Schneider Electric - Rapid Fire Design Tips for Using Alteryx in a Team Environment - Inspire 2017

You know that Alteryx can save you hundreds of hours each year. But as your organization grows, and more users across the enterprise adopt Alteryx, you need to think about how to use Alteryx in a team environment. In this session, we'll discuss 15 rapid-fire design tips for using Alteryx at scale and how you can transform Alteryx from an individual tool to an analytic workbench that drives enterprise-wide decision-making. You'll learn best practices for everything from labeling modules and documentation to data blending tips and how to train new users effectively.



Video Transcription


Tom Sturgeon
Welcome everyone. There's two keywords I want to jump into today, and every time I hit the button that happens, which is always a problem to go to the back of the presentation. One second here for a moment while I hit the reset button. There we go. Okay.

There's two keywords I want to focus on today. The first one is up here on the screen. It's team. It's this team concept. I'll step out of the way here for you guys. When I think about team, if I did a word association to team I would think about maybe sports. Sports gives us a great example of how to build good teams. If you think about sports, if you're not a sports fan, there's always a winner and a loser. It's very defined outcomes. There's no ambiguity with a few exceptions, there's no ambiguity. At the end of the season there's always a winner and a bunch of losers, essentially.

What's interesting about football compared to other sports is you'll notice this is an interesting example where we have equal distribution of teams, 32 teams, 16 in each conference, 4 divisions. What's different about football is that you cannot spend a bazillion dollars to go get all the players you want, you are limited by how much money you can spend. Every team in the National Football League has to spend the same amount of money to get the same resources. Salary cap.

The idea behind this is that you have the same tools given to every team. The same tools are being given to every team, but are the results the same? Now, you would expect one year, one team's going to do well. Two years in a row, do well. I was thinking about this recently, and I thought to myself is it equal? Are things equal? Over the last 10 years I went to Google trying to answer this question, and what came up was... I apologize if you're from St. Louis. This is double salt in the wound. Not only is your team over the last 10 years the worst, but they also left your city so I apologize profusely for saying that to you. These guys had the same resources, the same tools, over the last 10 years they've lost 113 games out of 160. Not a great percentage.

Now, these guys, which sometimes if you're... I live in Denver. If I show this picture sometimes people start booing. If you want to get your boos out, fine. A lot of people don't like these two guys especially, but they have opposite results. Last 10 years, they've won 80% of their games. Of course, when we want to find the answer where do we go. We go to Google of course, and the word that I was looking for, and this is the second word, the first word is team. The second word is system. These guys, whatever they're doing have a system of taking the same resources, the same tools, and producing better results.

Congratulations if you've just bought Alteryx, you're going to be an icon in your organization, but how do you build a team of icons? How do you build organizational success? More importantly, how do you avoid being the St. Louis Rams of Alteryx? How do you avoid being in Loserville? That's what we want to avoid today. When you're building a team, we want to make sure we're building it for success.

Hello, I am Tom Sturgeon. I've been using Alteryx for about 12 years. I work for a company called Schneider Electric, we are probably one of the biggest companies you've probably never heard of. We're the global leader in energy management. We have 180,000 employees in 100 different countries. We produce tons of data. We make stuff that we actually sell, but as a derivative of that we make lots of data, but more importantly to kind of bring this closer to the team concept, we currently have, in my domain, probably 50+ users. Hard to gauge the exact number of users we have. We have a lot of servers out there, and we have people in all different locations. Some dealing with different languages, which I was glad to see we now have Alteryx in French and German. We also have people in different countries, different continents, different time zones. We have lots of production modules that we built over the last 10 years. A huge library of intellectual property that we want to preserve.

The reason why I call out this last thing out here, this every night, some workflows running for seven years is because people naturally progress through your organization. If they're doing great work, you want to capture that and put it into a system that you can build upon. You do not want to lose, it's the worst feeling when you're an analyst or a manager and your analyst leaves and all that intellectual property has left. How do you preserve intellectual property that lives in Alteryx? Those are the things I'm thinking about as I lead our team.

That's our approach. I promise this is going to be as rapid fire as I can make it. We're going to go into some tips. I know when I come to these conferences I really want to walk away with some concrete things I can do. Nothing conceptual. Nothing theoretical. List of some concrete actions. I'm going to talk about tips in two spaces. One is environmental, and the second is design. I'll dive into design in a little bit. First I'll jump into the environmental side of it.

When I say environmental, what do I mean by environmental? I'm talking about decisions that you make as a team before you jump into the tool. These aren't things that are necessarily user tips, design tips, put your unique tool here, these are things around the tool. I'll talk through each one of these things. Some of these are very simple. This is not rocket science, little tips and tricks, workflow names, train the new guy, SQL, we'll walk through some of these concepts as we go through this. A little zoom in there. Decisions on how to make things before I start designing.

The first thing I want to talk through is workflow names. Something such mundane can cause you some pain. I was once asked, "Who is the worst employee you've ever worked with?" Who is the worst employee you've ever worked with, and maybe some of you know this answer. I don't want any names, and the answer, I thought about it for a minute, and the answer was me from six months ago. Me from six months ago had a project to do. Me from six months ago had to get something done on time, on deadline, and delivered. The me from six months ago wasn't thinking about the me from today who had a, "Oh, you want to do that again? You want to repeat that process?"

Here's a simple example of a project folder. We have project folders where all of our stuff lives, and if it's six months later you walk back into this and you're thinking to yourself, "Okay, I've got some,"... First off, you're using Alteryx, which is great. If you're using Excel and Access, then you're even in a worse position. What are these doing? What's a sequence? I have a four or five step process. How do I label these things so I come back in six months and know how to do it again.

Tip number one is, this is probably the simplest tip you could ever do, but we label all of our modules in the sequence that we run them in, all of the outputs, all of the inputs, all of the modules. 0-10 is the first step. Why do I start with 0-10? Because sometimes you go back and you think, "Oh, you know what? There's a step before that." So it's 0-7. We just arbitrarily take 0-10. Now everything is sequenced properly. You know it's 0-10, outputs, inputs relate to that. You're already halfway through, and you're not hating yourself six months later. That's tip number one.

Tip number two, super fast tip, if you didn't know this YXTBs are very compressed and a very great way to store intermediary data. You have CSV files and Excel files. I would avoid Excel at all costs to store information in between steps. YXTB is fantastic because it maintains all the field dimensions, field size, naming, and it's super fast. It's actually less disc space. My accident I came across this, and I realized that a CSV that was 100 megs would be the same YXTB at 30. I don't know how it works, but it compresses very highly, and it goes very fast. Tip number two.

This is another simple one, folder organization. As a team, again, we have 55 people. One of the rules we have in our team, and I'm actually not a huge proponent of desktop in the enterprise. I know that's probably contrary to the prevailing wisdom here, but we like to have people log into a controlled environment and use Alteryx for data security, speed, consistency of experience. IT gets involved, and you start dealing with people's laptops and latency issues and things like that. What we do is we have a centralized folder structure that keeps everything organized by date, by month, by the analyst. You'll see all the initials. Again, we have 55 people. We want to know who did what, but also I want to make sure that if that person leaves, I know where all their work is. Also, if I need to access it while they're sleeping I can go take a look at that. Very simple structures around maintaining modules.

This may change in a year with the new features coming out in the gallery and so forth, but for now this is how we operate as a team. The key takeaway here is... You'll see team takeaways, I'm trying to point out what's the team takeaway with all of this, and it really is nothing is saved locally. The rule on my team, and I kind of scare my new employees, I should be able to take a hammer to your laptop and crush it if I wanted to, although I won't, and give you a new laptop tomorrow, and you should be working as if nothing ever happened. Ignore what I just did to your laptop. This keeps folks in IT very, very happy with data being in the data center.

Go ahead, question.

Crowd question:
[inaudible 00:10:52]

Tom Sturgeon
With which one? ODBC connection? The question is how do ODBC connections, and what's the rest of the question?

Crowd question:
As it relates to what you just said, you'd take a hammer to it, and you're up and running the next day [inaudible 00:11:11].

Tom Sturgeon
We don't use Alteryx on the desktop. All the ODBC connections are centrally located in the server. We have people all around the world, but we work in what I call St. Louis, which is where our data center happens to be. Everyone logs into the servers, and you can also do the same with virtual desktops too where you actually deploy Alteryx at a virtual desktop environment. It makes your life... If you don't have a lot of IT support, it makes your life a little bit easier to manage that. The ODBC connections don't live locally on your desktop. Does that answer question?

Crowd question:
[inaudible 00:11:46].

Tom Sturgeon
Okay, sorry. They're on their server so that... I'm not crushing a server. If I crush a server, I'd be in real trouble. That would be a problem. Yes please.

Crowd question:
How do you do version control?

Tom Sturgeon
Version control. Good question. With adhoc project folders, we don't. Usually how a project works through the process for us is you have a project, I've got to get it done, it might be a one and done type of situation where you're not really concerned about version control. When that project becomes something you want to repeat, then you can leverage the Alteryx gallery to deploy that into the scheduled module. We have lots of modules that are scheduled, and that's where we would handle that. We don't think about version control too early on in the process since we're trying to get the... Does that answer your question?

Crowd question:
Yup.

Tom Sturgeon
Again, simple. Again, not rocket science. We like to keep things organized, AYX Modules, AYX Outputs, and the frame of reference here is 55 people trying to work in the same sequence. That's my reference here. If you're one person, do what you want. It's when you have a second person is the concern.

This is the next question. This is number four. This is the closest thing to Republican-Democrat, Pepsi-Coke, that Alteryx has created, which is vertical vs horizontal. Many new users use horizontal. We were before that using vertical, and that's just how I learned the tool and how we learned the tool. I'm going to make one pitch for vertical. I'm going to make one pitch for vertical, I don't want to offend anybody so please don't throw anything at me, but our pitch for vertical, and I was talking about layout. Just to be clear, the layout of the modules, just to make sure we're talking about that.

My pitch for vertical is simply this, when we have new employees come on board, the most common tool we use is the join tool. When someone's using SQL in the past they always reference Seq. Outer-left and Outer-right. I love Alteryx, and I love the horizontal, but I have a tough time with the translating Outer-left to Outer-right. I've got to do this. That's one of the reasons why we still do vertical, and also for us it's more of a web user experience where you'll see some of our modules later on, they're very long, and you want to scroll down sometimes. It's a little bit easier for us to see that, but whatever you do I'm not pressuring you. No peer pressure, but just adopt a standard. I can tell you if some people are doing horizontal, and some folks are doing vertical, that's very confusing. They don't always transfer that well. There's always some things that get out of whack, especially with documentation and things like that. Pick a team standard.

Number five, I'm not sure if you can hear the audio. This is what we do with our new employees. Hey, welcome to the team. Here's Alteryx videos. Go watch a video, and they come back, and they're excited. "Wow, I wish I had this tool at my last job." Alteryx does a great job of providing all the materials to help you learn, but then, we give them this. This is intimidating sometimes for new users, it's intimidating. You have to go do a project. Go deliver this project. Here's your first project. We want to make sure we ramp that person up, but we also... Everyone's busy. All of your senior analysts are busy, so they can't sit there for eight hours next to the person and say, "Okay, do this. Do this. Do this. Do this."

One simple thing we do for newbies is we do breadcrumbs. Simple little tactic where you're a new person, what we'll do is we'll simply have an experienced person throw the tools in the palette, make some notes, don't configure anything because you want the person to learn how to connect the tools, and sometimes that's slow in the beginning. We just do a simple breadcrumb trail of how your module should look like and have them connect the dots, configure the tools, make the mistakes they're going to make, but it at least pushes them farther along in the process, and it's a little bit less intimidating for new employees coming on board.

Number six, which way is the right way? Two modules do the exact same thing, this may be very obvious, but sometimes I have to explain it to my team. Yes, you've achieved the goal. Thank you, but I can't read that. That's so much code. So much stuff. If we can do it this way, then let's do it this way. Less tools is better. That's just the rule that we have. The less tools you use to achieve the same role, the better. I count to that as an obvious thing at times, but then I realize that wasn't always the most obvious thing to say, so we have a rule in our team, our standard in our team that says whoever does it with the least amount of tools is the most efficient way of doing it, but as in Alteryx, there's multiple ways of doing the same thing, as you all know. Just having that put together in writing.

Last one for the environmental stuff, and then we'll jump into the design. Again, thinking about expanding Alteryx in your environment, I categorize people in two groups, data analysts and business analysts. Data analysts generally spend 80% of their time in Alteryx. Business analysts who are really communicating with your senior leadership, your leadership team, helping translate data into insights, and you want them in Alteryx, but they're not necessarily going to be living in Alteryx 80% of the time. How do you translate this? If I'm a business analyst and, "Here's Alteryx. Just go learn Alteryx and go get your data. This is your schema," for example. This is very intimidating for a part-time Alteryx user. Expecting people to go into SQL server, ODBC, terabyte, whatever it is, it can be very intimidating.

One of the things that we do from an organizational perspective is we process data left to right. We take the data, load it into our data mart, put into our SQL server. We enhance that data. All of our data analysts are working in the back end to build this nice, clean, hygiene data. That's our analytic sandbox, but at the very end of the process, we're actually making... I call them Alteryx indexes, but the technical term is they're called Calgary indexes, but I just use Alteryx indexes. Can I have a simple show of hands of who uses Calgary indexes in the room today? Okay, not a lot. This is a great way to get people who are a little bit nervous about using Alteryx into the tool.

Make them... If I get the word right, it's a de normalized dataset, super table we call them. A non-technical term, super table. We connect all these things together and make a nice simple excel file-esk dataset, and all they need to do... This is really cookie cutter stuff. Give me the data from Kansas City, Kentucky, whatever, leverage these things as much as you can. The only time I see these things not being very useful is if realtime data is needed, this isn't your solution, but most scenarios don't require that. This opens up Alteryx to more people in your organization. The more you have people using data and Alteryx, the better it is for the entire company.

That is the environmental stuff. Those are things, again around the tool, around the tool, and now we're going to talk about inside the tool. When you're inside the tool, what are some things you can do, simple things you can do to make it more collaborative, standard. The one thing I want to point out is these are user tips, but they're also an agreed upon way that we use the tool as an organization, as a team. Some of these are user tips, but I'm going to try to highlight the team effect of what that means because I know you have other user tips out there for sure. This goes back to the system concept. I apologize, doing presentations on Wednesday is always more difficult than doing it on Tuesday because of the event last night. I didn't pace myself vocally this week.

Workflow layout. Who's the famous painter that does all the crazy painting?

Crowd question:
Picasso.

Tom Sturgeon
Thank you, yes. You could make a module that looks like this. Again, simple concepts. The flow and the shape should mean something. Without me looking at anything, I can see I have data inputs, everything kind of comes together. We have some join, some data enhancements, other join, I can very quickly see the flow of what's going on without actually touching anything. As a manager, that's the first thing I see. Zoom out, show me the whole thing. What's going on, give me a picture, and then dive in. The shape must mean something. It's a bit... It's a tedious thing, but it's also you're losing some functionality in Alteryx if you don't leverage the visual aspects to the fullest extent. That's one key visual aspect, is the shaping of your modules should have some type of consistency to them.

Number two, tell me why, not what. I love it when someone puts a join to this. I join this to this. No sh... I can see that visually what's going on. What is the reasoning behind this? We have some notes. I want dialogue. I want normal spoken English. Don't give me the whole technical stuff. What was the reason? Who did you talk to? When did that happen? Why did we do this? We have some pretty standard processes on when we do peer reviews I'm looking for this type of information on our modules. Super helpful, especially when that person gets promoted because they're using Alteryx, and they go somewhere else, and you're trying to manage their process.

This one is super basic. Colors, I was once on a train, and I had a fire drill, and I actually talked somebody through a module by describing, "Go to the fourth box down next to the purple letters, go two tools over, change that tool." Super helpful to give some visual references. We have greens equals inputs, grays are logic, yellows are things to watch out for. We all agree that we're going to highlight things to be careful about for yellows. You can also do the test tool for yellows. We don't use the test tool probably as much as we should, but the test tool is a very good cautionary things to put in your modules.

Last piece, everything has an operator's manual. I've been asked, "Where do you put your documentation? Should I write a PDF? Should I put it on your website?" No, put it in the module. Put the instruction manuals in the module and document it there.

Flags and filters, flags and filters. In the very beginning we all... Not we all, I shouldn't say that, but many users think of... They're coming from Excel reference points, and they're thinking about worksheets. What I've seen happen over the years is they'll take a data source, and they'll try to make what they know in Excel. I'll filter for this data source, filter for that, filter for this. Again, this is just generic stuff here, but do some work on the tools. The first thing about this is that one, I see four filters. Is there five fields that we're filtering out? Are there five values that we're filtering out? Are there six values? What are the other values that we're not missing? Is there data in here that's also in here? It opens up some questions. This is a very nuanced part of the tool.

What we standardize on is this. Right away, I can kind of see I don't want to make multiple versions of the same table or I'm not sure. We take the same piece of code and do filters and filter off "F". Then, as a quality assurance person I can see is that something we should be looking at? Is that a problem? Don't know, but at least I could put some words around that. Even better, a more efficient way... This is usually how early onset... Folks, think about the world as worksheets, and I've got to make these tables. Then I've got to do these thing, and then I've got to bring them back together. The more you separate data, the more likely you are to mess up and to make multiple tables, and you have to keep track of record counts and duplicates and so forth. The more you can keep things in line, the better.

If you notice your filter code, your filter tool, up until recently I should say, and your formula tool use the same library. You can copy the same code. You take your filter code, and you make fields on your data. We call those flags. We're flagging the data, keeping it in stream versus filtering it, doing something, and then trying to bring it back together. This took me awhile to get in the beginning. It was so difficult to wrap my brain around adjusting it from the Excel worksheet environment to this. Once your team gets more mature, this is how you operate. It's again, less tools. Less tools also falls into that.

No nudity. No naked joins. What do I mean? I saw this on the community. Perhaps you've had this experience. I only found this once, I'm sure there's probably a lot of these situations. This is the beginner issue in Alteryx and so forth. You can see the comments here, "I joined this and my data," we call it an exploding join. I'm sure there's other terms out there for it. It's a big headache. I was with somebody in the solution center yesterday, and this is what they had. This is not their module, but this is what they were doing. They conceptually want to join data together, the join tool. That's how it works. Don't trust your data. 99% of our joins have to have a unique tool on one side of the other, by default. Even if your data is unique, one day it could not be unique, and then you're in trouble.

If there is, our standard on our team, this is tip number six, is that if you don't have a unique tool, you need to explain why. Why are we trusting our data to do that. I've probably put about several thousand joins on a palette and 1% didn't have a unique tool on one side or the other.

Next one, this is, again, not a mistake in Alteryx. This is not a mistake in Alteryx. You're not going to get an error on this. You can continue to do this if you wish, but everyone knows the right fields when you put two fields together. The standard on our team is we... I don't want to see any rights. A right to me means did you address that situation? Is that a field that we want to keep going forward with? If it is a field that we want to keep, then we want to rename it to something meaningful. Do something intentional on that field. Don't let it default to right because as a QA person I'm looking at this thinking now I need to ask you a question what's going on here. If I see you rename it, then I know that... I'm not sure what you're doing, but you're intentionally doing something, and that you meant to do it, and you're keeping it going forward, and you deselected the other stuff. Field management in Alteryx is critical. Managing fields properly is so critical to keeping things organized and clean.

Number eight, and we have about three more to go here I think. Number eight, always join from... This is the same filter. First off, there's not unique tool, so that would be a (buzz) right there, but they're joining from the right and the left. Again, not a mistake in Alteryx. You can do this, but it becomes very... When you're building data, it becomes more challenging to, "Okay, these fields I started with, and then I added these fields over here, and then I added these fields over here, but these fields got added over here, and then I deleted these." Your field management starts to get a little bit cludgy.

The rule of the team is always join from the right. I actually have this symbol. I say, "You've got to go like this, and you got to go like this," any my team now makes fun of me when I walk in the room, and did you do this. They all go, "Yeah, we did this." Always join from the right. It keeps your fields being added on conceptually and of course unique.

Number nine, this one's going to take a minute to explain here, so bear with me. When you're working in a team environment, we've allocated our analysts based on the source data. We have so many sources of data, you name it we have it. ERPs, 40 different ERP systems, SAP, Oracle, salesforce.com, just everything. No one person can know all that stuff. It's just impossible. What we've done is we want to build data that brings together... We want to make it look like we're one company when we're not really one company. We had to bring all this data together and harmonize it. What we do is we want to... You can imagine a big union at some point coming together. What we'll do is we need all that data to look alike in the union tool.

We'll create a common field architecture. When I ask you to rename fields, you're going to think of the select tool. The select tool, we have one process where we have 505 global reporting partners who every month give us their sales of our products in all different languages. We have to talk all those 505 and make it look like one system. What we've done is if I gave everyone a select tool and said, "Hey, rename it to this common format," we're going to have some mistakes along the way. We're going to have people put the B string at 250, and maybe this guy's going to change the label one way. It's just a little bit more difficult to manage. Instead of asking people to do renamings, we ask them to do mappings. Here's an example of salesforce.com. You can see we want to rename account owner to company name and there's probably an address field in here somewhere to address 1, address 2 and so forth.

What we do is we develop a simple formula tool. As opposed to you going into your situation, selecting and renaming data, I'm going to give you the way I need the data in a formula tool. You simply map your data. It's all pre configured. Here's my field names, here's my string types of whatever the field type you want to have, size, it's all going... I have 20 people, here's a formula tool. Take your data and map it to this. When I say map it, all they're doing in the statements, as you can see here hopefully, is they're simply saying this field equals that. Achieves the same purpose of a common field layout, but in a much more consistent process across multiple teams, and development in pieces. I can see some faces looking... Hopefully that helps.

X Files, the last one. We did promise to be at 10:30, right? We're getting there. Actually, this is my lack of skill in any sort of artistic tools. There you go. It took me a half hour to figure that out. X Fields. X Fields, this is again, a team standard that we have. Keep the truth in your data. What does that mean? A lot of times you're going to be asked to format, clean, hygiene, standardize, do whatever it is you're going to do to data. Don't do it on the original data set, on the original field. Our recommendation is not to do that I should say.

Our recommendation is to build a new field, and we always put the word X in front of it. That tells me that that's a field that we've created, that our team has generated after acquiring the data versus that exists in the source system. This allows us and everyone else to compare before and after. If we're making hygiene, especially an address and company name cleaning and things of that nature. X Fields are always what we call them. We can do it here in the X account type on the regular formula tool. You can put that in there too, but everything is labeled X something. It just implies a simple standard of what we did to that field.

Thank you.

There is a summary here. If you're interested in the recap of that or photos. There you go.

Speaker:
Great presentation, thanks Tom.

Tom Sturgeon
Thank you.

Speaker:
Any questions for Tom?

Crowd question:
Do you have any standards to impose data integrity?

Tom Sturgeon
Standards to impose... On source systems?

Crowd question:
No, on the... IO controls, Roe counts, whatever, just to make sure that what you're bringing in you're controlling all the way through the process as a team.

Tom Sturgeon
Nothing codified. Nothing that says I need some tools and reporting. We will do... There's two answers to that question, one it's part of the training process. Records in should equal records out unless they shouldn't. There's a record count process that we have to go through training, and the browse data tool helps us with that. The second answer is for certain processes that are high redundancy, high importance, we build a module that does the work, and then we build attached to that module on the left, all the reporting. Did all these fields join, for example. What's the record count? It summarizes it in to the reporting tools inside Alteryx that will send an email to the analyst responsible. I wouldn't say for everything, but that really, really important stuff we'll go that extra mile to design reporting solutions about the data and on top of the process. Did that answer your question?

Crowd question:
Yeah.

Tom Sturgeon
Okay.

Crowd question:
For your flagging fields. Do you, at the end of your workflow, eventually just filter those all out, or do you leave those in and then in your dashboarding phase use the Tablo filters to take those out? Or do you just leave them in there and just note that they are flagged?

Tom Sturgeon
Usually for flagging fields, we find it's easier in Tablo to make information about the information. We'll usually leave them in Tablo. We'll usually leave them named X something so that the analyst in Tablo can then use it. We will then usually rename it in Tablo to something a little bit more user friendly. Nobody likes to look at X something. It's kind of ugly to look at. We'll keep it all the way consistent until the end, and then rename it. That's if it's going into Tablo. Most times the flags will stay in the output. The answer I would say is 90% of the time, yes. Yes.

Crowd question:
[inaudible 00:36:21]

Tom Sturgeon
It's really up to the analyst. I would say that I'm always hesitant to take out data before it goes to visualization because then you don't know what you don't see. I would rather them actively take it out in Tablo with the field filter in Tablo than me in Alteryx remove the data and then to never see it in Tablo. I probably prefer to do it that way, unless there was a technical issue, size of data, performance issue where Tablo wasn't going to work because you just had a mountain of data in there. Under those circumstances I would adjust the delivery.

Speaker:
Sure.

Crowd question:
Hi, how do you feel about replacing right, underscore with the name of the file that it came from or some other prefix to everything?

Tom Sturgeon
Sometimes we'll do that if we have a clean version of a field we'll call it C something. The challenge with that is if you have it reliant on a file name and the file name changes a lot, then your field names are dynamically changing on you, which can be a bit of a pain. In that situation what we'll do is on the input tool, bring in the field name as a field. That's an option to do that.

Crowd question:
Thanks.

Speaker:
That's good stuff.

Tom Sturgeon
Thank you. We can throw it like a football.

Crowd question:
Are there any standard exception process that you guys have devised in your course of development?

Tom Sturgeon
A standard exception process to which?

Crowd question:
To handle the workflow fail, or to notify. What are the best practices around that?

Tom Sturgeon
This is where Alteryx, I think the server tool is developing some more robustness around this. We've had to develop some kind of hacks around some of these things. What we do is... This is what we might not do in the future, but what we've done is on our 1200 workflows... There's nothing up there important anyways, unless you want to see that. On the 1200 workflows we have three tools. It's called module runtime. Those are pink boxes. On that, we have a text input tool that says the name of the module, a short description of what it is, and then the next tool below that it goes into a formula tool that says daytime now. The last tool is simply an output to a SQL table. It captures just three fields, the names of the module, what it does in normal English, a couple sentences of what it's doing, and then when it last ran.

We were capturing... We built that into every module so it captures the runtime. There are other solutions out there that you can look at the logs, and you could probably think of a few other solutions. That's just what we settled on. In the production modules, we call it our pink box. Our module runtime box, and it captures it as a SQL table, and then we use Tablo to visualize when things are running, and if things didn't run we want to know that for sure. Does that answer your question?

Crowd question:
It does.

Tom Sturgeon
Okay.

Crowd question:
So, to start off you templatized workflow for people start with.

Tom Sturgeon
Yes, and we build it into the process off to the side.

Crowd question:
All right, thank you.

[inaudible 00:39:52] For a team that's just starting to get into Alteryx pretty heavy, how long did it take you guys to [inaudible 00:40:00].

Tom Sturgeon
We've learned this... There wasn't a book I read. We've kind of stumbled into these things through mistakes a lot of times. It really depends on the leader. I feel that we've had other teams in our organization who use Alteryx for the same amount of time, and they haven't had the same success because they haven't adopted some standards. You need to have a champion who lives and breaths it, and then animate that to the organization. It's really a management culture. The last thing I'll say is we have very... I have employees all over the world, so we have to have very specific guidelines.

People want to know what do I need to do to be successful, I'd document these things in a document saying... If I gave it to my wife, she'd fall asleep like what does this even mean, it's all Greek, but you know, LD modules, IT modules, join from the right. We actually write this stuff down so people know that that's what we're looking for. It took me about seven years to get to that point, and I realized I wasn't doing a very good job at communicating it as a manager, so we wrote it all down and shared it.

We also have very open conversations about make your mistakes, encourage mistake making, let's bring it all together. I always honor the people who are going to step up, show their modules, and get exposed for all their mistakes. We want to make sure mistakes are awesome and enhanced and okay and really create that culture of people feeling okay that the boss isn't going to get mad at you. We want to make sure that we create that culture of failure, do it quickly. That's what we've done.

Crowd question:
Thanks. This was really useful and practical. I'm thinking of a variety of workflows. We've got something in them that is hard coded so to speak. This year it says 2017, and next year when roll the year, I need to update all those years. I think probably the best practice is to try to get rid of those hard codes wherever we can and engineer around them, but do you have any tips on using, some method of all this stuff needs to get updated at a certain time?

Tom Sturgeon
Hard coding, I might be able to answer your question. If you have lots... If you have a huge if-then statement, is that what you're talking about? If it's this 2017, then do this?

Crowd question:
Some type, yeah.

Tom Sturgeon
What I would do, what we normally do is we build reference tables. I find it easier to manage things via a Google sheet or an Excel file and build that into your process as a join, and then people can edit the Google sheet or whatever it is you're trying to do, and your analyst doesn't really need to be responsible for all that logic. We try to take it outside Alteryx, but then feed it in. When you need to make those edits, like this account is no longer owned by this person, they go into the Excel file or the CSV file or whatever sharepoint site and make the change there. That's your join versus hard coding, if it's Bank of America it's owned by Frank, and the next share you've got to change it to Bob. Hopefully it helps you.

Crowd question:
Yeah.

Tom Sturgeon
Okay.

Crowd question:
Thanks.

Speaker:
All right, hey thanks again everyone. Thanks for joining us. I'm sure Tom will standby for a little longer to answer questions.

Tom Sturgeon
Thank you.

Speaker:
Thank you.

^Top

TESTEZ LA
PUISSANCE.
D’ALTERYX.

Pour Démarrer