DELL | EMC - How Soon Can You Get Here?! - Inspire 2017

When your access to data is compromised, so is your understanding of your business. Alteryx and Tableau play a crucial role in ensuring that the DELL|EMC service force swiftly supports customers during times of need, when fast response time is essential for customer satisfaction. Learn how the DELL | EMC Field Scheduling system is optimized by utilizing Alteryx workflows to blend and analyze disparate data sets, and Tableau visualizations to innovate an informative, customer-friendly web interface. Steve will also share how DELL|EMC utilizes unconventional inspiration sources like pizza, TV and lemonade for guidance in providing outstanding support for their customers!



Video Transcription


Steve Scales:
My name again is Steve Scales. I am a Senior Data Intelligence Engineer. I'm not sure who came up with that whole thing. It was just Data Intelligence Engineer, and then I used the acronym DIE, and I thought, "No, no". So, I like... They put senior in front, I just think it sounds cool.

But, I get to play with data a lot. And I've been working for EMC for about 13 years, and when I typically tell people I work for EMC, you know friends or family, "Where do you work?".

"I work for EMC".

"Oh. Well, what's EMC?".

I get a lot of that. Nobody really understands, or has heard of EMC. So, I explained, it's data storage, we do a lot of data storage. And I get the blank, kind of deer in the headlight, "Ah, data storage. Yeah...".

So, I'm like, " Well, it touches your life in every way, you just don't know it. Everywhere there's data, it's the physical storage of it. It's the cloud. If you think about the cloud, the physical data that's in the cloud, is stored on physical pieces of equipment. EMC makes that equipment. We are the cloud.

And then everybody goes, "Oh! Okay, all right".

So, I've always had that for the 13 years when I try to explain that I work for EMC. And then, this past year in 2016, there was a big merger, in fact it was the biggest tech merger ever, and Dell bought EMC. And so now, after this merger occurred, I work for Dell.

So, now when I have family or friends ask, "Where do you work?".

I say, "Well, I work for Dell.", and now I get, the light bulb comes on right away. They say, "Dell! I know Dell. I have a PC problem, can you help me with it because..."

So, from one extreme to the other. So, one of the things... We're going to talk about a few things today, and one of them, we're going to talk about is the difference between aggregated data, and detailed data, and how I used lemonade to get past some of that. So, we'll go into that in a little more detail.

But, once we're able to get some of the detailed data, I was able to open up new doors, and look at new measures inside of our corporation. And so, I'll show some of those, some of the insights that we've got to. Using both Ultrix and Tableau. And so, I'll show you the workflow that I built for Ultrix, because you guys probably haven't seen enough work flows this week yet.

So, I'll show you that, and we'll just break it down a little bit. And then I'll talk about we did some unique things, doing some prototyping. So, when we talk about the lemonade though, one of the things that I always had trouble with is, I wanted to do analysis on the data sets that we had. We have these data sets, and typically I was doing them in Excel before Ultrix or Tableau came along. And, all the challenges, we're all aware of the challenges with that.

But then once I got the new tools, I said, "Okay, I want to go find some new insights in our data. I want to go explore, and so I'd go to IT and I'd ask, "Okay, I need this data set".

And they said, "Okay, how much of that data set do you need?".

"Well, I need all of it".

And they would say, "Yeah, but it's huge. You'll have to cut it down. You just need a portion of it, so how much can we cut it down for?

I'm like, "No, no, I have these new tools. I can handle all the data, give me it all".

And they would say, "Oh, silly man, silly man... You can't have all the data. So, I'll tell you what? We'll help you out, we'll aggregate it down to whatever you're looking for. So just tell me what you're looking for in the data, and then we'll give it to you."

I said, "Well, I don't know what I'm looking for yet. That's the whole point, I want the whole data set, so I can explore the data, and find some new insights".

I just wasn't getting through. They couldn't understand that the whole, you know, aggregating the data... That causes problems.

So, I said, you know I've been to a lot of the conferences. I've been to the Tableau and the Ultrix conferences, and one of the themes is always 'Tell a story with the data'. It's important to be able to tell the story. So, I said, you know what? I'm going to use that whole tell the story thing, maybe I can get through to IT with a story.

This is the story I came up with and you're welcome to use it if you have the same challenge. It starts off with a school district and this school district had three schools. The superintendent of the school district used to go and visit the three different schools. At lunch time, she would notice that there would be students sitting together, and there would be some students that were off alone, and they weren't eating.

She asked questions about it, "Why are those students not eating?". Said, "Well, those are our underprivileged students. Our funding got cut, we used to have hot lunch programs for them, but they can't afford hot lunch, and a lot of them, truthfully, they can't afford to bring a cold lunch, a lunch from home. So, they spend the hour, not eating, when all their friends are eating".

This broke her heart. She said, "OK, I know the funding got cut, but maybe there is something we can do." So, she came up with an idea to have a contest. She said, "I'm going to have each of my three schools compete, and they're gonna do it with lemonade stands. They're gonna raise money at these lemonade stands to help with a hot lunch program".

So, she said, "Every school is gonna have a lemonade stand, and when they sell a cup of lemonade, they're gonna have a three by five card, and they're gonna write some information down. They're gonna keep track of all the sales that they did. We're gonna do this for an entire month. We're gonna do four weeks of selling lemonade.

The four weeks went by, she said, "OK, now I want to know, which school did the best". She said, "Send me all those cards".

And they did, what I got, was, "oh, we can't send you all those cards. You know how many cards there are? It would take up your entire office, with all these cards. We got boxes and boxes of it. Tell you what? Instead of sending you all the cards, we'll have each of the schools just take a card and kind of condense their data on to one card for every week".

She said, "OK. If I got that, I'd get basically three piles, four cards in each pile for what they did". She got those, and she laid them across her desk and said, OK, let's see, which school did the best.

School A, they did 50 the first week, 50 the second week, 50 the third week, and then they ramped up and they doubled it in the last week. She said, that's pretty good.

School B, she looked at that, and that was nice and steady. 50 and 50 and 50, all the way across.

School C, A little bit of a slower start, but then they kind of ramped it up and got going and did 50, as well, all the way across.

Having these cards across her desk, let me ask you guys, looking at the data, which school did the best?

Crowd question:
(Inaudible rumble from the audience).

Steve Scales:
A? It's kind of obvious. School A did the best. She said, "I want to recognize School A, for doing the best. But, I also want to kind of do like a little ceremony and recognize the other schools for participating, as well".

She went to each of her school's and she said, "Can you send me a picture of each of the teams so I can hold that up and show people?".

She said, I'm ready to recognize, let me see those pictures that came in, and when she started getting the pictures, she learned there's a little more to this story that I had originally thought. She got the picture of School A, and it was a team of five students. She said, all right, well each of those five students, they sold you know, about fifty per student. That's pretty good.

Then she got a picture of School B. She went, whoa! I got a lot more volunteers with school B, got twice as many volunteers with school B. Hmm, interesting.

Then she got the picture of School C, and she said, Whoa! Wait a minute. One? One person? So, she picked up the phone and she called the principal at School C and said, I got the picture. I'm only seeing one person, what's the deal?

They said, "Oh, yeah. We had a hard time getting any volunteers, and the person that actually ended up volunteering, that's Patrick. Patrick felt very passionate about the whole hot lunch program because he, himself, is one of the underprivileged students. And, he has a younger brother that's also in the school district that is, as well. Couldn't get lunch, couldn't eat lunch, he's one sitting off to the side.

She found out that Patrick, not only went to school, but he lived with his mother, and his mother was disabled, so she couldn't work. After school, Patrick would go to work, and he provided for the family. He provided for his mother and his little brother by being a janitor at a exercise facility. So, he really didn't have a lot of time to do the lemonade stand, but no one else was volunteering.

He said, "I'm still gonna give it a go, I'm gonna see what I can do". He set up his lemonade stand outside of work. He got done with work, he set up by the main entrance. That first week, he sold ten cups of lemonade. He was there every night, and he sold ten cups. He was disappointed.

He was like, "All right, not only am I not selling a lot, but it's taking me away from my family. But, I really feel this is an important thing. I wish there was more I could sell, I'm gonna make a goal, I'm gonna double this for next week. Hopefully, I can do it a little sooner, so I can get home and spend some time with my family."

Then he thought about, well, how am I gonna do that? And, as he was outside thinking about it, he noticed that people were leaving from the rear exit. And, he said, oh, the rear exit, that's where the hot yoga is going on. He's watching all these people coming out, they're all sweaty, they're hot, they're walking....

He's like, huh? So, he set up his lemonade stand back there. He was able to sell fifty cups in just one night. He said, all right, I more than doubled what I wanted, I made my goal. Now I can spend the rest of the week with my family.

Hearing this, the superintendent said, all right, I'm gonna change up my presentation a little bit here. For recognition, I do want to recognize school A. They did sell the most, so I'm gonna have a special recognition for them.

School B, now that I saw their picture, I have more input in there. They had more volunteers, that says something, so I want to recognize them for having more volunteers.

With School C, she wanted to make that one special. She not only wanted to mention Patrick because it was the greatest number sold by a single student. But, he was the most independent by being the only one, and he had the best creative marketing idea as far as selling the lemonade.

When looking at the cards across her desk, she only saw a piece of the story. But then, once she got all the other information, she learned...The rest of the story!

Now, we probably have some people in here that don't know who Paul Harvey is. Paul Harvey was a radio guy, and that was his catch phrase that he would end with. With his radio programs, he would tell a story every day, and so, for anybody that doesn't know who Paul Harvey is, and I just did that whole 'The rest of the story', I nailed that. Sounded just him.

Crowd question:
(laughing).

Steve Scales:
For anybody that does know who Paul Harvey is, and had heard that, I really apologize for butchering that. That was pretty bad.

Now, what does that have to do with data? Well, with our EMC, we have a huge service force. We have thirty-one thousand service experts, and twenty-eight thousand partners. We support customers across the globe, 165 different countries.

What that does, is that generates a ton of information. Because, we handle tens of thousands of customer interactions every single day. One of the systems that we have that helps support that, is a scheduling system. We use that scheduling system to be able to identify where the customer's needs are, and then, where our resources are to help those customers. Then, match those up in order to make sure that we can get there as quickly as possible and resolve the customer's issues.

But, in order to do that, the system needs some information as well. We feed it different information so it can do these calculations. I said, you know what? That's a lot of information, I might be able to use some of that information. The machine's using it for certain things, but maybe I could use it for different things, as well.

One of the things that we do with that system, is our service engineers have a mobile device, and they give us input that feeds to that system. Typically, we would use two pieces of that data. We would use the travel time, so we would know how long it'd take to get to the service event. Then, we would say, okay, how long have you been there? And that's how long it takes to do a repair.

But, when we started looking at the data, we were finding these huge swings, and trying to identify where those swings were coming from as far as time. What we realized was, even though we're using these, these are the aggregated pieces, there's more detail, but we weren't using the detail because it's too much data. Well, it's not too much data when you talk about using Ultrix and putting it into Tableau.

So, we actually gather five different time stamps from our customer engineers. They tell us when they are leaving to go to a site, and they tell us when they arrive on site. We know that's their travel time. When they're on site, they're not doing the repair yet, sometimes they can go right to the machine and start doing the repair, but sometimes, they may have to interact with the customer, they may have to wait for the customer to power the equipment down, they may have to identify where the part that was supposed to be sent to the customer for repair, where that's located. A lot of times, they have to go through security systems before they can even get to the equipment. So, all that took time.

When they're on site, they don't actually tell us that third time stamp of start work until they're physically in front of the box, ready to start doing the repair. Then, when they're all done with the repair, the machine's working, they tell us that it's repaired but they haven't left the site yet. There still might be other things they need to do with the customer. Test it, make sure that it's powered back up, in different, they may do... London, for example, they'll have tea with the customer. That was time, and we can capture that between then and when they actually leave the site, which is that fifth time stamp.

Since, they were giving us all those time stamps, we could say, all right, now we know what the repair time is. Knowing what the repair time is, we can take some of that data from our system, and we can do some calculations if we blend some other data sources with it.

We took those two time stamps they give us, the actual repair time, and then we take logistics data and say, all right, what part were your replacing? So, I know how long it took, I know what kind of part we were replacing. Then, we take from our duration table, which gives us the product that we're working on. That duration table is what's feeding that system, so we gotta feed back into that duration table and update that every once in a while. Finally, we grab our SQL database, and that gave us some customer information that wasn't in our scheduling system. Our scheduling system was all lat and longs that didn't tell us who the customer was.

Based with all that information, we put that all together, we could get some insights into our data. We used Tableau, and we brought up... Actually, our engineers use this Tableau dashboard. Using the product and the part type, and then we can compare repair times. So, I'll give you a little closer look at that, at what were talking about here.

We use the box pods, box and whiskers, but behind that, you see there's an orange bar. The orange bar tells us the volume of that particular product and part. So, we can see which one's matter, and which ones have lower volumes with them. Then, of course, we've got the median. The median was important, because when I talk about that system and knowing how long it's got to be to do a repair, the system needs to know that, so that median tells us how long it should take.

The, of course, we've got our core tiles and our outliers. But, the median, we were actually using to feed back in to our system, tell the system how long something should take. We had to do that prior to Ultrix, we had to do that with Excel. We were doing it with Excel and bringing in different pieces and trying to manually map all that data together.

We could do it, we could only get a hundred thousand rows at a time, I don't know if you've ever done that. Well, that's all I can get, so I'll pull that, and I'll pull that, then I'll put it all together, then you get duplicates in there, and it's a mess. So, you have to clean that all up. It would take us about thirty-two hours every quarter, because we updated every quarter. So, doing it manually, thirty-two hours.

Once we got Ultrix involved, we were able to get that down to fifteen minutes. We've heard that story, using Ultrix, a lot. That one really means a lot to me, personally, because I personally was the guy doing the thirty-two hours of work. By getting it down to fifteen minutes, it bought me so much time back.

If you want to see the workflow, I'll show you the workflow, go behind the curtain and show what I did. It's not super fancy, but there's a couple things that I do, that I typically show people internally in the company when I'm helping them learn Ultrix. I do training on Ultrix and Tableau, as well.

One of them is, that first little box up at the top. This is a best practice that I use, feel free to use it as well. Whenever I start a workflow, I put a comment box in the upper left-hand corner, and then I type in "the runtime=", and then I update it after it runs. Because, when I'm working on a workflow, I know this is going to take me about five minutes every time I hit the run, or two minutes, or thirty seconds, but two-weeks from now, when I open this workflow back up: Is this the one that took thirty seconds? Or is this the one that takes ten minutes? I never could remember. By opening it up, I always have that little box up in the corner. What's neat is when I start to get workflows from other people now, I'm starting to see their workflows have that box up there, as well.

I also asked the product development team to make that automated so that we could have that. They thought it was a good idea. If you guys like that idea, put it on community, let's see if we can get some votes up for that.

The second thing I do, is I put comment boxes for anything that might need for me to trigger in my mind. I always color them yellow, so that before I run it, it's things like, go make sure that this file has been updated and is in the right location, 'cause it's gonna feed from there. Go make sure that your data if you're doing a certain date range, make sure that date range is updated, and things like that.

Then, I've got my inputs from those different data sources that we talked about. I separate them out because I want to do different calculations based on the part-product combination. Then we also have service events where there is no part, those are a different type of service event. I want to analyze those differently, so I separate those out and I have them go through two different flows.

I also take those outputs, instead of doing YXCB, well I do YXCB outputs, but I also do TDE outputs. If you're not familiar with a TDE output, it's for Tableau and if you're using any of the data for Tableau, I highly recommend using a TDE, which is a Tableau Data Extract. Because it'll make your Tableau processors or desktops run much, much faster.

Then, I go and I filter out some of the outliers. I do some calculations as far as vidpoint, because we need to also identify where that certain ticket, that certain service event occurred. When did occur? Sometimes we might start at 10:00 pm one night, and go until 3:00 am the next day. Was that on this day? Or was is on this day? We have to narrow it down, I use what's called a mid-point, I take those five time stamps that we talked about, I find out where the exact middle is and I call that the mid-point. Wherever the mid-point falls, is where the majority of the ticket happened, so if it was on this day, and the mid-point fell on this day, I would call it this day.

I go back and say, OK, well, I need to make sure I got enough records when I go and update, how long it's gonna take for that product-part combination. If I've got 200 or more records, some of the parts, no problem, that happens in a day. For that quarter, I'll use the previous two quarters, so six months worth of data, and figure out what the median is. But, if I don't have 200 events, then I was getting swings, so I want to bring in more data, let me go back another quarter. I go back one more quarter, and say take three quarters worth. If I get 200, then I use that, if I don't have 200, I go back another quarter, and I grab that. And, I keep doing that until I get to five quarters. If I don't have it in five quarters, I say, bring me everything. Which I can do now, 'cause I've got the data.

We have to send it to... The ability to have the system updated. So, I want to notate the product-part combinations and whether it's the same. We checked it, for the last six months, it's the same as it's always been, so it's nice and steady. We don't have any change. There are some product-part combinations that don't exist anymore, so we can take those out. Then there are the ones, we say, we need to update this. It's gotten faster to our service engineers. The engineers, for the machine may have changed something in bracketing so they can take it out and put it in a lot faster. Or, they may have code upgrades that they have to do, so it got a lot slower to da, so there might be some differences there.

Finally, it automates and email, sends it out, and the person that gets that email then does the update into the system. That's internally. We use a lot of different things internally with Ultrix.

We also got some feedback from our customers that said, you know what? We want some information about our events. We had one major customer that a colleague of mine sat down with, and they brought out a television TV Guide channel. They said, this is what we want to see for our equipment.

And we said, "A TV Guide?".

They said, well, not a TV Guide. That's kinda how we want it to look, though. So, instead of having the channels, we want all our boxes, 'cause I have a box at this location, and a box at this location, and a box at this location. I'm trying to calculate them all, so I want to see each of those all in one thing. Then, going across the top, how we have the times, I want to be able to see, when is our service technician gonna come to those different locations. And, when they're there, what are they doing. What's the status of that event?

He said, OK, that's very interesting because, not at the same time that they were asking for this, we were talking about using those five time stamps. We said, you know what? Have you ever ordered a pizza from Domino's? Have you guys ever ordered pizza from Domino's online?

If not, you get a bar, and it's so cool because now you can tell when the pie is in the oven, when's it being quality checked, when's it out for delivery, you can see who the person is that's coming to your door. It's all one, two, three, four, five, we said, we have five time stamps that we get, we could do something like that. We could show our customers where we're at with our service events. Then, we could put it into a format like that television guide. Yeah, let's do that.

We got the request to say, yeah, let's do that, and by the way, we're updating our portal to our customer. We're gonna want you guys to do a little prototype, because we want you to build this in for this portal. We said, OK, yeah, we can do that. It'll take us about six to eight months to do a prototype, you gotta gather in all the people that know the data, you gather in all the customer information, you gotta lay out different things.

And they said, "No, no, six to eight months, no. The portal is going live in a month. We need you to have it all done, in a month".

We said, "What? No, we can't do it in a month. Well, wait a minute, you know, maybe we could do it in a month".

If we limited our team down, and just got like our SME for the data, and we could build a prototype using that SME for the data. And, then we'd need an SME for the customer. We could bring in a data source, I said, hey, Ultrix. We can bring in those different data sources, we can use that as our prototype data source. We said, OK, well then, how is it going to look to the customer?

Well, this is what we need from the customer. We actually brought in a Tableau consultant from Tableau, they offer that service. We had them help us build a customer interface. That customer interface gave us all of the different data points and kind of a Domino's type thing at the bottom. And, by the way, it gave us a little calendar at the top too, which was kind of a bonus thing, so they could go in and see what was happening in the future for like their installs and upgrades.

With our team, in order to do that prototyping, not six to eight months, we did it in twenty-four hours. In twenty-four hours, we had a template that we could say, here's how it works, here's what you click, here's what you see, here's the changes that occur. Now, the portal that's being used, is actually in D3.

We don't actually use the Tableau dashboard as our customer interface, but it was our template. We were able to give that template to the designers at D3 and that - that you see on the right-hand side, is what our customers see today. We were able to do that, and get it turned around within that one-month process to get it out there and online. Our customer's have said that that has been incredible.

In conclusion, I just want to summarize what we did. We talked a little about aggregated data, and how much aggregating can change the story. We talked about our analytics and some of the data, or the insights we were able to see with our business by using that detail data.

We got inspiration from different sources like pizza and television to be able to do some rapid prototyping, things you don't normally think of when you're talking about Tableau and Ultrix. With that, that's my spiel, and if we've got any question, now would be a good time to ask the questions.

Crowd question:
(clapping from the audience).

Steve Scales:
Aw, shucks. Any questions?

Crowd question:
I'd just like to say that that's a very impressive presentation, and actually the lemonade, that first story, that's very powerful. Because, I've always been in that position, on the business side where I'm asking for the data, but IT is always pushing back.

Steve Scales:
Yeah, and they think they're helping by pushing back.

Crowd question:
Yeah.

Steve Scales:
I'll help you out, I'll go aggregate it for you.

Crowd question:
Yeah, thank you.

Steve Scales:
Well, thanks. I appreciate that. One of the... When you talk about the story, because I did, I wanted to come up with something to kinda resonate outside of data and talk about a story type thing to get that. But, I had done this presentation once, and I was at a bar afterwards and somebody said, hey, great presentation, let me ask you a question, when you gave that to IT, did they give you the data?

I had to stop and think about it for a second, well, let's see, I know how I got the data, you know, no, they didn't give me the data when I brought that story up. But, what I ended up doing, is going into our system, into the production, finding out that they were doing snapshots, and I took the raw data from those snapshots and that's what I was pulling in.

I said, I was Patrick, I was going to the back door where hot yoga is going on. That's what I did, so, the story came in handy, just in a different way than I thought.

Thank you.

Crowd question:
I might have missed this part, but is the story a true story?

Steve Scales:
It is not a true story, it is a story that I wanted to make sure would tug on the heart strings because that what I learned in the different Tableau and Ultrix conferences, is you gotta make it personal, you gotta make it hit home. So, that's why I did it. Could be a true story, I was doing this one time, and I saw somebody sniffling. I'm like, no, no, it's made up! I made it up!

Crowd question:
Hi.

Steve Scales:
Hi.

Crowd question:
So, your flow about those five time stamps, I work on operational analytics and it's incredibly similar where you know, you've got the first log time, you've got pick up time, first warm response time and everything like that.

Steve Scales:
Yeah.

Crowd question:
What are you using to kind of communicate that at the end? You know, total resolution time. Are you putting in a dashboard? And are you showing it operationally, or are you showing in more client facing, or perhaps both?

Steve Scales:
The client facing we do now with the prototype that I showed you earlier... Or, what we do now. The stuff that I was showing here, we were using for engineering, because they like to know, when they make a change to the product, what does that do to the field? Does it make it go longer or shorter? But, when you talk about those five different things and what we use them, and how we use them internally, there's probably about thirty different ways that we use those five time stamps.

For the presentation, I had to pick one, so that's one. We use it in a whole bunch of different ways, one of my favorite one's is to see the movement of our service force globally. I built a Tableau dashboard, it shows the globe, and then I use pages which I think is a feature in Tableau that's very underrated, but if you're familiar with Tableau, you can put pages in there. I would show it in fifteen minute increments of where those five time stamps were for every single one of our customer engineers.

So, when I talk about those tens of thousands of things, I had dots all over the globe. You would see them in the America's, and then as... I had a little time bar at the bottom, as it would go across, and it would get late at night for the Americas and the dots would start to fade. Then Europe would come on line, and then as the bar went farther, Europe wend online, you could see Asia come online, then as Asia faded out and it looped back around, you could see the America's come back on line.

It gave us some insight there, as far as the movement of our CE's, just another example of it.

Crowd question:
Hello, congrats for the great presentation again.

My question will be, you're leveraging out Ultrix and Tableau for building tools for the customers and also visualization for the corporate.

Steve Scales:
Yes.

Crowd question:
My question will be, I guess, are you also utilizing and leveraging the tools for some type of predictive or prescriptive work in terms of utilizing those other tools that they're having, teaching us here at the conference. Or those types of analysis?

Steve Scales:
Yeah, we're just starting to do that to a degree, right now. One that comes to mind is a... We're trying to predict the work load for a given service area so we can identify, do we need to increase our manpower in that area? Or decrease our manpower in that area? We want to see what's been happening, and what the predictions are. We're just starting to get into that right now, but that's a perfect example of how we're using predictive.

Yeah, good question.

Crowd question:
No more questions?

Steve Scales:
All right. Well, I want to thank everybody for coming. I really appreciate you coming and listening to me and y'all have a great, wonderful rest of the conference.

^Top

ERLEBEN SIE
DIE WELT VON ALTERYX
SELBST.

Los Gehts