What Do You Need for Successful Self-Service BI?

Click here to watch this episode on our YouTube channel.

In this episode, Brick and Caleb discuss self-service BI, what it is, and some of the considerations for making it happen in your company.

Blue Margin helps private equity owned and mid-market companies organize their data into dashboards to execute on strategy and create a culture of accountability. We call it The Dashboard Effect, the title of our book and podcast. 

 Visit Blue Margin's library of additional BI resources here.

For a free, downloadable copy of our book, The Dashboard Effect, click here, or buy a hardcopy or Kindle version on Amazon.

#BI #businessintelligence #selfservice #PowerBI #Azure #Synapse

Transcript

0:04

Welcome to the Dashboard Effect Podcast. I'm Brick Thompson. And I'm Caleb Ochs. Hey, Caleb, are you doing? Pretty good. So it's been been a few weeks since we sat down here together. Yeah, it's been a while. I got a lot to say. All right, good. What's our topic for today? So let's talk about self service BI. All right, so what is self service BI? That is a good question. So if we had to boil it down, it's putting data and business intelligence in the hands of people that don't necessarily have BI technical skills, and letting them explore the data and get answers from it themselves. Yeah. Okay. That's how I see it, too. So you may have people with technical skills, but really it's, you have probably heard this term, Democratization of Data. Making data available, just to regular business folks who maybe aren't highly technically skilled, but giving them access so they can do their own analysis and be able to make decisions, that type of thing. Sure. Okay. And we're hearing more and more companies really want to move towards this. And I think it really is because if you can allow the data some freedom within the company, you do speed decision making, and you get ideas that you wouldn't have gotten otherwise. And I think the key differentiating factor between regular BI and self service BI is really, you don't have to go to IT every time you want some data, right? Yeah, when I worked at my previous job, one of my first jobs out of college, I was an analyst, and I had to wait on IT for data. And even self service back then was becoming something that people wanted you to do, and it was this hot topic and stuff. And that was right about the time where these visualization tools were just kind of gaining popularity. And... what we're seeing is that it's becoming more and more accessible for companies to actually do this. Even though it really hasn't taken off, at least in my view of things, in the last few years. Even though people want it to. Yeah, yeah. Well, it's not as simple as you might think, at Right, Yeah. Well, there's a lot there. Why don't we start with first. It's very, I mean, there's a lot to it. There's definitely a lot to it. And like you said, when I was waiting on IT, like, I had the technical skill. We've progressed a little bit since then, I think. I hope. But, you know, I still had to wait two weeks before I got a little data change in my dataset. And I was still building the reports, but there's both pieces to it, right? There's building the reports, but there's also gaining access and knowing what data you have access to and usually having some sort of control over it. just talking about self service BI in the Power BI world? So, at Blue Margin, we've devoted ourselves to the Microsoft Stack, Azure, Synapse, Power BI, those types of things around BI. So, what are the things that Power BI does to enable self service BI? So there's some things that are very close to everything else that other BI tools do for you. Like dragging and drop. That's one of the big selling points of a tool, like Power BI is that you can just drag things, it's really easy to create visuals and stuff. And so Power BI does that, which is great. Yeah, yeah. But as you said, other tools, do it. Tableau does that, Click does it. In fact, most modern tools do that. I do think Power BI does a really good job of making that simple, displaying the data model, what measures are available and so on. But you're right, a lot do that. There is a thing in the Power BI world, which has been there for a long time, and doesn't get a lot of traction typically, although I think it's possible it's going to start to get more and more. And that's Natural Language Querying. So the user can go in and actually just type in and do this right now, "Give me sales by month for the last 12 months," something like that. And assuming your system is set up well, so you've got good metadata, so the system knows what measures to go to, and what data to work on. You can get interesting results there. It'll even do simple visualizations, and so on. Yeah. I mean, it's a really cool feature that they've rolled out. And it's been out for a couple years. But you're right, you have to get everything set up the way that, that tool is expecting it to be set up. Otherwise, it might get confused. There are some times where you'll see in data models where you have two have the same column. They're named the same thing, but they're in different tables. So then you have to name the table in your question. You have to set things up to where it works well with that. But a really powerful option for just getting people used to using the data and, you know, just going and asking a question. It's that easy. Yeah. And I think as people are more and more used to using their personal digital assistants, you know, Siri and, and Google and that type of thing. And we're starting to see these generative AI, text, chat engines and so on, I think people will start coming to this more and more. So this is probably going to be something where I'm sure Microsoft is spending a lot of effort on that. And I think implementers such as Blue Margin, such as ourselves, we'll need to really start to focus on that, because I bet users are gonna start expecting it more. Yeah, that's really interesting. That's kind of broader than just self service, too. If you think about just society, people are just wanting to do things themselves more and more. You're just going to Google. You just want DoorDash to come to your door, you don't want to go out and deal with it. It's all about convenience, and easy and quicker and getting the answers you need as fast as you can. And that's exactly what self service is meant to do. And something like a natural language query is is kind of the perfect entry point. Yeah, perfect on ramp. Because you don't have to be technically skilled, as long as it's set up well, and giving you the answers you're looking for. Another thing about the Microsoft ecosystem that I think is useful for self service is that you can have certified datasets. So IT can actually publish datasets where the data has been certified as clean and correct. And the measures that are available in the cubes, and so on, that are acting on that data also are "certified." And so users can know if they're using a certified data set that they're likely to get good data. They don't have to worry about cleaning the data, for example. Yeah, right. I mean, that's a huge piece. And we're kind of touching on that governance side of self service, where you want to make sure people are looking at the right numbers, even though they're going and getting them on their own. Yeah. And, you know, something like a certified data set will hopefully get rid of that problem where an analyst is pulling a bunch of data out and then doing all kinds of transformations and filtering and stuff to get the number that they want. Hopefully, that's all done on the backside, before it gets to the report. And then you know how to do that. And you have good reports. Yeah, and it's not just that the analyst has to do all sorts of transforms and cleaning and so on. But we certainly see that an analyst will do some work on data, and then present that and have someone else in another part of the company say that's not right. And now there's a discussion or argument about how they're calculating certain things and where they got the data. And it just sort of adds noise to the system, as opposed to, if everybody's working off of certified datasets, you have a lot less of that. Yeah, that's another really good aspect of this around governance is choosing what a measure means and aligning on it as a company. We just see it over and over where people are like, "Oh, we have two different definitions of the same measure." If they're lucky, it's only two. Net profit, you can calculate a bunch of different ways depending on what you're including in the in the calculation. Exactly. Yeah. And then also, I think, it's great to have the certified data sets, but also then the ability to put your own data in there as well and even join it to certified datasets. So you can enable your users to actually bring their own sort of custom role data. That will have some of those problems that we were talking about with the governance and so on, but sometimes that's really necessary to do the analysis that, you know, citizen BI developers want to do. Right, it's kind of a journey to get to self service, because, for that example, you're gonna still have budgets in Excel that you don't upload to some system yet. Maybe ideally, in the future, you could do that. But right now, you don't want to just say, "We can't do that." So you want to enable that ability still, and what you're talking about is exactly how you would do that. That's a great example. So maybe there is an official budget, but maybe your department head has asked you to do some analysis against a proposed budget that they want to work on. And so it's not uploaded anywhere. So yeah, thats a really great example. So, we've been talking about, you've mentioned governance a couple times, I guess we could get into what are some of the adoption challenges and the considerations for going into self service. So when you think of governance, what does that mean to you at sort of a high level? I think, we've kind of touched on it already, but it's knowing what people are using in terms of data and how they're using it. And feeling confident that it's correct. Right. Okay. Yeah. Would you include measures? I guess he would, you would include that as well. So it's not just the underlying data, but how are they doing calculations. Yeah, right. Yeah, because you can filter data out and stuff in your calculations, that's for sure. Yeah. Okay. So another big consideration is obviously training. So you can't just set up a system and say, "Hey, good news, everybody, you can now do self service BI." You're gonna get a lot of blank stares. So you've got to have some kind of an adoption plan to actually train analysts. And we were talking before we started recording about sort of, where do you start in a company, and really, you almost have to start with enterprise wide BI. And by enterprise, I don't mean huge company, but just the company wide BI. So that you've got, you a foundation, a basis to build a self service off of. And from there, you can start moving to more and more granular user groups. So you can go to, you know, departmental level BI, or team BI, and eventually get to individual BI. You don't have to go through all those, you could jump to individual BI. But you really need that initial base of data warehouse, data lake, certified data, or at least clean data, and accept it as,"This is what we're going to use," before you can really start training the users on how to get the most out of it. Right. And governance and training are closely related. You want people to use the tool in a way that's going to work really well for their data and how you've set it up and actually get the most out of it to hopefully, increase the likelihood of adoption. It's interesting, when I worked at this other large company where we rolled out BI, the way that we did it was just straight to departmental type BI. So I was an analyst in a department and I got trained on how to use Click View. And then I had to go figure out how to make things happen. That was basically it. I just got trained and then I went and did it. And you could see how in some departments that worked really well. And in others, that just fizzled out. And I think the reason why is they didn't start with that, kind of core foundation to get the buy in from the company first. And then enable the users like the analysts, like myself, to be able to build specific reports and self service stuff for the departments. Yeah, that makes sense. So you've got to have that foundation, then you've got training, then there's a concept of a center of excellence. And we think this is something that's going to be really important in order for self service adoption to happen. Can you describe how you see a center of excellence? Yeah, the best way I can think to describe it is, it's like a chat room, that you would be able to go and ask a question. And there would be people that moderate the room and would give you an answer, and help you out. Help you make progress and into what you're doing. I think it's a little bit more than that, but I think that's the bulk of really what it what it's supposed to do. Is kind of be that support that support center for your BI initiative. Yeah, so giving self service users somewhere to go to ask questions. So usually, it would be populated by a group of more expert users who really know the data, know the business environment, know the business rules, that type of thing. You know, in our case, we can provide that center of excellence as the company is getting up and going. But you have this concept of a chat room, or office hours, or just someone you could go to that's going to help advise you on best Right. Because no matter how good your training is, there's practices, and how to use the data, and how to connect and get to the training, and solve a specific problem. All that stuff. going to be some real world scenario that your training didn't cover. It may be very, very simple. But if you don't have the answer, and the quick answer to it, can really hurt you. Yeah. And then another really important thing is just to have good support. You know, the whole point of self service BI is not have to go to IT every time you need something. But guess what, there are going to be times you need to go to IT. So you might as you're doing your analysis, figure out,"Okay, I do need another set of data that's not available," and need to go and talk to someone about making that available. Yeah, so you have to have a system for that, right, of That's right. pulling in new data sources. And even further than that, this And getting that backup in there is going to have to involve some kind of fits into support, but not really, but I'm gonna say it anyway. Where, let's say you've got a self serve made report and it's starting to gain traction in the in the company, and it becomes a report that everybody wants to see. You also have to have a way to move that report from the self service build, into like a production ready, foundational report. IT support. Yeah, that's a really good point. And it's something we might have mentioned to start with. A lot of times a self service BI does result in a more widely distributed report. So someone's doing some analysis and they discover some really clever way or useful way to look at the data in order to drive decisions, and someone will realize, "Okay, we need to make this available more broadly." And that's where you need the support for, "How do you make that happen?" Right. I mean, it's a great thing. That's ultimately what you want, right? That'd be fantastic if you got that all over the place. So you definitely want to make that as easy as possible to enable it. Yep. Okay. So how do you get started? Let's say you want to do self service BI, at your company? Yeah. So I think it really asked to start at that Enterprise BI level with building a foundation. You're gonna have to have your IT, your technical folks, do that for you. Get you into a place where you've got some pillar reports, you're starting to see the adoption. And then at that point, you can kind of move your way. Hopefully, you'll be able to skip a few rungs, maybe you don't have to go to departmental, you can go straight to self service. But you have to start with that foundation. Once you have the foundation, then you can start adding things and optimizing your foundation for this self service model. Yeah. And obviously, you can start with beta users, maybe slightly more technical individuals, so that you can start to figure out, "Okay, where are the rough edges we need to sand off before we start to roll this out more broadly." But then you've got to think about the training, the governance, Center of Excellence. Those things all don't have to be perfectly in place, you can do it in an iterative manner, so that you're bringing on some users, you're learning some things, you're iterating those things and then ramping it up as you get it more broadly. I mean, hopefully, you get to a point where I mean, it's at some point, every department has people that are able to do this on their own self service analysis. Right. Yeah, that's the ultimate goal. And I guess that, that kind of touches on something that we didn't really speak to here, but really about the culture of the company. If people don't really want this, good luck, you know. You need to make sure that you have some sponsorship, and you've got a culture that is using data now and just hungry for it as is. Gonna make your life a lot easier. Yeah. Well, if you don't have that, you're probably not gonna spend the effort to make it happen. And I think in this time with how important data has become, and continues to become more and more important, you're probably going to find that in most companies. But you do need to have that. And I think having champions, people who are actually using it, and sort of showing how useful it is and able to sort of lead the way, is really important. Yeah, right. Yeah. Having a few quick wins, goes a long ways to getting people excited about it. Definitely does. All right. Anything else you wanted to cover? I don't think so. All right. It's been good sitting with you. We'll see you soon. Thanks Brick.