Southwest Airlines - The Points are the Point - Inspire 2017

Southwest Airlines has been repeatedly recognized for having one of the most rewarding loyalty programs, both in terms of value and travel flexibility. The airline's Rapid Rewards program allows members to accrue points in a variety of ways, with points redeemable for flights as well as other products and services. After a year, the customer's points expire, and they can be removed from the company's books. In this session, you'll learn how Southwest used Alteryx to streamline the extremely complex data collection process and more effectively calculate the probability that a customer's points would expire. You'll hear how Alteryx ultimately saved the company $80+ million and cut eight weeks of work across 12 full-time employees.

Video Transcription

Michael Overly:
Thank you. I just want to quickly say thanks to [inaudible 00:00:07] for the introduction, but for inviting me to share the story that I'm going to today. I also want to thank Linus Cotter, I think I see him back there. He's our account rep. When he first heard the story we were sharing, internally, he recommended that we get in touch with Roman to see if we wanted to share it, perhaps, at a broader scale. I'll let you guys be the judge of whether it's worthy or not of sharing. We've been pretty pleased with the progress we've been able to make in such a short period of time.

I also... I would be remiss if I didn't thank one person in particular. He's here in the front row, Tom Lainey. For those of you who were at Alteryx 2016 Inspires Conference... Alteryx Inspired 2016's Conference. Tom gave a fantastic breakout session presentation. He's also the reason that I'm an Alteryx user. It wasn't the Inspire presentation that inspired me to use Alteryx. It was actually about six months before the Inspire Conference. I got an email invitation from Tom.

Tom is a colleague of mine who sits in a different part of the organization. We're a big organization. We have 50,000+ employees. It's hard to keep up with what everyone's doing. But Tom sent me a calendar hold, about a year and a half ago, that had a subject line, "Hold for Alteryx Demo." And I didn't... If I'm being honest I didn't know what Alteryx was. I had never heard of it. As Robin said, I was a SAS advocate assessed now, but I didn't care about it if it wasn't SAS, and my toolbox was all stocked up. I didn't need anything else, and I was going to decline Tom's appointment, and being the clever guy that Tom he inserted a line noting that there would be donuts at the demo, and that earned him a tentative response instead of decline.

So three weeks later came and went, and turned out I was hungry about halfway through the time that was scheduled for that block. I thought, "I'm going to go work my way into the back, grab two donuts, and be on my way." And as I was grabbing my donuts I kind of was hearing Tom talk to the audience, and he had his screen projected, and he was saying something about like, "Look how easy it is to take these two inputs and do a join. You can do a left join or right join." And I was kind of scanning the audience. So this the SQL for people who don't know SQL. Is this what were talking about this? I'm thinking this is the last thing I want. I'm on my way with my two donuts and I never thought about Alteryx again.

We had a former teammate of mine here, Sam Goodwell, who had at the time she was a token Alteryx in marketing. She was the only person out of a 200 person department that was using Alteryx, and Sam is, and I don't say this because she's here or because I'm trying to be nice, but Sam is the best SQL expert that I've come across in my decade of predictive analytics. She is an expert when it comes to writing complex queries that are efficient and run smoothly, and I imagine that's why they chose her to be the Alteryx evaluator for marketing, and I knew she had been doing some automation of these dashboards or taking up a lot of our analysts time.

I didn't really know what it was, but it involved Alteryx workflow that streamlined a lot of that work. She was promoted into a different position. My director came over and said, "Hey, I need you to adopt this license from Sam's computer so we don't lose it, and oh by the way on the fifth business day of each month can you kick off these workflows?" It just sounded very administrative and I said, "Thanks, but no thanks. I think you can go find someone else." He said, "Oh, by the way I have this non-refundable ticket to their annual conference in two weeks in San Diego," and I said, "Okay, I'll take it." "Would you do the fifth business day?" "Okay, yup done."

I went home I told my wife, "Clear the schedule, we're going on vacation to San Diego." Two weeks later I pack up the car, we drive to the airport. My wife and two kids and I are on a plane to San Diego. I have a plan where Tuesday morning I'll go in, grab the marketing materials, listen to the keynote, write down a few points to prove that I was there, and then I'll go be with my wife and kids to the beach. I was sitting there, I thought I was the only Southwest employee in this room of... For those that were there last year, it's a very similar look and feel to Dean's opening this morning. The setup with a big stage, a huge screen, and I thought I was the only Southwest employee in the room, and Dean gave his opening remarks, and he was transitioning into the next part of his talk, and behind him one of our 737, 800 newly remodeled planes shows up across these six movie screens behind him. And I've never been so queasy in my life.

My face just got so like sweaty I'm like, "Oh my gosh is he about to call me on screen? Like what is... Or call me on stage?" I didn't know what he could be about to say, and he went on and he said, "At Southwest Airlines they are using Alteryx to turn their analysts into data warriors," and he says it with this fervor. I was so blown away I'm like, "This guy is a freaking liar. What is he talking about?" I'm in the process of texting my leaders I'm like, "Dude, this guy is full of it. I don't know what he's talking about. Is he talking about Sam's dashboard, is that what he's talking about?"

I so I was so confused. And then he's like, "Be sure to checkout the breakout session tomorrow whatever." And I'm like... And then sure enough Tom and along with like 40 other colleagues are there, and I emailed Tom we go meet in the hallway, and I'm like, "What is all this about?" He's like, "Why weren't you at happy hour last night?" I'm like, "What happy hour?" He's like, "The Southwest meetup." We had our own meetup at Inspire 2016 and here I was thinking that I was going to have a nice vacation. I'd blend in with the crowd, nobody would know that I was there, and so I spent the next probably 80% of the next two days latched to on anyone that would listen to me in the Solution Center.

I was like a guy that had been converted. I'd seen the light, and I tried to soak up as much as I could from every single one of those people in the Help Center. I walked away from that conference with so many possible use cases in applications, and I'm about to share one with you that it was the bane of my existence for the first year I was at Southwest. It was the most unnecessarily complicated process that I had ever been involved in, and that was scheduled to kick off in July, and I told you that [inaudible 00:07:27] not only because I think it's kind of a funny story, but because it gives context to the timeline that we're talking about.

This isn't something that we had been implementing for years and it had turned into finally a matured process where we supported Alteryx. I hit the ground running a year ago when I touched down in Dallas, and I had four weeks to figure out how I was going to leverage what I had just seen to make my life easier come July. So this happened quick. It was fast, it was furious, and I think the time saving and benefits speak for themselves, but I just... I've been so.... The way in which I work has changed so drastically in the last 12 months that I think it's worth noting that everything Dead said this morning it's so true.

For anyone out there that is considering this product, don't be as close minded as I was with only using SAS or not needing another tool because it really is a powerful... Ironically the only tool that I now use outside of Alteryx is Rodeo, which if you don't know, is owned by Y hat, which as of four hours ago I just found that Alteryx acquired Y hat. Now I can say I only use Alteryx. I don't have to go to Rodeo too, it's exciting. I think that for the Python users out there, I'm sure you know Rodeo and it's... Excuse me. I love Python and if that gets integrated, then I have no other complaints to share with Alteryx.

I want to get into... That was on their website, so if I'd just gone to their website last year that I would have seen that we are in fact data warriors, and I just didn't know it. The story has to do with our loyalty program, and some of you might be members of our Rapid Reward loyalty program. We've had, what I would say is a evolution of loyalty programs in the airline industry. It's really the industry that's credited with the loyalty programs that we have today across all those industries, and I have a iteration of this deck that actually goes into this workflow that I built that uses the Google developer API to tap into my Gmail account, and index all the loyalty program emails I have in my inbox, and the remaining point balances. I just do it because I can, and it's a easy way to track all the points that are about to expire, but the point is it's hard to manage all of our active point balances.

There's a series of financial accounting practices that have evolved to manage all of these loyalty programs and whether someone will or will not use those points matters because the points are worth money, and so when you have billions of outstanding points that can add up to a lot of of liabilities. So knowing whether someone is or isn't going to use those points can be valuable when it comes to writing those points off sooner than just letting them expire, or if they never expire, then you could have those liabilities sitting on your books for a very long time.

So I'm not giving you a financial accountant advice, you should go talk to your accounting department if you want to start spoiling your points, but it's a process that is used widely across the airline industry, and our story is about how we've made ours more agile and pragmatic, and you'll see here in a second.

The way in which we brought the process to life looked great on paper, but it was a nightmare when it came to actually executing it. And then, like I said, I came back last year with an idea that Alteryx could potentially solve the problem that I had, which was just it taking way too much time. I mentioned that airlines are credited with loyalty programs. In 1981, I think it was, American launched AA Advantage program, which again many of you are probably members of. It actually hasn't... The name and branding it seems pretty consistent from '81 to present, and they've paved the way for the rest of us.

United followed suit right after, and Delta was shortly behind. Southwest arrived fashionably late, like we do to everything, six years behind with our Company Club program. I don't know how much about it. I was like four years old, so I got it... I don't know, but I do know in '96 they launched the Record Rewards program, and we still have that today. We went through one rebranding in 2011, but we like to think it's pretty good. It's flexible. I can sit here and try to sell you on it, but the reality is with that that comes a ton of point earning activity and not so much point burning activity, and that leads into the dilemma that I mentioned earlier. How do you account for all of these outstanding points that are clogging up your outstanding liabilities, and can you pair that down?

American had, in the 80's, created this academic model that the rest of the airline's adopted. It was a good model. It served the right purpose, and it provided accountants with a fixed rate percentage that they could use to spoil those points. And everyone more or less used a very consistent fixed rate, and everyone was comfortable with that. It made life easy. Now we're talking about something that was developed before the proliferation of the internet, and you're using something that hasn't necessarily kept pace with... I say you, we were using that hasn't been changed in a long time. I can't speak for the other airlines maybe they were updating theirs, we were not.

We were using the same thing that we had been for a very long time, and so in 2014 an executive in our finance department challenged the business to say, is there a better way to be doing this? Can we be smarter in the way in which we're spoiling these points and choosing which ones we spoil? And I think it was a great kind of provocative challenge, and so the only way to figure that what is to developer a perfect concept.

This is Dean and I at least years Alteryx conference. I think it was the last day, and I introduced myself as Richard Hendrickson Pied Piper, and for the Silicon Valley fans it's a fictional character from an HBO series, and Dean shook my hand and said, "Thank you for your business." And I thought, "Okay, he's not a fan of Silicon Valley." Seems like a nice enough guy. So I went on to share with him I was really excited, but forgot to ever mention that I was really Mike [Overly 00:15:31] from Southwest Airlines. So I walked thinking, "Oh my gosh, now when I meet him next do I say I'm Richard or do I said I'm Michael?" I haven't had the chance to run into him yet this year, but I'll clarify that with him if I have a chance.

So, I'm going to use Dean as an example about how do we model out whether someone's points are valuable or worthless, and by worthless I mean points that will never be redeemed, can we distinguish that. When someone goes onto and books a flight there are these standard pieces of information that we collect and add face value. They're not necessarily... They don't seem to be things that would be predictive of someone's likelihood to redeem, and so we started to develop a series of hypotheses that might be behaviors or demographics that would be more indicative of someone's likelihood to redeem.

When we first built out this model we had... Here's an illustrative... This is an illustrative list of all the potential variables, but a lot of them are intuitive. The frequency, and recency, and total sum of points a crude share someone's flying a lot, and accruing a lot then you think they might want to use them, but do they have a Record Reward credit card? Do they... How many months have they been in the program? Are they new? Are they... Have been with us for ten years? What's the average [inaudible 00:17:24] value? These are all things that we wanted to test to see if we built this logistic regression to say do these have any predictive relationship with someone's redemption activity, the variable that we were testing was like have they ever redeemed yes or no.

To play this through we actually pulled up Dean's Record Reward account balance over here, he's got plenty. Way more points then he needs to buy... I think he can buy like six tickets with that many points, but he's never redeemed them. The challenge I have going back to the process that we manage is, for those 60,000 points that Dean has, should those be on our books or should those of our books? If you flow that through to the model, actually Dean has as a very low percentage chance, 24% chance of redeeming these points. So when I get back I'll send out financial accounting department an email and say, "Hey, we got 60,000 back on our books. We'll be a little more profitable."

I'm being sarcastic, but the idea is that there are members that have never redeemed with us who based on their flight frequency you'd say, "Hey, they're still actively engaged. Let's not write these off." But if they never have any intention or don't even know that those points are available then we shouldn't be letting them take away from our profitability, and so this is one example across 100 million people to pass that through our doors every day, and how do you do this on a large-scale.

Once we got buy in from a third-party arbiters, and our financial accounting executives who had the answer the questions is it legal? Is it accurate? Do we have the resources? The third bullet is like, it was me. They I didn't care whether I had to work twice as much they just wanted me to go make it happen, and so we developed this... This is what it looked like on paper. It involved a series of checks and balances that were required by Ernst & Young to make sure that we were doing everything we were supposed to, and 2016 last year would've been the third year where we came up with discrepancies.

No matter how many QC checkpoints we put in place there was always room for human error, and it wasn't any one person it was just Dead said, "Excel hell" this morning. This was CSV file hell. There were just so many transfer and USB drives, and upload to this server, and so much room for someone to forget to change variable without any paper trail of the person before me. One our business intelligence handed that data over to me, I was tasked with appending variables, and then modeling it, extrapolating that out to the entire populating. Having financial accounting doing a series of checks and balances, and then turning it over to data scientists on Ernst & Young's team in Philadelphia who had to replicate the results, and it just never tied out.

The time it takes to reconcile 100 million... Your probabilities and finding out why one of them doesn't line up. I say it was the bane of my existence, it was eight weeks, two months of time every year that was spent running this process. After 2015 our auditors wanted to file a audit deficiency because it was taking more of their time they had budgeted to audit us essentially, and so our executives said, "So we don't run into any errors, lets do this every quarter" And so now you're talking about eight weeks times four, and so entering into 2016 we did run a in Q1, we were getting ready to do our Q2 official run, which is our audit, and I mentioned coming back in June. I thought to myself, "If... Oh yeah, I just skipped [inaudible 00:22:00], sorry.

So, I find about Alteryx. In that four week time period, I quickly piece together this workflow that you see here, and it's essentially automating all of those handoffs that were previously files that were transferred via a very secure transfer devices. USB, file sharing, and putting this in a single workflow. So we're extracting it right out of our database. It's doing all the calculations. It's eventually flowing in through to a run in the logistic regression. Taking those score probabilities, scoring back across our database. It's spitting out three files, one on Ernst & Young's server, one on our internal server, and one over in financial accounting secure shared directory.

So with the click of the button, I'm doing those eight weeks of works, and having the files saved out there, and it took 26 minutes to run what previously was eight weeks of work. I'm not a auditor, I'm not an accountant, but the transparency that this brought to the table it spoke for itself. There was not need for anyone to audit me or to audit someone on our business intelligence team. They needed to audit the workflow because if something wasn't adding up it meant that one of the calculations in this workflow that was documented here was wrong, and so you can easily follow the paper trail if you will with this process.

So I went back and I shared this with our accounting department, they didn't care as long as it was right. They just wanted it to be right, so once we proved that it was right. Once we could show the auditors that everything was properly scripted, we can run this week if we want to and see if the probabilities are changing or the redemption activity has changed enough for us to be more conservative or aggressive in the way in which we manager our liabilities. And we can automate and have it run on our server, and just the fact that we have what once was two months of work boiled down into 26 minutes it freed up the need for... Across four organizations we had ten full-time resources that were involved in this process. A series of contractors and third-party fees that essentially went away.

They weren't made obsolete is just meant that they could then go focus on all that backlog of work that they couldn't do in their normal lives or at least have until the second shift to get chipped away at their outstanding list of action items. I think the key here was that this eliminated the not fun stuff that was part of my job, and it allowed me to focus on the things that we're being asked of me that I just didn't have the time to do.

It wasn't just the time savings improvement, the previous process ran on a sample of 2 million customers, and then was extrapolated out to our entire population. This new process runs on our entire database. There's no room for biased sample size or a biased sample rather, and so what we saw is that we were still being a little bit too conservative, and when you model it out on the entire population there were a lot of points that we didn't feel like were going to ever be used. The net impact of doing this over the last 12 months has been over 80 million dollars worth of liabilities of these spoiled points we're able to get, "Taken off the shelves." If you will.

Without getting into the nitty-gritty of our financial... That had a significant improvement of our bottom line, and I think it would on most peoples [inaudible 00:26:19]. Again, not being an account, the whole idea that it's kind of like funny money. Liabilities that we don't think we'll ever use. And the reality is if we were wrong and all of a sudden tomorrow they did get redeemed, then we would just take a negative hit. It's not like we're... They're all part of the generally accepted accounting principles that the securities and exchange commission acquires businesses to hold to, and so I think in cases where points have a shelf life, ours do. They expire after two years inactivity, we can be a little bit more aggressive because the reality is they would roll off on their own after two years, and we can kind of go back and check, but if you have points that never have a expiration date it can be a little bit more tricky.

In that Gmail workflow that I built, I found 34 Papa John's points. I didn't even know belonged to Papa John's loyalty program. I had 34 Papa John's that I found from 2009. They never expire, and you only 25 apparently to buy a free pizza, and so two weeks ago I ordered a free large pizza from Papa John's using 25 Papa John's points. I didn't know. My point is if Papa John's had spoiled my points, which they probably should have, because I couldn't remember the last time I ate at Papa John's. And I literally stumbled across them in my inbox.

I'm not saying that you should go talk to your department about whether you should spoil your points or not, this is about how Alteryx saved us a ton of time. Again, it's something that really launched in a four week period then has matured over the last call it 11 months, and it's just been a wild ride. Tom's team was recognized last year for all the ways in which they're implementing this throughout the business. I'm just in marketing. I know that everyone's doing some exciting stuff.

I think we have time for ten minutes of Q&A if... Happy to talk about this or other ways in which we're using them. If you have any.

Crowd question
Questions from the audience?

Michael Overly:

Crowd question
[inaudible 00:28:54]. How have you reconciled the points after the fact that they've been spoiled and then someone actually does redeem? And how have you gone back... Do you run the flow on a daily basis then to check that or how have you managed that?

Michael Overly:
Yeah, yeah. So in the proof of concept... That's a good question. So in the proof of concept what we did to... The way we defined accurate was we did exactly that. We found people who had never redeemed and let their points expired, so there's that two week period. And then we went back and said, "Well okay, if the model had predicted that they would expire, did they actually expire?" And what we found was that using the two million person sample we were at 85% accuracy meaning the people who we'd flag actually did end up expiring. So even the 15% that we were off, they were false positives, so it was you know... If anything in that we were being ultra-conservative.

It got into the low 90's when we moved that to the entire population. It's easier when there is an expiration date because you can go back, and we don't actively go back and audit that. We just need to periodically check that we're not way off. Anything else, yeah?

Crowd question
What's your [inaudible 00:30:16] after a certain point [inaudible 00:30:21]?

Michael Overly:
Threshold in what sense?

Crowd question
Like at what point did you decide okay [inaudible 00:30:28].

Michael Overly:
Oh, I see. So like in Dean's example. His 24% probability. There wasn't a... Another good question. I thought you guys were paying attention. So there were two statistically valid options we had. One was Pickett threshold, 70%. Anyone over this score gets spoiled. The other is take their probability and multiply it by the number of outstanding points they have, and what's remaining we'll keep those. So in Dean's example we'd take 60,000 and keep everything but the 24% of the 60,000.

So we tried both of those alternatives, and it... But to answer your question the threshold we played around with was 70%, and that was...I can't cite a scientific reason why we picked 70, but it's not the way we [inaudible 00:31:35].

Crowd question
Was it more of like an automated [inaudible 00:31:36]?

Michael Overly
I imagine. When I say... I just can't recall the reason, but again I wouldn't be the one to advise on that. It was just one of the options we had.

Crowd question
I believe we have a follow up question here.

Michael Overly

Crowd question
I can just take it offline.

Crowd question
Okay. We will take it offline. Any other questions? I have a quick question for you.

Michael Overly
Sure, sure.

Crowd question
So I know you thought you were the sole Southwest analyst. Do you have any advice for other folks here in terms of connecting with other Alteryx users within the organization, and just sort of how to best share tips and tricks?

Michael Overly
I do. I have to be honest because I have colleagues in the room that can keep me honest. Here's what I'll say to that [inaudible 00:32:34]. When I talk about the pros and cons of SAS against Alteryx, I think if you had talked to me two years ago I would have cited the SAS user community as the reason that they're superior, is they have a support network that it is so rich with information that there's never a question that you can't quickly get an answer on because of that community, and it was of one of the reasons that made it such a powerful tool for me. And that was before I had found out about the Alteryx user community, not just within Southwest, but with just the general community, and I go there, not because it's the first Google search result because it's a place where I know that people have shared their solutions.

I have to say that if you're new to the tool like check the community because odds are someone else has run into the same issue. Within Southwest, I would say we're still, clearly we're still maturing. If I can show up somewhere, I think I'm the only one, but I think Tom's team has developed a good... Leslie Mitchell too, she's our Alteryx administrator. We have emails [inaudible 00:33:57], we have onsite support, I think it's really a matter of getting people up and running on it, and then questions are good because it means they're using it. Here we have department champions that they can go ask.

I think it's still taking shape, but having a self service support group on the Alteryx community, personally it's been where I usually go first, and if I can't answer it there a last resort just because I hate... To me asking a question is like a sign defeat, that I couldn't figure it out on my own, but I know not everyone is that way. I'm sure that you guys get questions. Anything eels you want to add to that?

Crowd question
[inaudible 00:34:44]. Within Southwest we have a edge analyst, we call it edge analyst meeting, and it collects all our analysts from all over the company in one spot quarterly, so when you see with other... And that Michael, you presented that. That's how I found out about this presentation. I'd never known Michael until that meeting, so that's kind of how the word spreads about what other people are doing within our department. And I think that's kind of how a lot the Alteryx gets spread as well. When people see other analysts doing those kinds of things.

Michael Overly
[inaudible 00:35:25].

Crowd question
Yeah, here.

Michael Overly
Leslie Mitchell.

Leslie Mitchell:
We went from 30 Alteryx licenses in February 2016 to about 375 desktop licenses now, so we've been kind of behind the curb in setting up these kind of user communities. But now we do have an IBM connection site where people can go and share their workflows, and share tips and tricks, and we also have other levels of support. Level one support is your department champion, and then the level 2 support would be somebody in IT. So we're still rolling all of this out because we've grown so quickly, but we have lots of great ideas we just need a bigger team to put them into place.

Crowd question
I was the first... I mean two years ago I was the only had the expire. I was the only one. That was it. I'm going to brag, yeah. But there's 40 of us now today.

Crowd question
Thank you for that. We have a question back here.

Crowd question
So I like Alteryx, and I have some colleagues who really like SAS, and I'm wondering if you can share what was the hardest part of that transition, and what helped you through it?

Michael Overly
I was asking myself this when Dean made the comment about the guy in Boston saying that Alteryx could take over SAS. Up until maybe three months ago I would have been lying if I had said that Alteryx was a superior solution to SAS. I think I had adopted it because it was the tool that my employer had invest in, but if my boss had come to me and said like, "Hey, we just freed up some money in technology for SAS." I would've been like, "Yeah."

Honesty, even today, knowing though [inaudible 00:37:33] position. Like my list of excuses is dwindling. All of the reasons I used to love SAS and enterprise guide, and... We're using Malcolm Gladwell's definition of an expert being 10,000 hours, like I am a SAS expert by no other reason than the month time I've sunk into learning the nitty-gritty details of the programming language. I wouldn't have to do that with Alteryx. I could have spent eight years of grad school doing something else with my time. I think that if Jim Goodnight doesn't have Alteryx on his radar then he should because I think that... It's not faster, it's not strong, I mean those are a result of the IT resources that you put behind it. To me I can't come up with an answer.

Edward Givens is here. He's a skeptic, like I was. He sent me a text this morning saying that he thinks... He hasn't installed Alteryx on his computer yet. He came here without ever opening the software, and he's on my top ten list of the smartest people I know. He got a perfect score on the Wonderlic test, which means he's like in the crazy wild genius category, and I've witnessed it with my own eyes. He's a freak of nature, and he told me this morning that he thinks Alteryx is a powerful data blending tool, but he's not sold in the predictive capabilities. I was like, "Alteryx doesn't have predictive capability. It leverages R, which he loves." He lives in R. I'm like, "You can have your R console right in there and it can do everything that you're doing there."

The benefit is all that data's flowing in from all your disparate data sources, and no more ODBC connectors. So now he hasn't texted back yet, I'll see what his response is. My point is Alteryx isn't a language, it's not a programming language. It's a tool that just makes the things you are leveraging easier. Whether it's something you really like in Excel. Like I have this ongoing challenge that there's nothing that Alteryx can't either solve for or enable you to do and I honestly... Now with the Y hat acquisition I... And I'm not getting paid up here. I don't get kick backs from Alteryx. It's just I gush about them because of it making my life easier over the last year and being sold on its capabilities.

I have no... I don't care if you guys go with SAS or Alteryx, but I can tell you I don't care today whether Southwest invests in SAS or a year ago. That wasn't the case. That's what I can say.

Crowd question
I have a quick question follow up on the SAS, Alteryx comparison.

Michael Overly
Yeah, bring it.

Crowd question
First of all, what SAS products were you using? And secondly, is it the contrast between the two as... They're not really mutually exclusive right? They can be complimentary.

Michael Overly:
By all means. That's a fair preface. It absolutely depends on what you're using SAS for and to question, I'm a SAS enterprise guy. I don't know what version I run, at the time maybe 5.3. I don't know what they were up to, but it was sitting on top of three terabyte server underneath my desk that I had exclusive rights to, and it processed billions of records of HBO at lightning quick speeds, and I like that I could do the analytics and modeling in database essentially. So that's the use case I'm talking about.

I know that folks in technology have... Whoa. I thought there was a step right there. That would've been funny. I know there's other use cases that one can solve for another, but for me as a data scientist dealing in predictive analytic my use cases for SAS versus Alteryx they would fall in the center of that venn diagram.

Crowd question
Any final questions?

Michael Overly:
Well thank you. I appreciate the time. Feel free to reach out if... My contact info if you just want to further debate SAS versus Alteryx. I think you're supposed to ... I'm supposed to remind you to rate me as... Give feedback on our session so that they can...

Crowd question
Yes, awesome.

Michael Overly:
All right, cool.

Crowd question
Thank you Michael. That was a fantastic session.

Michael Overly:
Yeah, I appreciate it.



Pour Démarrer