece right here, this is probably one of the most significant pieces of the workflow.

So, this is using the in-database tools to go and get that usage data because there is a lot of usage data and if we could not leverage the server that usage data sits on... As it stands this job typically runs bout three hours. Prior to having the in-database tools available to us we would have to pre-aggregate that data outside of Alteryx and typically we couldn't get the job to complete. The data was just too, it was too much. Trying to pull it in on any machine that we could run it on. All right, without me going out and selling my soul to get servers from our server admin guys. So, this is a huge win for us and you know three hours sounds like a lot. We're literally talking about hundreds of millions of rows of data, so doing it in three hours is pretty amazing. And this is one those things we're really excited about because eventually this data is gonna be moving to Hadoop platform. We expect to be able to pick this workflow up, point it at Hadoop, do the in-database connections, and then work on that platform. And probably see this shrink from three hours to a half hour I would guess.

So, a lot of work went into this getting it worked out. Working through the usage types. Getting the data pulled together. And then the next interesting piece, another piece also no possible without Alteryx is the spacial. So, we've gotta find the usage that came off the towers that are essentially within 25 miles of the actual race course, right? That way we know we are just getting the data that are associated with the race because we have multiple towers that are out there, but some of them are 50 or 100 miles away. And if we pull data in off them we would be pulling data that may or may not be race related. So, we wanted to pick a limit and set up the spacial tools and again without this, we would be guessing, right? Either that or we would have to turn this over to an external spacial tool, something like Esri and ask it to go and give us these towers back and come in here and manipulate this by hand. So, being able to work with this... When we brought Alteryx in before Siege I pick them all up with spacial because we had one use case for spacial for some FCC reporting that we did. I amazed at how much are spacial capabilities have grown because it turns out to be so applicable in so many aspects of what we do.

All right. So, a little bit more data and then another spacial analysis so that we can create spacial objects for the checkpoints. So, each of the checkpoints that's along the route as well because some of them are incredibly small. And again all of that gets generated out to a TDE, then we pull the list of locations. So, lots of data from, you know in this case just a couple of different data sources, but a great deal of data getting generated to get us out to several sets of data that we can analyze and then pull together. So, I want to go back.

So, what were some of workflow challenges? So, obviously the large data sets, right? So, the in-database tools playing a key role in helping us to hammer on those really large data sets. We have some pretty hefty database servers in our shop were fortunate to be right sized. And so, this allows us to take full advantage of the power of those databases. Let them do all of that crunching. The spacial analysis also key. Getting that tower data close to the race course, getting those checkpoint mapped out, and being able to tie the checkpoints and the towers together. Getting us exactly what we need in that space. And then of course data clean up and migration because as it turns out the Iditarod's data is not clean. And I'm not casting a stone at them because guess what? Our data is not clean either.

So, there's a lot of cleansing that ended up having to go into this even though we control the hardware and the infrastructure that generates records off the tower. The towers don't always generate great records and sometimes the, you know based on configuration or the way the call comes into it, it generates wrong data. You know, we would go back to network engineers and point this out and they will say, "Yup, we know about it and we're working on it." And they're getting it because in their world it's not... You know, for us for analysis it's important for them it hasn't gotten to a place yet where it's creating a customer related problem. So, they have to get those taken care of first. But, there was again looking back at the workflow at the sheer number of regexes, parses, clean ups that we were doing there, you know, speaks to the fact that very little of this data comes in sort of ready to work with. And without the power of Alteryx it would have probably taken five times as long to generate that data and it probably would've been full of mistakes that sort of the old ways we had done things.

So, let's go out real quick to Tablo and just see a little bit of what our results look like and I can show you the race path for this year so you can see it's a little bit different. So, as you can see there as we start in Fairbanks you end up with a little bit of a different course. This is Anchorage down here. Willow is roughly about here. So, normally you'd start here in Willow and... Walk out here. You'd start out in Willow and make your way through the Alaska Range up towards Nenana and Manley in that direction there. So, it is a significant change when the race can't go, but it is charted out so that the mileage is essentially the same.

So, we put together a little bit of a teaching visualization here. You can pull up based on the Iditarod data we pulled in we realized there some interesting information in here. You can pull the mushers, you can see where they finished, where they peaked at. Down here you can see how their times rated against the average times compared to the overall average speeds of all the teams. How long it took them, what their average speed was, little bit of a biography about the mushers. There are some interesting folks who come to Alaska and mush dogs including a former runway model. It is a different collection of people that is for sure. Actually let's go here.

So, this is data from March 9th and data from the Ruby checkpoint. And what we're looking at here is right here on the ninth of March, 56 mushers came through the Ruby checkpoint. So, the largest single day we had mushers there and sure enough we're able to see peaks from 12 AM to 1 AM on the network all throughout the ninth because what's happening is as those, and you can see it down here. Some of its 2G data, some of its 3G data, a peak in the SMS events, so text messages as well as voice on that day. So, you can see a significant peak in Ruby. And what's happening is as the mushers are coming into town, right, their support teams are getting there ahead of them. They're calling back to Anchorage. They're calling back to Fairbanks. The media is with them, so the media is sending out its stories, its pictures.

So, we're starting to see a buzz, right? You start to see that expected flow of traffic that goes along with the sort of the front of the race, right? It's almost a little bit like a wavefront. So, as the racers move and as they go through checkpoints in mass, you see the traffic pick up in that checkpoint.

So, this is the last day. You can see here on 14th, right? So, four mushers checked into Nome. Mitch Seavey was the first musher, he checked in as I said it took eight days, 23 hours. We see now a spike in 3G traffic, so we have a 3G network in Nome. We see a spike in 3G traffic and then essentially it stays, right? Because now coming in late on the 14th was Mitch Seavey and three other mushers who were just right behind him, but then the next day 45 and 28 after that right?

And there was a just so... You know, so the guy I helped to do it 20 years ago, when he crossed the finish line in Nome they asked him what was the best part about knowing that he was getting close to Nome and he said it was when he could see all the trees. So, yeah there are no trees outside of Nome, but when you haven't slept in five days you see trees right? So, as these guys are coming, amazingly enough, Nome turns into a bit of a part town. I don't remember how many people live in Nome, but it's not very many. But it's population essentially triples overnight. We could probably do just usage study on Nome alone at the end of the Iditarod, but again we saw in this case we sort of bore out our assumptions.

We saw what we expected to see as Mitch crossed the finish line in Nome, traffic spiked, and essentially stayed there because now he has won the race. There's a lot of fanfare around that. There's gonna be a lot of TV coverage. There's gonna be a lot of local television. There will be will national news there. Typically, the end of the race brings out lots and lots of folks to that area and people sort of start piling into town to take part in the celebration. And again down here you can see that this is the cell site in Nome and again tied in with the race path. All right.

So, what did we learn by doing this? Because we are a self-service shop and we are sort of non-standard. It is always interesting on keynote warnings to listen to Dean speak because we don't fit the standard Alteryx mode because we are IT. I am the Director of Data Analytics in our IT shop and we brought Alteryx into the organization, right? So, it didn't come in through a line of business. It didn't come in, you know, but we brought it in specifically to give it to the line of business. So, one of the reasons we do these types of studies is so that we can learn a little bit about our data, but also we can pass those learnings on.

So, one of the things we learned is that there are no limitations working with large data sets, right? The in-database tools will let us leverage the power of our databases to do heavy crunching on that side you know making this type of analysis possible. Spacial analysis is crucial. Having those analysis tools, having them be easy to work with. You know, lets us do this type of thing where we are taking two data sets that are geographically tied, tagging them and then bringing them together. As I said, we learned that not of our data is clean. We're gonna work on that. We're never gonna get it all the way there, but you know this is one of those things where the benefit of this is when we find data that is particularly challenging to work with we typically put together tools or clean up efforts to get it done and then we can share those out with our user base.

And then that's the last point there right? We learned a lot of important techniques and concepts because the way that we work now... So, in the old days we had the warehouse, you needed a report, you came to us, you asked us for the report, we wrote it, and then you told us that was exactly what you asked for but not what you wanted so we wrote it again. And we did that three of four times, right? Today, we provide Alteryx and Tablo. We provide a significant amount of training for it and we work as a consulting group. Now, if someone comes to us and says, "I don't have time and I just need this done," of course we do it. I'm not gonna turn away work, but more often than not they come to us and say, "I've gotten this far. Can you help me get the rest of the way?" And we have several users in our organization that we can go to, and say, "We've gotten this far. Can you help us get the rest of the way?" Right? Because they have you know they've had the time and ability to really lock down on the tools and get that.

So, that's important for us because our self-service initiative has been very successful to date to the point where it's been recognized by our CEO, our COO. I have all the support in the world that I need especially from my boss and that's important, right? So, once you have that you know it's going.

We started out like I said with four Alteryx licenses and we probably had about 20 roughly around 20 Tablo licenses. Today, we have 170 folks using Tablo, we have 60 folks using Alteryx, I'm expecting that we will have 20-30 more of each by the end of this year as we continue to bring in other lines of business. They see the benefit of the tools, they see what the other lines of business are doing. The wireless people ruined for everyone, which is awesome. They put their quarterly business reports that are given to our CEO into Tablo. In the middle of the meeting he stopped them and he said, "What are you showing me?" and they said, "Oh, you know this is Tablo." And he said, "Okay, well now I want everything in this." And so, okay then I got really super busy, which is a good problem to have.

And we had expected when we picked up Atleryx, we bought an enterprise license, we wanted access to the server. We're server users. We wanted to have this. We had forecast maybe 30 users by the end of like 2018. Then blown that forecast out of the water, right? So it's 60 users like I said, we're probably gonna add another 20 in the next few months. So, if I'm out in the casino and I'm playing a game where you can bet against me you should do that.

So, we share our experience. We have an internal Alteryx users group. We're right now, I think Denali Federal Credit Union has one or two seats of Alteryx, but there's only the two of us in the area right now. If we get more traction, we'll take it to a public users group, but right now we're doing an internal users group. Once a quarter we do brown bag lunches, management presentations whenever we get the oppurtunity to. I really, really push the personal contact with my staff. I ask them to go out, sit with the people that we work with who need our help. Go meet with them face to face, every time they get that oppurtunity. Right? And we want to start looking at the gallery in our server as a repository. So, I was thrilled this morning to see Alteryx connected. It fits right into our plans with the direction that we are going, which is always nice when it happens.

So, our next steps we're gonna continue year-to-year to do this study. It teaches us. It's a good foundation. The work that you saw in Tablo in particular, by my newest employee who spent the last year interning with us. So, we give our interns an oppurtunity to hit the ground running with Alteryx, with Tablo. Getting their feet wet with the tool, right in the middle of it. So, when they graduate I just pick them up and hire them and they all ready to go.

We're gonna expand this model throughout the state. This is not the only place where have demand management issues. So, a cruise ship comes into Juno. Juno is about 20,000 people. A cruise ship like the size of, I mean the things are huge. It pulls up right to the harbor there and it disgorges 8,000 who haven't used their phone because it's super expensive to use your phone on a cruise ship. So, the very first thing they do when they get on dry land is connect to the network and call home and start sending pictures. So, there's lot of places where seasonally we deal with this sort of bursting technology we believe we can provide some insight to our network folks.

We're gonna correlate this data with other network events. So we want to take a larger picture of what's going on the whole network. Is the Iditarod affecting the network away from the race? Probably. Right? But we need to know where and how. And then we want to expand this knowledge into predictive and prescriptive models. We just put our first predictive model into play this month. So, we built a model that basically looks at our internet and our cable TV users and predicts when cable TV users would dumb cable TV because they watch TV purely over the internet. So, we're gonna see how that works out. We're running that model. It is all built in Alteryx, using our tools. So, we're really excited about that and we feel like it's gonna take us a long way and we want to take this model and go a little bit further. And really push the self-service and let our users know what they can expect to see.

And with that I'm gonna open it up to questions.

Crowd question:
Thank you Matt. Are there any questions from the audience? A silent group.

Crowd question:
Testing. The business justification for putting this all together was learning more about the main analysis or?

Matt Childs:
Yeah. It was to understand how we could assist like our engineering teams with looking at network analysis purely through the usage data. So, being able to tie it to a specific event because we are also one of the race sponsors, right? So, this is an event we've always been heavily involved in. So, the justification was this give us a lot of lockdown event at which we can look at and either validate or throw out the assumptions we've made that traffic increases x amount across the network during the race.

Crowd question:
Any other questions? I actually have a question for you. So I know you mentioned the Tablo vises in addition to putting out beautiful dashboards, what other tips and tricks can you offer the group today to get top leadership buy-in so that you can foster this culture of analytics throughout an organization?

Matt Childs:
So, I think the biggest thing is transparency, right? So, a couple of times a year I get the oppurtunity to present to our executive management team and whenever we do that we try to find where the fine line is between telling them what they need to know and telling them too much. But I've gotten to the point now where I rather air on the side of telling them too much. So, we talk specifically about the tools and what they do. We don't just say, we take it and magic happens and ta-da. We take it and I've shown our CEO. Our COO have been in meetings where I've pulled up Alteryx and run a workflow, so they can see it.

I've talked about in our consumer group we had one of our analyst work our consumer folks to take a process that was munging about 40 spread sheets every Monday morning. Taking some ridiculous amount of time about 20 hours a week and tying them all together into one data set and then extracting a report from that. Letting there analyst do what they were paid to do, which is analyze the data, not spend 20 hours a week cutting and pasting it together, which there are gonna make mistakes. So, we took that process and literally we showed it to them, so they can see it. This is what the tools do, this is why we need the training, this is why it's important to have it.

We just built this spreadsheet, ironically, we just built this spreadsheet to hand up to our management team last month with everything that we do in Alteryx and Tablo and all of the other self-service tools that we use and showing what the benefit is. How many hours we're saving each extrapolating that out to a yearly savings because I'm not a revenue stream. I'm never gonna make the company money, but I can make the company money by helping them save by cutting costs and helping them save time. So, that's how we're doing it now. So, be open and transparent. Don't be afraid to show them what is that you're doing with the tool and why it's important.

Crowd question:
Thank you and I think most would agree. Saving money is just a as important as making money. And we have a question in the back. One moment.

Crowd question:
Were you doing this type of analysis before Alteryx? And also you mentioned you do some heavy lifting on your databases. What type of databases are you using with Alteryx?

Matt Childs:
So, we were not able to do this level of analysis before Alteryx because we didn't have a spacial tool, and our usage warehouses sort of peaked to capacity. So, we're a sequel server shop. So, all of our warehouses are sitting on, and warehouses sort of a probably a misnomer. They're more like operational data stores and we have a couple of small data march that go along with them. But our usage warehouse it's pushed to capacity. It's the largest database in the entire company and I sort of treat it like it's a one-way store right now. Data goes in, kinda doesn't want to come out on its own though. You have to coax it out, which is hwy we're moving to the Hadoop platform. But yeah this type of analysis wasn't possible to us before Alteryx. The in-database tools combined with the spacial tools make it possible. We probably could've done something like this, but it would've been months worth of work.

Thank you Matt. And with that I would like to thank you for sharing your story and sharing a bit about the GCI analytics experience. I highly encourage you to connect with Matt after the session and before we leave I do ask that you pull up your conference app and rate the session so that we can continue to improve.

Thank you for attending and enjoy the rest of your day.



Los Gehts