Scaling Power BI Adoption Within Your Organization | Governance 101

Whenever you have a business problem, it is very tempting to think, if we just buy this new tool, it's going to solve all our problems. Because, of course, spending money is much easier than thinking. The reality is, though, that buying a new tool is only the start. You need to train your staff on using it. You need to change your processes to incorporate it and generally have a plan to get value from that tool. And Power BI is no exception. So it's a fantastic tool for business intelligence, but just. Just buying licenses won't solve your data problems. So today we're going to learn how to bridge that gap from just buying stuff to having business impact. And in order to do that, I've got two fantastic guests for you. First of all, Ginger Grant is the principal consultant at Desert Isle Group. She's an expert in advanced analytics solutions, including machine learning, data warehousing, ETL reporting, cube development, Power bi, Excel, automation, all sorts of things. And Microsoft has also recognized her technical contributions by awarding her a Data Platform mvp. And on top of this, Ginger is a Data Camp instructor. So she's created the intermediate SQL Server course, and she's also acted as an advisor for much of Data Camp's Power BI curriculum. And secondly, we have Stephanie Bruno, who is a business intelligence architect at Eyelink Digital. Stephanie has over a decade of experience in all aspects of the business intelligence workflow, from modeling data to implementing enterprise CI CD processes to reporting management, and of course, lots of working with Power BI. In 2020, Stephanie was also given the Microsoft Data Platform MVP award, and she was previously a senior Business analytics specialist at Microsoft. I'm hoping she still got some inside tips there on what's going on with Power bi. And she was also Associate of Informatics at the Elizabeth Glaser Pediatrics AIDS Foundation. All right, so two esteemed guests, let's kick things off now. I'd like to start with a bit of inspiration. So first of all, can you tell me, like, what a good business intelligence setup looks like? Ginger, do you want to go first?

I'm sorry, could you repeat the question?

We're having a lot of audio issues today. Technology is hard. Okay, so the question was for a bit of inspiration, can you tell us what a good business intelligence setup looks like in an organization?

So what business intelligence looks like? Well, business intelligence is basically looking at providing answers with your data. Everybody's got a lot of information that they have gathered. What are they doing with it? Is it providing a lot of insight? Is it helpful? I had a client last year that I was talking to and they had been using SSRS reports, and the SSRS reports, they were looking at 155 page reports. Like, no, no, no, this is providing value. And I'm like, no, it isn't. Nobody's looking at that. You need to provide information, and it's just not there. So you need to do that by elevating what's important to people. And you can do that through things like Power bi, which is designed for that.

Okay, yeah, I can certainly see how. 150 page reports. Not many people are going to read it all the way through line by line. Like, who has time for that? Okay, so, Stephanie, just on a similar subject, have you seen any good examples of business intelligence use? Like, any success stories?

Yeah, and so much of what Ginger said, you know, we're all kind of drowning in data. So I know for me, when I started, I was just excited that I could actually put some data together and throw some visuals on a page and say, look, it's pretty. But that's not enough. Just like Ginger was saying. Right. We really need to enable people to get insights from the data quickly and easily. So, yes, definitely, you mentioned that. I was at the Elizabeth Glaser Pediatric AIDS foundation before, and that was definitely a huge success story. As a matter of fact, there was so much manual effort to taken months, every month, every quarter to provide reports to our donors. But it didn't allow enough time for people to actually gain insights and think of new ways to make change and improve our processes. Until we were able to automate that through Power bi, it wasn't about getting rid of jobs. It was about freeing up people's time to do something actually valuable. So that was really my first aha moment with, wow, these tools can really make a difference.

Okay. I like the idea of automating stuff and freeing up people's time in order to focus on something else that's going to be more valuable. So on that note, since you've both mentioned we're kind of drowning data. There's always just tons of reports going on. Many of them are too long, many dashboards. Do you have any advice on where you can spend your time in order to have a big business impact? Are there any sort of common themes around this? And it's a big question. It's probably going to vary from business to business. Go on, Ginger. Do you want to take this one first?

Well, it really. It's. It's all really about listening, because how are you supposed to know what's important to people unless you ask them? I Know, this sounds really novel when we're talking about tech, but I find that a big problem a lot with like, I come in to fix things that are broken primarily. And so there's a lot of data warehousing projects. They're like, yeah, marketing people thought this was important. We don't care. So if you're not tracking the information that's important to the business, I don't care how pretty it is, it's not going to provide value. So it's, it's a lot about getting that information that people want and then finding out where that is available and presenting it to them. But I think there's a really big disconnect between what people put together unless they're talking to the users and finding out what they really want.

Okay, that sounds terrifying. Speaking to people in other teams, each other, they've got different incentives. Okay, we're definitely going to have to come back to this in a moment, but Stephanie, do you have any advice on where you're going to have a big impact when trying to improve your BI capabilities?

Yeah, and it's kind of two parts. I think there's, I think viral adoption is great and you want to really be able to empower users to just go ahead and get started. And you're going to find that some people get really excited about what they can do with these tools and they want to learn more and that's fantastic. So you want to empower people. You don't want to kind of lock things down too much so that they're, they can't do what they want to do. But on the other hand, without training, they're going to run up against some roadblocks. They might do things inefficiently, they might focus a lot on making pretty visuals as opposed to really building up that foundation of a good trustworthy semantic model data set, for example, in Power bi. So I think focusing on making sure that you can have these endorsed and trusted data sets that people can take and do lots of things with, they can go make their pretty visuals, they can analyze that data to their heart's content. That's an area of very important focus, that foundation. But like I said, at the same time, you do want to let people maybe go off and make some mistakes while they're learning the tools and get excited about the possibilities.

Okay.

It does seem like there's going to be a trade off and I think different organizations sort of fall on, have different areas of this way. Some organizations want things like very strictly governed and very tightly controlled and make sure everything's like done right. And in the same way in other organizations have more of a sort of wild west free for. It's like each individual is in power, but maybe you've got a little bit more chaos going on. Okay, so maybe we need to get into some of these things in a bit more depth. So Ginger, before you mentioned the idea that you need to communicate with your colleagues and find out like who actually wants to look at the dashboards you're creating. Do you have any advice on how you can get data teams and perhaps like the commercial teams or other end user teams to communicate, communicate better?

One of the things that I think is important to have happen is it really does not need to be. It needs to be a collaborative relationship and not a competitive relationship. And I think that that's again, we're down to talking about people again and getting it to be collaborating with different business groups. Because I think that Power BI is designed to be important to have central data management so that you've got one version of the troop. However, individuals are the ones going to be knowing enough about the data to create the most meaningful reports. So it needs to be a collaborative effort when it comes to gathering requirements and also teaching people. To Stephanie's point, yes, people need to learn a lot of stuff because you miss things if you just, you know, are using your search engine every time you have a problem. So resolving some of those issues by providing perhaps a center of excellence within your organization where people can find the information that they need. I've helped set them up to starting out with Power bi. A lot of times we encourage them to generate monthly or bimonthly meetings with experts in the organization to say, hey, this is the model that we have the data that you're looking for, so you don't need to go create it again. So save everybody some work too.

I like the idea of training. And you're going to want people to be able to use the tools you just bought. Maybe. On that note, Stephanie, do you have any more ideas of, for example, what do the end users need to know about Power BI in order to make use of this?

Actually, that's a great question. Before we started this, Ginger and I were talking about all the training that she's about to be presenting and for a customer. And I wanted to mention I had a customer who asked me to do a half day end user training and I thought, oh, that's how, what am I going to do? How am I going to fill up four hours with end user training? It turned out to be very easy. There was so much that isn't obvious from when you just start using the tool as an end user, but so much functionality that's available. So I do think that valuable half day end user training just so people can understand, how do I, how do I drill into this visual? How do I make sure that I'm understanding how this visual impacts this visual and what else can I do? What other insights can I gain is actually pretty key, I think. Yeah, end user training just for people, just for consumers, not even developers of reports, purely for consumers of reports is pretty important.

Okay, can you just go into a bit more detail on what did that four hours consist of? What are the things that end users need to know?

Yeah, I mean, I think a big one for me is that people, when they start looking at the visuals, they get excited, they want to answer a question that they don't have the answer to readily. So they'll click the three dots and see, oh, I can export to Excel. That's a default thing. We're all, you know, that's a habit. Let's export to Excel. But no, we don't want to export to Excel. That's last resort. Because the moment we do, then we've got stale data. And then if we want to do that, answer that same question next month. Well, now we have to export to Excel again and do whatever process we did. But built into the tool, you can actually instead connect Excel to your data model. Then you will always have fresh data. But that's not necessarily obvious. That might take a couple minutes to show someone. Just little tricks like that to make sure that you are using the data and your time as efficiently as possible.

Absolutely. It's very easy to have manual processes just creep in there by accident. That's a good trick. Ginger, since you're doing training as well, do you have any advice on what do end users need to know about Power bi?

Well, end users really need to know is you can find a lot of information on sites such as Data Camp, of course, for how to learn about Power bi, which is very helpful. And I think that's a baseline that people need to have. But on top of that, it's like, where is the thing that we use in our organization? One of the cooler things about Copilot that I saw inside of Power bi, technically it's fabric, but documenting here is a measure. This is what it does. Because so often developers go through and they create models and people are like, what's that mean? What does that name represent? Kind of thing. And so if you, if you have just labeled, you know, where the stuff comes from or I've actually developed reports for organizations and we have a glossary page. And the glossary page is just text where it's like this came from, you know, the ACME internal system. And then we list like. And the data for, you know, all of the HR came from, you know, our human soft system. And that is where that and that data is updated daily. So that it's very easy for people to go look at it because it's the last page of the report. It's a glossary and the important concept and also the purpose of this report. Because a lot of times, especially with large organizations, there are thousands of reports, people be like, well, I. This, you know, this calculation, this KPI is not in that report. No, it's not going to be. It's in this report. So listing the purpose of the report so people don't look for like, how come I can't find finance stuff? Well, because you're looking in the accounting report and if you, if it's something silly like, I'm like, are we seriously going to be doing glossary pages for all the reports? Like, yes. And then talking to the users, they love that. Very low tech. Fill a page up with text for people.

I do like that. A lot of those things are relatively simple to implement and yet almost always not done just yet. Documenting where the numbers came from somewhere near your dashboard so people can find out.

I'm confused. I thought developers love documentation, don't they? It's not the first thing they do.

You know, interns are great for that. Just saying. The best documented system that I ever worked with was in New Hampshire. And the universities in New Hampshire mandate that all of their graduates have interns. And so the interns rule the doc.

That's a nice solution. Lots of interns. Also possibly quite expensive. I'm wondering whether, since a lot of these generative AI tools are being built into Power bi, can you use AI to write some of the documentation for you?

Yes, yes. Yeah, it'll put all the field documentation for you. Like anything else, it does generate it. So, like, you don't want to be advised to use glue to put pepperoni on pizza. You do need to review it to make sure that the information that it generated is accurate and reflective of what people are interested in knowing. But yeah, it can generate. Copilot can put in a label on every single table and every single field inside of your data model, which is a kind of A cool feature that I think is great and hope more people use. I've not used it enough to be able to attest to the quality of it one way or the other. But most generative AI, they say it's 90% accurate, which means you need to fix 10% of the things.

Okay, so that's your tenfold productivity boost right there. That seems pretty good.

And then to take that to the next level, Microsoft Fabric, which Power BI is now a part of, has all these new features available so you can get really inception like with it. So you can use fabric to actually query your model and get those descriptions and then put them back into the model. Boom. So you actually don't even need to do it by hand anymore. But to Ginger's point, it's a copilot, it's not a pilot, so you don't want to rely on it without double checking.

Okay, that seems pretty cool. So you're using AI to query the model, to improve the model, and then hopefully things are getting better. All right, so we talked about training for end users. You had some ideas, a couple of ideas there for people actually developing with Power bi, so the data teams and so forth. How should organizations approach this? Like training the developers to make sure that you've got high quality dashboards and things for everyone else to use. What do developers need to know? Stephanie, do you want to expand on that?

Yeah, and this one is tricky because you're not always going to know who the people that are going to really click with the technology are. I think like Ginger mentioned having a group, some kind of a community of practice where people maybe they take a training and they start developing semantic models, which is the name for the underlying data set for Power BI reports. And then they can talk to each other, ask questions, maybe get some more training. Because, you know, you do training once and then if you don't actually put it into action, you're going to forget. But you know, you're going to have to come back and forth a little bit. But I know for me, I have found that sometimes the best semantic model developers are, they're not in it, they're somebody over in marketing. Then they just got really excited about what they could do. You know, they saw the benefit, they saw how it was going to save them time. So I think even just finding the people that these tools resonate with is an important step. So you can't plan to just, you know, pick the people out. Do us do a single week of training and then let loose. It's really Got to be iterative.

Okay. Yeah, I like that idea of having ongoing training throughout just to make sure that people's skills are up to date and they have the right skills at the time when they need to use them. All right, so sounds like this training is going to be like you need the data team people to know about this. You can use many users to know about this. And it's maybe not clear what everyone needs to know. Wants who should be responsible for organizing like making sure that everyone has the right Power BI skills. Is this going to come from a chief data officer? Is it going to come from someone in learning development who should be in charge of this?

Our favorite answer. It depends. Right? Okay. I think it depends on the organization. You know, it really depends on your structure. Is this going to be kind of a bottom up evolution or is it going to be top down driven? So deciding who needs to be responsible for that, that's that it depends. But it's definitely a role that somebody needs to be able to focus their time on. Without managing training and governance. Things are just going to sprawl and get out of control.

Totally agree with what Stephanie said. Yeah. The one thing that I will also say is that you need to also make sure that it's not just it. Because Power BI along with fabric is designed for the subject matter. Experts who know the data best would be the ones who are best qualified to write appropriate reports. Hopefully it has made that easy for them by providing them the data that they need. Well documented. But that means that you are telling people who maybe aren't very it like that, hey, I know you've been but, but I can tell you these people have been using Excel. So it's not like they're not you know, doing have been doing reports, they've just been doing them in Excel. And I think it's important that training not just be for it but also for anybody you want to create a report. And I think anybody that includes anybody who's responsible for any kind of data driven decision making within the organization. It's not just it, but every person who is in the organization that handles data should be doing Power BI reports. They shouldn't wait for it to develop them for them. That's the best way for them to get what they want. Make it yourself.

Okay, so that sounds like pretty much everyone in your organization. Then it's going to need some kind of training around at least how to understand dashboards and how to understand visualizations of reports and things like that. So this isn't just Me like a small scale training, it's going to be pretty large.

I agree.

Okay. And I guess in order to get that kind of adoption and like make sure everyone's comfortable with this, a lot of processes are going to have to change. So typically what you need to change around your processes in order to make sure people are actually making use of the dashboards you're creating, making use of the business intelligence tools. Stephanie, do you want to go first?

There are a lot of monitoring tools within Power BI fabric. I may use those interchangeably now that Power BI is a part of fabric. But there are a lot of monitoring tools. They're really important to watch because you've gotten some requirements, you've followed Ginger's plan of talking to people and getting everything you think you need. Then you build this report and then you find out nobody's using it. Well, how do you find that out? Because you have these monitoring tools so you can watch, you can see how many people are looking at this report, what are they doing with the report, and if you find they're not using it, then it's time to go back and talk to people some more and understand that there's also. You may find we didn't talk too much. We've talked a lot about end user training. We didn't really talk so much about the semantic model developers. Their training. That's going to take some really important skills that they're going to have to learn because without it, it's very easy to create inefficient mod that perform slowly and hog up all of your resources. So you also need to be monitoring that you need to look at. Okay, I do have this report. It's being used a lot, but it's hogging up 80% of my resources. Why is that? So Frequently I find that we unleash Power BI on an organization, but there's nobody who's got part of their job responsibility dedicated to actually monitoring what's happening with the usage and the capacity. Then they think, well, Power BI is not great, or we need to pay more for a bigger capacity. When in reality you might just need some more training for your model developers so that they can make a more efficient model. There are so many things to monitor. It's not super easy and straightforward where to find them all right now. But I do think that is improving quickly and it will become easier and easier to find this information. It does exist, but you have to know where to look for it.

Okay, I do like that idea of just monitoring what's being Built like who's actually using dashboards? If no one's looked at the dashboard in a few months, then probably means that no one cares and you can get rid of it. And certainly like monitoring resource consumption as well is going to make sure that you're not just crashing half the servers or getting a very expensive cloud bill. Okay, so Ginger, do you have anything to add to that? Like, are there any areas where you think, okay, we're going to have to change this process if people actually make use of the business intelligence tools?

You know, I really think Stephanie raised a great point because I've not been any place where somebody's job was to monitor what's going on, which is, you know, unfortunate because you have that with databases. I will say to help you with that, Microsoft does have a fabric capacity application that they have created. Now, that's not going to help so much with Power bi, but if you have a whole bunch of people doing who have created semantic models that have the same data in it, you'll be able to see that there. That is a great tool that is free. You can download and implement right now if you're using Fabric to help you with this big monitoring tool. Because there's a lot out there for people who are doing Power BI reports, though they can all look at their statistics. I agree with Stephanie. I don't think very many people are and they should be just a couple of clicks. You can go to the ellipse menu, find out the utilization for your reports and I highly recommend anybody who's created a report to do that.

Okay, nice. So since you've mentioned Fabric a few times, maybe we just need to take a short digression into tooling and where Power BI fits into this larger ecosystem of Microsoft tools. Ginger, do you want to go first this time? What's the relationship between Power BI and Fabric and everything else?

Sure. Well, as people have been using Power BI for, I think it's 10 years now, Microsoft realized that there's a much larger data picture, that Power BI is merely the tail end of the and that is all the data that they have in the organization. Power BI uses the end and Fabric uses the back part of it. Where you're like, I need to get data from my on premise Acme system. I've got data in an Azure SQL db, I've got some stuff in Snowflake. It's scattered here and there. But yet as an organization, we want one version of the truth. Fabric is designed to bring all of that data into one location and that One location is the onelake and then use that as your source for your Power BI models. This is a switch from a lot of the Power BI development that people have done up until this point. Because a lot of people I have seen some just amazing M code that has been written to do a ton of data manipulation so that I've got the perfect semantic model for Power bi. And kudos to you for doing that. But maybe it might be more efficient and easier to track and easier to incorporate with other data sets. If you did that maybe in Spark and then just put that cleaned up end process semantic model inside of the data lake and people connected to that, maybe it's moving your tooling around. Maybe that there is a better choice than M to do a lot of your data manipulation. And that's what fabric brings to the table.

Okay, so to me that sounds a bit like it's kind of the World Wide web for data. Like if you want to go to a website, you don't want to care like which service hosted on, you just want to say, okay, this is the web address. And so in the data example, you don't want to care which data silo the data set stored in. You just want to say, okay, I want this data set. So is that the general idea?

Yeah, the general idea is to have one place where you can go for all of it. And it doesn't all have to be within Microsoft space. You have data in aws, fine. You have data in Snowflake, fine. They don't have a connection to Google as of now, but I would look at six months and see if that changes. Because they really want people to be able to centrally locate their information because that's a pain point organizations have. You can't report on it if you have it in different places. How are you going to do that? Yes, you can do that into semantic model and Power bi. But is that the best choice? Especially if you're talking terabytes with the data?

I can imagine a very big data set, so you don't be manipulating it in Power BI at all. That's going to be horrible. Lots and lots of data manipulation tools out there now. Well, yeah, that's a whole separate data engineering webinar. But yeah, it's good that you can make use of that and then feed everything into Power bi. All right, Stephanie, is there anything else about the Microsoft tooling ecosystem that you want to tell us about?

Yeah, I mean, there's lots to talk about with fabric. We could go on all day with Fabric. I'm pretty excited about all of the developments with fabric on top of Power bi. My background was SSRS Power bi, not data engineering. So when fabric came came around I was a little bit intimidated, like, oh my gosh, now I have to learn all these new skills. But it's actually fairly straightforward that I want to say that the modules, the learn modules from that Microsoft has developed are incredibly helpful to kind of get your feet wet with understanding all of these different parts for fabric, especially the data engineering pieces. But beyond that, a couple of things that I think really make it worth looking at even if you're just focused on Power bi, are that the new licenses, the licensing model for fabric has a much lower entry point cost wise. And you get all of these amazing benefits that you get with Power BI Premium, but at a much lower entry point. That was always prohibitive before, but with fabric you get all of this and then secondly, it really does unlock your data. So even if you are doing a bunch of data manipulation in M through the model, which again I agree with Ginger, push that upstream if possible, maybe do that in Spark. But now with fabric you can actually access that model data with tools other than Power BI and Excel it opens it up. We're talking about self service. That doesn't just mean self service Excel users, that might mean self service data scientists. They want to access the data in the form that maybe the Power BI developer has done all of this manipulation and logic on. They couldn't do that before. They could only access it with Power BI in Excel. Now you can access it with a notebook. You can just directly hit that semantic model through a notebook in fabric. So it really unlocks things and again, like I said, makes it an easier entry point cost wise. So if you want to try out, if you're new to Power bi, don't overlook fabric. There are a lot of amazing features.

This is really interesting and I do like the idea that you can use notebooks as well. So the data scientists get to do similar things to the data analysts because different end tools preferred. I'm curious as to who needs to care about which bits of this. A lot of the fabric stuff sounds like it's maybe more targeted towards data engineers rather than and business intelligence people. Like, so what do. Like who needs to know what and do business intelligence people need to know anything about fabric or, or not? Ginger?

Well, I do. I mean you do have to know something about it. But I think if you are focusing on business intelligence at some time, you know, you may think, well it. I would be best suited to providing reports in my organization. And maybe somebody else can get that data for me faster than I can. So again, working out with, hey, maybe somebody can help me with this, and they could do it in three hours and it would take me two weeks. So look at the skill. And also other people might benefit from that centralized data storage. If I need to get data, like, I don't know, census information for some of the reports that I'm using on, I'm probably not the only one who could. Could, you know, benefit from that within my organization. So maybe again, communication and working with people who have different skill sets, except in very small organizations, a lot of places have people who have different skills utilize them. Not everybody can do data models, and not everybody is good at presenting data visually in appealing and compelling format and work to your strength.

Okay. Actually, I have to say, as much as I like working with data, having other people do work for me is. Is brilliant. So I do have asking other people to do stuff. So I guess this leads into, like, what does the ideal sort of team set up like? So you said you need, like, someone who's good at presenting, maybe someone who's good at building dashboards, maybe someone who's good at, like, organizing data. What's the. What's the sort of complete skill set you need across an organization? How do you divide them up into teams? Go on, Stephanie.

Yeah, again, that's tough.

It's a big question.

Yeah. And I think it depends on the size of your organization. I mean, in some organizations, you know, you might be the one to wear all the hats, and you're going to have to learn a little bit of everything. But maybe in a larger organization, you might have some dedicated data engineers and dedicated semantic model developers and dedicated visual developers and dedicated scientists. So you might have all of these different roles. I think it's important to have, you know, like we talked about, certainly someone even. Even in a small organization, there has to be someone who's. Part of their role is really this governance and monitoring and making sure that you're getting the best value out of your investment. But in addition to that, I feel like I'm really hitting that one hard today. But in addition to that, you know, you also. I think a key skill is going to be that semantic model development. It's important to be able to develop an efficient model that can answer people's questions and, you know, not make them stare at a spinning visual, you know, little, little timer waiting for something to load for 30 seconds. And that does take some skill. And Practice. So I think that semantic model developers is really a key, key role. And then for the things like the visuals, you know, expand out and teach a lot of people how to do that. That's fun. I mean, who doesn't like making a pretty chart? So I think you're going to get a lot of people who want to do that.

Yeah, definitely. Database is one of the fun parts of working data. Since you mentioned the idea of a semantic model developer, can you just expand on, like, what actually is a semantic model and what does a semantic model developer do?

Yeah, so we used to call it Dataset, then Microsoft renamed it to semantic model. So we're all trying to get used to saying the proper terminology. So when maybe if you come from Excel, you know, you're kind of used to working with a spreadsheet, like a big wide spreadsheet that has all of the columns that you might want to use to build visuals. Power BI can handle that up to a point. But if you get larger and larger data, having a big wide table as your main underlying source is going to be inefficient in Power bi. It's going to perform slowly. So it's important to understand the idea of a star schema. So you want to have your fact table that's mostly numbers, it's kind of narrow and really long, and then all of your little lookup tables around it. That's the very basic basic. But that's a concept that I think is really hard for people to jump over that hurdle if they've come from maybe an Excel background. And up to a point, you know, they start developing models and they're great until the data gets bigger and bigger. And now you've got a billion rows and your model isn't supporting it anymore. So understanding that basic tabular object modeling skill is very important.

Okay, so, yeah, as I understand it, basically if you're a database designer, you want lots of really small tables that have specific stuff in. And as an analyst, you don't want to worry about joining stuff together because you can never remember how to join things properly. And so you just want one big table with everything in it. And reconciling the two is always sort of tricky. So I guess the semantic layer is just like, hey, this is what it should look like from. In order to solve the business problem. Is that kind of right?

Solve the business problem and be performant. Because the moment somebody has to wait 30 seconds for a visual to load, they're not coming back to your report, you've lost them. So you need to be able to have it answer the questions quickly.

All right, I think I need to go back and take the data database design course again because this is stuff I can never hold about good task the audience. Actually the database design costs very good. So one of the sort of dreams for business intelligence is having self serve analytics. So the data team isn't constantly being pestered for like, hey, can you do this, do that for me? And people just go and create their own dashboards, answer their own questions. It seems like a pipe dream in many cases. Ginger, do you have any idea on like is this possible now can we get to self service analytics?

You can to a point. Power BI has had for years a Q and a visual. And I encourage everybody out there to go put that on your Power BI reports. And if you have a dashboard, go make, you know, show people how to use it. Because you can put a visual on your report today that will allow you to ask, you know, which region had the highest sales in January 2024 and it will give you the answer. You can do that right now. You don't have to spend any more money for an expensive premium license to make that happen. You can do that now. And I don't think that the tool is very widely adopted. Also the tool works better if you have what's called seeded it. So some people might call it just like people throw around different names for things. And if you can tell Power bi, which you can, you basically say, well, when someone says, you know the I work for a company that they throw around location and community and it's the same word. It just depends on the system where it came from. So if you have, if you decide I'm going to call it location and you throw the word community in to Power BI so that it knows that that's also location and community means the same thing. I can ask you which is the most profitable location or which is the most profitable community and it'll automatically know it's not as smart as generative AI, but you can do it right now with the tooling that you have and make it better yourself for a lot of different needs people have today.

Okay, I like the idea that just fairly simple questions, those should just be anyone can do this. I guess I presume you have to have the training and things in place to make sure that happens. Okay, Stephanie, do you want to add to that? Have you got any tips for how you can make self service analytics a reality?

Yeah, definitely. And to Ginger's point and kind of bringing it Back to what we were talking about earlier with user training. That would be a great addition to any user training that you might do is show people how to use that Q and A visual. And another one that I agree with you, Ginger, that it's underutilized that the Q and A visual is personalizing visuals. I don't see a lot of people using the personalized visuals feature, which is where you can enable this on a page in a Power BI report. And it will let people click a little button on, let's say a bar chart. And they say, actually I want to see this as a line chart. And I would rather have my X axis be months rather than regions or whatever. Instead of them having to go out and create their own report, they can just tweak one that somebody else has built for them by this personalized visuals feature. So that's another underutilized one that I think can be added to end user training, certainly. And then definitely teaching people not to go out and make their own models if something already exists. Connect to something and make your own visuals. But don't reinvent the wheel by by building more and more data sets.

Okay, so yeah, Q and A visuals and then personalizing visuals. Those both sound like incredibly important features. Now we've got a ton of questions from the audience, so we're going to go to audience Questions in a moment. For anyone who wants to ask a question, please do put them in the chat. We'll try and get to as many as we can. Before that, we mentioned a few cool features of Power BI so far. What's the one feature or tip you wish everyone knew about Power bi? Ginger, do you want to go the.

One thing that people know about rbi? I like bookmarks and I don't think any very many people know about those. You can create your own personal bookmarks. Let's say that I only care about stuff in America. And all reporting says it's got a drop down box and every single time I have to go and pick it, I can set a bookmark inside of the report that I'm looking at@Power BI. So I see it the way I want to see it. And I don't see users using that feature hardly at all.

Okay. Yeah. Having to do the same pointing and clicking over and over again, that kind of spoils the point of having things automated for you. Yeah, it's going to be very tedious. I like that. Stephanie, what's your favorite feature you wish everyone knew about?

I do have one and I also want to just piggyback on Ginger's and say, yeah, and you can actually do those bookmarks with personalized visuals. So maybe every time you want to see that chart in a different format so you can personalize it and then add that to your bookmark, every time you come to the report, it's going to be in that format. Totally agree with you, Ginger. And then for me, it's definitely connecting Excel to data sets, either through Analyze in Excel or there's a newer feature called Connected Tables. So with Analyze in Excel, you have to make pivot tables in Excel. I don't love pivot tables. I'm not always a fan of pivot tables, but being able to create my own connected table. So it basically just have a table in Excel that looks like any other table, but it's live connected to that data. So when the data is refreshed, my Excel file is also refreshed. So, you know, helping people avoid that whole export to Excel, wasted time of doing the same thing over and over. These Excel connected tables are really cool. Okay.

Yes. And just making sure data is fresh all the time, that sounds like a very useful feature.

Okay.

All right, let's go to some audience questions. All right, so it's kind of a big question from Hamza, but where should we start? So perhaps you're just getting started with using Power bi. Like what's, what's step one that you need to know?

I, you know, I would say there are these Dashboard in a Day trainings. They really kind of give you an overview of Power bi. They touch on all the different parts. They don't go deep, but they're going to show you exactly how all the different parts of it work. So I think just starting with some of these trainings that are available at Data Camp. And Ginger's an excellent teacher. I would, I would definitely look at anything she's got, but you can get started really quickly. But just be aware that you're not going to know everything you know within a day. Once you start getting more complex, you're going to need more training. But there are, there are lots of great short trainings available out there.

Yes, I suppose the obvious answer is DataCamp's introduction to power BI courses is the place to start. And if you want to go even more basic, then perhaps some of the courses on understanding data visualization, a good idea. All right, so next question comes from. Matthew says, oh, this is about team building. So what should the ratio of Power BI report and dashboard developers to data engineers be?

I can take that one. So it this gets back to it depends because here's the thing, it takes a long time to set up an appropriate data source. If you're going with fabric in one lake to have the different models in place, it's a lot takes a lot more effort to create than maintain. So you need a lot in the beginning, you need less going forward. It also depends on who is creating lots of your dax. So if the DAX is being written by your data engineers, you'll need more of them. If your dashboard, if that is not something your report developers are doing, then you will need less. I do think that I have been in organizations though where they had a small cadre of people who did data engineering and everybody in the company did dashboards and reporting. So it can be really, really high. If you consider anybody who writes a report, it could be the entire company and then maybe 10 people who handle the back end.

That's interesting that it sort of shifts over time. So you need more data engineers at the start where you can building out their infrastructure and later on. You mean more people who are doing sort of closer to the end user applications. Okay, next question from Adnane. So I sometimes confronted stakeholders who want to report to get some insight about the business. But when I asked about what are the KPIs they seem not to have, I presume that's not a proper answer. So okay, that sounds like very much a business problem. Probably a far too common business problem. How do you make sure that you have like good KPIs that are, you know, clear to everyone? So I know Ginger, you mentioned the idea of having documented dashboards with this glossary of KPIs. Do you have any advice on how to get to that point?

Well, I've had organizations who basically like, oh, well, you know, we don't have any of those. And what I, what I've done to. And then like, and we'll get back to you. We'll get back to you. So I, what I have done to spark conversation is I create my own and as soon as somebody looks at them they're like, we don't like those. Great, what do you like? Because it's a lot easier to say I don't like that. And then they can tell you what they do like rather than blank sheet of paper. So if you're not getting good input for KPIs, say this is what I thought and expect people to hate that and it to spark a conversation.

I like that. Hate driven development. That's a fun paradigm.

Yes.

Stephanie, have you Got any other tips like that or.

I was honestly thinking the same thing. It's iterative, so giving. Giving people something. But I wasn't going to take it to that level of expect them to hate it. But I like that, so I'm sticking with that from now on.

Yeah, I don't see how we can top that. All right, let's go to the next question. This is from Abhishek saying, can you explain how fabric and Power Bi. What's been. Oh, yeah. So basically, how has the existence of fabric changed Power bi?

Yeah, that is a tough one. If you are going along and you're happy with your whole Power Bi environment, you can continue doing that. You don't have to do anything with fabric. But I don't know why you wouldn't. Just because fabric does bring a lot of really awesome features to Power bi. So I think that the differences are just. We're a little bit unclear on how the paying for a part is going to be in the future. We know that the power BI Premium SKUs are going away and you have to buy a fabric SKU for it, but the details on that are a little bit fuzzy. I would just say don't be afraid, you know, keep going along with what you're doing if you're happy with it. But. But try out some of these fabric features because they are really, really pretty amazing and really unlocking your data, making it available, reducing data duplication. So definitely check it out.

Okay, so it sounds like you can keep doing what you're doing, but probably you do want to start investigating these fabric features as well, particularly like as you get more data sets and stuck in all different silos all over the place. Okay. All right, so we've got a question from Lee. Oh, we've got a celebrity audience member here. So Lisa, a bit of a storytelling legend. So Lee asks, what kind of training do you recommend? Beyond tool training to get the most out of Power vi.

Beyond.

I'm going to hand this to J.

So, I mean, it's obviously, you know, I would say understanding good visualization science. There's. If you are creating good reports that is not tool specific. I have three or four books by Steven Pugh, one of the leaders in that. Does Data Camp have a course on good visualization design?

We have a course on. Yes, there's a course on understanding data visualization concept by me. So I like.

That is what you need. And that's beyond, again, anything that's. You said beyond tool. So that's layout and design and the Science of visual. That's what would be where I would go after. After tools.

Yes. I suppose to add to that, I suspect Lee might have been pointing us towards. You probably need some training around, like storytelling. Like how do you go from this is just a report to like, this is what it's telling me about the business. So lots more content, lots more webinars on that coming sometime in Q3. So yeah, stay tuned for that.

All right, Just add on to that one too. Also, I'm going to make a plug for accessibility here. I think that's always something to keep in mind. So on top of good visualization design, always think about accessibility. Should not be an afterthought. Just add that onto it too. So make sure that whatever you design has accessibility built in.

Oh yeah, definitely. Like, so if you're going to get the many, many viewers of your dashboard, at some point you're gonna get cold blind people. You get people with who have sort of bad vision. Yeah. Probably speaking different languages as well. So yeah, have that sort of internationalization thing. Lots of good stuff you can do around accessibility to keep your audience happy. Okay, another question from admin. So admin asks, I'm not sure what PPU is. Pay per user. Maybe if you're on a premium per user.

Yeah.

Okay, what is a good deployment pipeline? Maybe can you explain what a deployment pipeline is and then say what makes it good?

Yeah, so a deployment pipeline is a feature in Power BI slash fabric where it allows you to move the objects, like your semantic models, your reports, other fabric artifacts, dashboards, from one workspace to another. When it started, it was three. That was a hard and fast rule. So you pretty much would have like a dev workspace, a test workspace, and a prod workspace. So you might have your developers publish their models and reports to the dev workspace after everybody's agreed that, okay, these changes are good, we use the deployment pipeline to then push them through to test. Then maybe you have your end, certain group of end users test them. Then. Okay, we're good. Let's move it to prod. So then you just push the objects to the prod environment as well. So that's like the shortest explanation of a deployment pipeline. I think I could give a good one versus a not good one though. I'm struggling to think what makes a deployment pipeline good or less good. Ginger, do you have any thoughts on that?

Well, if you don't have Power BI premium, it is possible to migrate using the power bi rest APIs is not for the faint of Heart. You can also do it with DevOps. Again, not for the faint of heart. We're talking you're going to need to do some development to make that work. If you don't have a premium skill. I will say that I hope that everybody is currently using the you don't have to pay for it preview feature right now for all Power BI reports and making sure that you have them in GitHub if nothing else.

Okay, yeah. Tracking which version of things you have in different environments, like whether it's GitHub or somewhere else. That sounds like a good idea, I guess. Bad pipeline. You've probably got different versions of Power BI in your development environment versus your test environment versus your production environment or different settings. Making sure those are consistent probably sounds like a useful thing. Katie asks how does Power BI work with other cloud platforms? For example, data is on Google Cloud platform or Data Lake on Amazon Web Services. Yeah. Is Power BI sort of it works best with other Microsoft tools or is it sort of cloud independent?

I think Ginger touched on this earlier.

Yeah, I think GCP is maybe coming soon in a few months, but we're.

Not sure it's coming soon as a shortcut into the fabric lake. But I've created reports for in from Amazon. It doesn't really matter if you are not if you're not using so if you're not using Amazon right now as there isn't a solution for your limit or you are limited to a 10 gig limit. That would be the limitation that you would have if you've got your data on Google Cloud for your specific model unless you pulled it out of the cloud. So that's not that big of a. I've created many a model. I end up getting a lot of data that people have from third parties that just they get FTPed somewhere and that's. I get it's not the same thing, but it's a similar process. It doesn't matter if you are importing your data into Power bi, using Power BI to create your semantic model. I don't care if you've got data from aws, Google or a flat file. It's all the same to me once pull into Power bi.

Okay, sounds like have your data anywhere and you should be okay, nice. All right, this question comes from Slade. So Slade asks do you have any tips on getting around the issue Power BI has to do with publishing composite models? Okay. You can have explained what a composite model is and then yeah, how do you get around this issue or what is the issue and how do you get around it.

I'm not sure about the issue and I saw Ginger nod, so I'm going to punt this to Ginger.

The issue is that if you use composite models, you're in directquery. The resolution is please don't use them. There are better solutions than composite modeling. I recommend that if you're solely doing Power bi, instead use dataflows. There are many more options with fabric, but I do not recommend the use of composite models because they're horrible. Use data flows instead. If you're talking straight.

No, composite models. Composite models bad. Let's avoid those. All right, so we've got, we're coming to time, but there are some more great questions from the audience. So we're going to run a few minutes over for anyone who has to jump, please do catch up with the question. The rest of the Q and A on the recording, make sure you register in order the the recording comes through to you. Before we go to the rest of the questions, I just want to say we've got some great sessions coming up. So next Tuesday we've got a session on creating AI assistance with GPT4. Oh, it's me, me running you through that. And then next Wednesday, a session on data governance with Azure. So if you're working on a Microsoft stack, then that's definitely gonna be one you want to attend. And then after that, in a week, there's a session on creating data pipelines with airflow. So some great sessions coming up for you. You QR code on the top left of the screen, please do go to that. One other thing is that we've got the Radar virtual conference coming up on June 26th and 27th. So we've got a lot of incredibly high profile guests coming along with that. So it's two days of talking all things AI. I'm very excited for it. And please do register for that. So again, there's a QR code for you to sign up for that. All right, let me do two more questions and then we shall wrap up.

Okay.

So, all right, we're going to start a little platform war here. So Ryan asks what the benefits using Power BI over Tableau. Yeah, when might, how might you go about choosing which BI tools to use?

Definitely bias. I think we're seeing a lot of investment in Power bi. So many features being added with Fabric. If you're looking for an all in one analytics platform. Fabric and Power BI has all the tools built in so you can do all the data manipulation you need. From everything I hear Tableau has, you know is superior on the visual, on the visuals that it includes. That may or may not be true. I don't know. But if your focus is purely on visuals, maybe tableau. But if you really are looking for that whole analytics platform with these underlying semantic models that support all of these different analysis, you know, again, I'm biased. I would say Power bi.

I'll also think that cost matters too. And you don't want to start with a tool that's cheap that nobody uses, because there's probably a reason for that. I know that there's a lot of students who use datacamp. Get started with Power BI and don't pay any money for it. You don't need a license to get started with Power bi. It's the full version that you would get if you pay the premium edition, if you download desktop. So spend your money on training to get started instead.

Amen to that.

Yes.

I like that one thing. Probably if you are in an organization, you want to use whatever your colleagues are using. Not a completely different tool though. So good to be synchronized with your colleagues. You can collaborate. All right, let's do one more question. So Edgar asks, what are some trends in business intelligence that would you attention to?

Yeah.

What are you most excited about?

A lot going on with AI? I mean, I'm definitely gonna go with. Yeah.

All about the AI. AI is. Is making everybody's life so much easier. Can tell you what. There would be a development team on a previous project that I worked on that I would have replaced with co pilot to build plane reports. Because if. If you don't care that much about how. They don't need to be perfect. They just need to be great. You need them right now. Use copilot right now. You'll have your reports in an hour. You'll have something. You'll be able to show the progress and you didn't have to do the work.

Nice. Yeah. AI doing the job for you. Absolutely brilliant. I like that. So it seems like you both agreed on the same thing. So at this point, I think we'll. We'll wrap up as well over time. So thank you so much, Stephanie. Thank you so much, Ginger. I had a really good conversation. I hope you enjoyed it too. So many insights there. Brilliant. Okay. And thank you to Adele for moderating. Thank you to everyone who asked a question. Sorry we didn't get to all your questions. Very enthusiastic audience today. I'm very impressed by. Thank you again and thank you to everyone who showed up. I hope to see you all again for future webinars and for radar, again, please do have a look at that QR code link and sign up for it. All right, thank you very much.