Small Business with Power BI and Fabric – Ep. 437

Mike and Tommy tackle a mailbag question from a small business struggling to justify Power BI and Fabric over their existing SSRS setup. They break down licensing strategies, the value of the semantic model, and why even small organizations can find the right price point for Fabric.

News & Announcements

Tommy dives into his experience with the new Prep Data for AI feature now available in Power BI Desktop. The conversation covers which measures and columns to expose to Copilot, the role of perspectives in semantic models, and how verified answers can improve the AI experience.

• Natural Language Interaction Isn’t Magic—It’s Semantic Modeling Done Right — Data Marc explains why strong semantic modeling is the foundation for reliable natural language analytics. He highlights how grounding Copilot and data agents on semantic models improves accuracy, consistency, and governance.

• Automatically Populate Data Agents with Semantic Model Synonyms — This post shows a practical approach to enriching Fabric Data Agents by programmatically extracting synonyms and descriptions from a semantic model. The goal is fewer implicit measures, better measure selection, and more trustworthy answers from conversational experiences.

Main Discussion: Power BI and Fabric for Small Business

A listener working with credit unions and small businesses asks whether Fabric makes sense when the organization already has SQL Server and SSRS reporting, doesn’t want to pay for premium licensing, and is currently sharing PBIX files via SharePoint.

Prep Data for AI — First Impressions

Mike recommends exposing only simple, core calculations to Copilot — base sums, averages, and clean dimensions — while excluding visual-specific formatting measures and hyper-context calculations like those using ALLSELECTED. Tommy highlights verified answers and the ability to tune what Copilot “sees” versus what report builders use.

Mike argues perspectives are underutilized and should be reusable in the Prep for AI workflow. If teams already curated perspectives with third-party tools, they shouldn’t have to reselect fields manually every time they configure AI.

The Semantic Model Is the Real Value

The hosts emphasize that Power BI’s biggest value isn’t prettier reports — it’s the semantic model. With SSRS, every report carries its own query and locked-in business logic. Update a definition and you may need to touch dozens of reports. A single semantic model centralizes definitions so they can be reused across paginated reports, interactive reports, and even Excel connections.

Licensing Strategies for Small Businesses

Mike makes the case that an F2 Fabric reservation (roughly $150–$160/month) is a practical entry point for many organizations exploring Power BI + Fabric. He also suggests evaluating Microsoft E5 licensing in regulated industries: if users already need stronger security controls, E5 can consolidate tools and includes Power BI Pro.

Stop Sharing PBIX Files

Both hosts strongly advise against sharing PBIX files via SharePoint. It’s effectively the same pattern as distributing Excel files: no single source of truth, inconsistent refresh behavior, and limited governance. The Power BI service is where a shared semantic model, security, and reusable definitions deliver the real leverage.

When to Stay Put

Mike offers a balanced take: if your organization’s workflow is purely emailed SSRS output and it meets needs, migrating to Fabric just to reproduce that workflow isn’t worth it. Fabric shines when you need interactive experiences, self-service analytics, faster iteration, and/or you’re already evaluating modernization or cloud migration for your existing data stack.

Looking Forward

Mike encourages small businesses to segment their audiences (viewers vs. creators vs. model builders) to find the right licensing mix. Both agree that shifting from emailed snapshots to governed self-service is the real transformation—and that Fabric’s lower SKUs make experimentation feasible at smaller scales.

Episode Transcript

Full verbatim transcript — click any timestamp to jump to that moment:

0:01 [Music] Good morning and welcome back to the Explicit Measures podcast with Tommy and

0:33 Mike. Good morning everybody. How’s everybody doing? Oh, good morning Mike. It is another good week. We are in the throws of summer now. We are getting ready with all the kids at home. Changing our patterns a little bit here. So that’s that’s that is our life here in the US. I guess for those who listen internationally, not everyone does the same schedule, school patterns that we do. So, yeah. any any fun summer activities that you’ve been doing, Tommy? Honestly, they’re the kids summer has been so hectic. They’ve had

1:04 Vacation bible school. They’ve had a lot of pool. So, I haven’t full time. It’s great. I haven’t even activated a contract that we talked about where they got that work in the morning because I’m I’m going to be, gracious here where if they’re busy from like 9 to two, like we’ve had a full day. So, but by then they can kind of like chill and mellow out in the afternoon and stuff like that as well. Yeah. Yeah. My kids are at an interesting age right now. We’ve got 11 through 15year-old now and so we’ve been just kind of play we

1:35 Just did one where we just played out in the evening. We just kind of took it out a little bit e easy in the evening. We we shot some basketball hoops. Have a little hoop that we have on on the street that we play with. we had little scooters and we were scooting around making like tracks and racetracks around the neighborhood. We live on a culde-sac so it’s easy for us to like we could take sidewalk chalk and like I was like go out in the go out in the culde-sac and like make a little stop like a track and I’ll time you. And so that that took up like two hours of just making tracks and zooming around and like, “Oh, this is this is we’re

2:05 Favoring our lefthand side too much. We should probably do some more right-hand turns.” I’m like, “Okay, all right. Interesting.” Like, yeah, it was it was very fun. kind of getting outside and playing all together as a family, which was which is great. All right. Anyways, let’s for our main topic today, we’re going to do a mailbag question. This is also a recorded episode, so just be aware this has been pre-recorded for this episode. our mailbag question is going to Tommy will read it here. It’s going to center around small businesses and fabric and where does that fit and what are our options here for small

2:35 Businesses and should we start thinking about using fabric? Does does this help us at all? So, that’s one area we can we can look at here. This will be our our main topic of discussion. But Tommy, do you have any news any kind of beat from the street things you’d like to talk about? So, there’s a few actually nonfabric blog news items I wanted to cover. And then honestly just been working a ton with the prep data for AI feature now available in desktop. Yeah. And just came out. Yeah. Just came out and I’ve been diving in head first. So,

3:05 I want to get your take on it. I want to get kind of what your feedback is that workflow for if my goal is to have a semantic model that can be again a data point or the data source for someone to talk to their data with. Is that workflow and the features that they have right now sufficient? Do you like it? What are your thoughts? initial blush is I like I like the selection options, right? Which tables to emphasize over other tables and columns. And that makes a lot of sense.

3:35 There’s a lot of things in our models that we really don’t want the AI looking at or helping us with or producing information around like the ID columns right those just don’t make sense right there’s no reason to have those some of the measures and things you may you may take them out honestly because when you build a semantic model I look at the semantic model kind of two ways there’s like core calculations calculations that are calculating the sums the averages the the base things then you have more elaborate calculations where you’re doing like filtering and some kind of

4:05 Aggregations on things to be a little bit more complex. But then there’s also calculations that are like very visual focused like I’m doing formatting like when this number is above this threshold make it a red color or a green color. And I kind of would argue like a lot of the visual formatting measures don’t need to be in in place or you being used with the semantic model or for prep for AI. Those are things I want to skip, right? So I I think if we’re talking about like things I want the AI to be aware of, it’s the simple stuff, right?

4:35 That that’s the kind of those are the the simple things I want you to to use to aggregate by, right? I don’t want you using my really complex aggregated filtered table with a whole bunch of extra things like you have to understand that DAX very intently. And I think there’s this, and again, Tom, you can correct me if I’m wrong or maybe I’m going off the rails here, but but there’s this interesting interplay between what DAX you build and what does the visual do, right? So sometimes you’re building Oh, context, right? And

5:06 But sometimes you build the DAX very specifically for what the visual needs to output what you want. Yeah. Right. Right. It’s that kind of it’s that kind of stuff that I’m like I don’t really think the AI needs to know about that stuff. I I don’t think it helps the AI a ton to have these hypersp specific calculations just for a single visual. Yeah. anytime you’re using something like keep sele like all selected or you’re doing something that has that kind of hyper context it’s I

5:36 Don’t think those so far have not worked great with copilot because it’s not going to understand it when someone’s asking for a particular med metric right that it only really works under the context of these fields so but that’s the nice thing though about the I’m really enjoying this prep data feature I know there’s a few more things I would like to had. But again, just the ability to split out the semantic model both almost like a a perspective

6:06 Where I have a view for just what people can see in a semantic model, but then just for co for AI for co-pilot where what these metrics while they’re essential, I don’t want people they don’t need to really be communicating with that and that doesn’t affect people’s ability to actually use those metrics or those fields in the report. I feel like the whole perspectives inside a semantic model are kind of underused honestly. Like I think I think you could use a perspective because perspectives are like a subset

6:37 Of information that’s about the whole model, right? It’s it’s picking different measures, different columns that you would show inside that perspective. Now, I don’t know if this is what it’s doing. And again, I I’d have to look a little bit more closely at the timal that’s produced by the prep your data for AI. Tell me what what happens inside there. What is it actually doing when you do a prep data for AI? Is it adding annotations? Is it adding extra things inside the model? What is it doing to help the model get smarter about those like that the AI? So there there’s a few kind of features

7:07 That are all going in place. One again is the which column which fields to actually show for co-pilot to be aware of. But then there’s also Hold on. I’m talking about are you talking about timle? I asked about tindle specifically. Like like where is it like you’re prep? So, here’s my thing. I was I was trying to narrow you down. Trying to say, “Look, yeah, I’m prepping the model using the PowerBI desktop. Therefore, it’s doing something. It’s saving some information somewhere. What is it actually doing in timle when you use it?” That’s what I’m that’s my

7:38 Question. Not Not the actual UI screen. No, I I gotcha. I don’t I haven’t looked too in depth to see if there’s actually anything written in Timle when you do the data. So, because the other thing too is like one of the main features on that prep data is that verified answers feature. Yes. Which is basically tied to if someone asks a certain question, it’s going to show up with that visual. I’ve been loving that. But see, that wouldn’t necessarily be in the semantic model, I guess, like you

8:08 Know, what I mean? Like would that be in the semantic models definition? It’s that visual. I don’t know if that is supported in Tindle. I’ll have to dive a little more, but I have a feeling that’s one of the metadata things that lives in the service. The reason I see where you’re going with that where we can do some little bit of automation around that too, but no, it was it was not that. I was just I’m just purely curious like anytime we do something inside the model. So again, I was going what I was hanging on my comment there

8:38 Was around perspectives, right? So the perspectives are something that you can use the perspectives select a subset of columns. One of my feedback items here would be, in models that are really large, there are models, there are models that have a lot of tables in them. And so you could spend a lot of time combing through the different tables and trying to figure out what’s important, what’s not important. So, one of the things or feedback pieces I would have here is in the simplify schema page on the prep for AI, I think it would be nice to have

9:10 Like a drop down for this for the perspectives because you’re going to I mean, one, they’re underutilized. I think they should be used more. And then two, if you’ve already spent time making perspectives inside a semantic model using like a third party tool, like why wouldn’t you want to reuse them? Why wouldn’t you want to copy one of them and just make a couple more adjustments for the AI and say, “Hey AI, just use this perspective when we’re training tables and that way I wouldn’t have to go back through the UI of the AI every single time. I could just kind of slightly adjust the perspective as I

9:40 Need.” And there’s other things here like verified answers and AI instructions, which interesting. But again, those are like I see where you’re going. Yeah. you’re focus and and it’s perspectives are so underutilized what they could be too because imagine world too where you can tell co-pilot hey when people are asking about these types these types of questions or topics use this perspective right so it’s not just a so I I love that idea but that’s only that first step the ver the AI instructions I’m I’m really curious about how this works because I would

10:11 Love to see the backend kind of information that’s actually generated when I’m doing this because this This is cool and it I I like the workflow. It’s better than the Q&A kind of setup here, but I don’t think we have all of the features in desktop needed for this to be hyper effective. So, and that kind of goes with the from our our good friend Data Mark who actually has two really neat articles that kind of expands on this prep data for AI features that are coming out.

10:43 So we’ve talked already about the data agents. We’ve talked about the copilot standalone. data mark came out with two great articles just around using natural interaction and actually doing that in semantic modeling. So where he actually has a notebook that or that actually can push the cinnamon synonyms excuse me cinnamon synonyms. Yes. that has always been one of the worst words for me to pronounce. and actually

11:14 Pushing those to a data agent. So to me, I think we’re this co where copilot’s going to be great is we going to need the features available in data agents. I think for this to be hyper effective. Yeah. I mean it I mean think about what where I mean the more they apply to this the it’s it’s a double-edged sword, right? because the more you send to the agent, the more information you have to prepare and the more tokens you’re going to use, the the more expensive it

11:46 Would become, right? So, we’re playing a we’re playing a bit of a balancing game here, I feel like, at some degree where the more I enrich the the prompts that I’m going to give to the co-pilot, the better the answers will be. But on the other hand, it’s also going to cost me more money. This is not like a $20 subscription to an AI. this is every single time I send more tokens into the prompt, it’s going to cost me more money because it’s going to take more effort for the the prompt to to do its thing. So, not not quite sure how that’s all going to pan out at the end of the day. but I I do think to your point, Tommy, right, the semantic model is the

12:17 Definition for all the different properties, the various like everything you need to do is kind of now kind of centralized around the semantic model. And when we look at like other tools like snowflake or I guess the other one maybe be yeah snowflake they have the data warehouse thing. So Snowflake and data bricks those are other two systems for data warehousing and both of them just recently announced they’re both building semantic models they’re building semantic model definitions to

12:48 Things right so they’re they’re adopting what Microsoft has already had for like 10 15 years here with the semantic model and using this technology to store the metadata like just have just having lakehouse tables doesn’t really store the relationships doesn’t really store all the descriptions and other semantic level things on the on top of so that’s why they have unity catalog that’s why these other cataloging experiences these metadata things are being generated on top of the data that’s already there because you

13:18 Need more enrichment for the AI to understand the context of what you’re asking for yeah and the the limitation that we have right now with the semantic model is that prep data is you think about what it’s actually doing in the back end it’s trying to generate code to in a sense do a query to the model well the problem is it’s probably what language is that being generated in? That’s where I really like data agents because you can actually translate a SQL statement or a KQL statement or even a DAX statement and use that all in the same data agent because I want to query

13:48 My table in a certain way. Well, DAX is I as great as DAX is, I don’t know if it’s the best language to do quick queries or going to work the best for how Copilot works. Usually things like I don’t know. So that’s where we’re still at. but there’s a lot of features to kind of get the the phrasing the the way that we actually word things correct. I’m still looking more for how we actually weight things. But it’s it’s going to be interesting the ability and

14:20 I think where I am focused on is how well does a tabular model work for AI. when we look at all the different AI tooling out there in the world right now and what it uses for its back end, how well is that going to translate over to a table model definition or semantic model? how so that’s that’s that is currently where my focus is right now. Interesting. Yeah, I think it’s going to be a lot of experimentation and figuring

14:51 Out what works. I’m looking forward to the community bringing up whatever seems to be good there as well. Yeah. Yeah. Interesting topic. What other what other topics do you have for us, Tommy? honestly, those were the big ones was just prepping data for AI and just the two great articles on kind of feeding things to a data agent and then just kind of going over the prep data ecosystem right now and fabric. So, I I’ll make sure I include the links for these things. There’s two comments here from data mark. he’s got a couple posts here that the natural language isn’t magic. and then,

15:23 Automatically populating data agents with the semantic model synonyms. data agents is quite interesting to me. I’m not quite sure where we’re going to go with data agents. I think that’s one that has a lot of potential for me. Taking a semantic model, adjusting it, adding some extra stuff around that, prepping your model a bit more for AI, adding additional context. That seems very interesting to me. I I do think there’s a trend here. The trend is when you use AI against your data models, the AI is being used to generate SQL code and once you have the

15:55 Output of that, you’re then building visuals and tables and talking about the data that’s being returned. I think we have very quickly found that the agents are not good with numbers or AI is not really good with numbers. So asking a specific numberbased question from the model really isn’t going to get you the answer you want. So, I think we’re starting to already see a shift in how AI is being used. AI is really good to write code. So, we’re going to use the AI to write SQL and the SQL then will

16:25 Return the answers that we care about from the model or that to me that seems like what we’re going to be doing here a lot more thousand%. awesome man. So, I don’t know if you but I’m I cannot wait to dive into this mailbag today. All right, let’s do the mailbag then. So, jumping in today, our our main topic here will be PowerBI without fabric. navigating PowerBI in small business and and things that we can do there. Tommy, go ahead and read us the mailback here. I think this is a great question to ask. Alrighty. So, here we go. I enjoy your podcast, but so do I. I not I’m a fan.

16:58 It it is very medium to large enterprise focus. So, some good feedback there. I work I work a lot with smaller businesses, eg credit unions. In listening to episode 412, your justifications are great if you have the infrastructure to support fabric. For example, I have a SQL server for reporting usually built for SSRS for business. It

17:29 May be a cloud or not. They don’t want to pay for fabric. So my only two options disagree here are sending out the PBIX file or posting to SharePoint. The business does want to invest in PowerBI Pro premium licenses for the whole company to view interactive reports nor P PowerBI which has its own set of vulnerabilities.

18:00 They already spend six figures a year already just to get data snapshots out of proprietary systems once a day. Am I missing something? Yes, I’m referencing banking, but it could be any small to medium-sized business that needs reporting. They need to be able to justify how it gains them on something more than what they get now. Yeah, this is a good question. All right. A lot of lot of thoughts that kind of come to

18:30 Mind on this one. one I will also argue the challenge I think becomes a little bit more I think you get a bit slightly steeper of a challenge here because we’re also in SS we’re in we’re in SSRS land but we’re also in the banking industry and I my understanding of when I’ve worked with banking andor financials there’s a much harder line between like the data engineering and the SQL DBAs versus like the business and what you can go access inside the business of things. though there’s there’s a much stronger delineation

19:01 Between the two groups. and users of PowerBI in the financial space get a lot less flexibility to build what they want to build on top of SQL servers because it just there’s just a lot more I mean it’s it’s literally legal legislation. Yeah. That requires you to have height control who’s accessing what at what time. And there’s a whole bunch of other extra things here that fabric somewhat gives you but doesn’t

19:31 Always give you all the pieces of this. So yeah, this is a great question. Tommy, where would you take this one? What do you think your what are your initial thoughts or initial reactions? So first I I do appreciate the users feedback on I think a lot of our conversations are if you had the whole shebang. you’re already you’re using fabric and you have access to all the features, but I think there’s still a lot of people like this person here where man, we’re just trying to get out the door or just even have a door open to start using fabric. I to your

20:02 Point even with a baking sector, they have a ton of systems too and just the integration itself is a pain where usually a lot of the investment is made. But I I I want to for me I kind of want to focus here on that last sentence that I said where they need to be able to justify how it gains them something more than what they get now. Yeah. And I think this is the big thing for a lot of organizations. So this is probably where again when I listen to this question here, this is probably where I’m trying to attack this

20:32 Problem as well. I would almost argue like what is the SQL server reporting doing and does it have to be on prem and does it have to be in the cloud. So you usually when I start seeing organizations starting to shift more towards a PowerBI or more towards fabric it’s when they have an opportunity where they need to adjust how they are storing their data today like the SQL servers that are onrem they’re getting too old they’re getting unsupported and now there’s like a cost initiative of we’re going to now evaluate how many dollars

21:03 We need to throw at a new SQL machine a bigger SQL machine we want to store like what are the things you’re trying to do on your current SQL things on prem that you can’t do because like for example in financial sector world you’re going to do a lot of snapshots is there something there that you want to do in snapshots that you need to do where you’re not you can’t do the reporting the way you wanted to because the SQL server just isn’t large enough we don’t have enough storage space there so I see it as an opportunity when we get to that moment

21:34 Where things are too old we’re getting deprecated we got to update we got to move things to the cloud That’s when I’m thinking about okay let’s revisit the cost comparison between what would a fabric SQL server look like versus a Azure SQL server and even to that degree if you already have the SQL server inside Azure is there a reason why you may want to move into fabric the SQL servers in my opinion run basically the same way there may be some slight adjustments there and then the

22:04 Other the other kind of key point that I would actually argue you here as well is a lot of your data is coming from SQL servers, but that doesn’t necessarily mean the SQL server is only the place where you’re getting the data from. My Yeah, I would argue point there’s probably other things you’re going to want to go get data from. And if you’re spending so much money on getting data out of current proprietary systems, Mhm. maybe maybe we start looking at like maybe we should be landing data in a

22:36 Lakehouse. Maybe we should be using a lakehouse with file folders to actually. So again, is there a cost reduction there as well? So usually I’m trying to attach fabric to like look there’s a cost savings, there’s a speed to deliver the value of the data and how can I apply or basically make it a wash, right? what I would have spent in net new development for this new feature or more data or more storage. I could then shift that expense over to fabric and then adjust what we’re doing. that this this is the point that’s I I very

23:07 Interesting the way you put that because I almost would argue the opposite here where I think a lot of the I think that hesitation is a lot of organizations especially midsize they have more systems than they have resources for and they’re probably already doing some work in some to this point ar some archaic systems so why do I want to to add another step to that feature and okay, now I’m going to add

23:37 It to a lakehouse. Okay, now I’m going to add it to something another system in fabric. where to me they’re already having this huge investment on the old systems. That means also the upskill too. Most middlesize organizations it’s hard for them to keep pace with all the technology changes both from the investment and also the skill right. So I that’s the challenge that you find with if I’m going to why would I just all a sudden just jump to

24:07 Fabric? No one in my team has any of the experience. Yeah, they’re all SSRS or and old systems, but if I’m going to use fabric, then that means that team has to entirely become experts in this this new environment. And that’s a hard sell. That’s a really hard sell. It’s not experts. No, I disagree with you in that statement. You always think everyone has to be an expert before they use everything. That’s not the truth. That’s not true. That’s not true. What you’re saying what my whole my whole point here was I’m talking about

24:37 Shifting costs, not not absorbing new costs to do this. What are we doing today that’s inefficient? Is the SQL server old? Are we running on a VM? Is it a is it a managed instance of machines? There there’s potentially an opportunity to move things out of your old world into a new place where you’re saving yourself money. So what I’m looking for is I’m looking for what can how can I one provide more capability to my organization but then two remove some things from my existing

25:07 System. I’m not going to try and stand up some two of the same things and I do I would agree with your statement around like look I may be introducing the lakehouse for something but that the only reason that is there is because it’s it’s potentially a more efficient way to get data from proprietary systems save it but with more volume of data right so the reason we had SQL server is we’d bring we’d connect to this the proprietary system we would load what

25:38 Data we could and we wouldn’t keep all the data. We’d only keep the bare minimum or we reagregate or we would do some very heavy computing things on SQL to make sure that the the SQL database stays very efficient and stores only what we care about slowly changing dimensions all these things. The the cheapness of storing things in the lake is changes it for me. I mean that that frankly changes the ability for me to store snapshots of databases every single day. the whole thing at

26:08 Pennies on the dollar of what it would used to cost me to build a SQL server. So again, that’s where I mean it takes it takes some experts to kind of show up and be able to point out where the problems are. I’d also have to really echo and while you were talking there too, Tommy, I think the the evaluation of what are your users actually using from your ecosystem, like what are they doing? And one of the things that stands out to me here is this comment around the SSRS reports. So

26:39 In this question, if if all we’re doing here is a SQL server that’s spitting out a bunch of SSRS reports that are getting emailed to people and that’s the end of the report. Like that is our reporting. I might I might actually argue don’t migrate. Don’t stay where you are in the current, data loading SQL server space because yes, you can get SSRS reports out of PowerBI. It’s called pageion reports. So we we have the same

27:11 Effect, but I find much more value coming from a semantic model and then having the SSRS reports pull from the semantic model because then role level security works. You still get the SSRS reporting. you can then apply a schedule, right? So people can get a scheduled emailed something of those reports. Great. We’re doing that. But again, if you’re going to take all this effort to move from SQL and SS SSRS reporting, spend 2 months migrating

27:41 Everything to fabric just to do the same thing again with sending out pages of emails with SSRS reports. I I’d argue like that’s not enough of a value ad unless you’re saving money or providing some new capability. So to me like that’s not that’s not the reason you want to go to fabric. That’s not a reason to to transition there. And it’s funny you say that too because I I cannot tell you how many initial trainings I was going to do on page native reporting completely shifted to just PowerBI report building because the the idea of what they thought it was

28:12 Like oh it’s just the updated SSR like no no what you actually want to do is going to be achieved in PowerBI desktop man. And but here’s the thing and you said that everyone doesn’t need to be experts, right? But if right now you have an environment where people are building SSRS well if they were going to switch to PowerBI reporting and for and even right now forget the licensing issue here. Well again that’s a business where

28:44 Their cost on for data and for BI is probably limited. that’s not a huge investment in terms of or going above and beyond. Well, someone’s got to own that. Someone’s got to run with that. And if their only experience is building SSRS, how quickly does that translate over? So, but I think the bigger thing here is too where it’s also about how people are viewing these reports right now and and I think that’s the where I would think value add here is

29:15 I am try is dealing with the snapshot issue. It’s funny the book on Jamie Diamond that I was reading and he was talking about talked about how his number one investment when he got to Chase the number one thing he focused on was systems integration and it was about getting all their systems to be able to talk because there had like 490 different things just so they can actually have a single source but his reports were not anything on the cloud

29:45 But it’s still you Excel files it was still rows and rows and rows of data where for a lot and again this is part of the data culture thing too but I I can wow you with a bar chart I can wow you with these things but it’s also about this point too shifting everyone over to what a report can do compared to what we’re used to and what we know on any technology I’ll stop there so yeah you’re seeing a lot of

30:15 Things there so a couple questions that kind of come up in my mind as you were stating those things one was you mentioned PowerBI and I think your comment or the very first comment you made was hey I was going to do a bunch of SSRS reporting but then I turned them over to kind of doing PowerBI. I think it’s a great statement that you made, but I also would argue, especially for this question here, right? If you already have a large user base using the SSRS reports today, I’m working with a couple clients right now, they all their

30:47 Customers want is a bunch of SSRS. Period. End of story. Like there’s there’s no question about it. Like if we can’t support that, there’s no use in doing anything fancy, moving anything forward. Period. Right? So my question there is like even if we like what percentage of your SS SRS blah blah blah I can’t say anything what percentage of your SSRS reports could you actually replace with PowerBI reporting and then once you say 10% 30% 40% I’m not quite sure what that would look like and so

31:19 Okay fine if if we can only replace 30% of our reports with PowerBI reports and get rid of like physically turn them off no longer render them is what I’m trying to communicate here is that’s what we’re saying then then is that worth the effort to go in and do the PowerBI semantic modeling what other capabilities are you giving yourselves do you want to start just deploying and using apps in a way that’s going to do this I mean at the end of the day you’re going to need to start figuring out what’s the right blend of this information so again I’m kind of reading

31:50 Between the lines here a little bit in this question they’re doing SSRS and they’re dumping PBX files into SharePoint. Yeah, that hurts me. Like, so the one thing I would point out here for small business, small businesses should not be selling around PowerBI reports as PBX files into SharePoint sites. That’s that’s the same thing of just doing Excel files and just putting data in an Excel file and then saving the Excel file to SharePoint. I

32:21 Mean, that’s you might as well just do it there, right? There’s no reason to do anything fancy in fabric at that point. make a make a semantic model connect analyze in Excel do build the tables you need to build and just save the files to SharePoint boom done right only the people who can who have access to the semantic models can refresh and then game over right so if you’re if you’re doing that kind of stuff then I don’t I’m not sure you need to go full-on fabric but if you if you feel like you can transition your users away from getting emailed reports using SSRS and they do want this interactive report

32:53 Piece or you have the ability to make an interactive report that answers the same questions as the page reports then I think I think we’re talking I think we’re I think we’re then actually ready to start exploring what can you get from PowerBI experiences but that’s so that’s just one thing another comment here around the licensing piece so aside from the whole everything users and all this stuff I would also echo depending on how big your business

33:24 Is you’re already spending a lot of money on things for Microsoft Fabric. I also would double down and say do my users need a bit more security around their licenses that they have for the the software, right? So, the reason I’m bringing this up is if you have a Microsoft E5 license, you automatically get PowerBI.com, a pro license for free. And that opens up a lot of doors here for you. When

33:55 You’re in these very regulated sectors, sometimes you can kill other programs like you might be using Google or Google Sheets or some other things. If you can get rid of those costs of these other applications and say, “Look, we’re just not going to use them and instead we’re going to centralize all of our costs down to our users to get them an E5 license.” Boom. Like we’re there. Like if every if you did that that feature alone you get better security you get better security per laptops you have more control around those users there’s

34:26 A security story here and usually businesses are very willing especially in in the financial sector you can kind of turn that story very quickly to let’s add additional security for these users by bumping them up to an E5. So again we’re trying to we’re talking about shifting costs here as well. So, what software can I get rid of where we can give users E5 licenses and then everyone gets pro licenses and then we’re we’re off to the races. We can do everything we want with PowerBI and pageionate reports inside PowerBI.com. So, that

34:57 That would work there by itself. and you don’t even need fabric for that. Yeah. And honestly, I I’m always lost when people say the licensing issue just strictly PowerBI, especially when they’re talking about they’re in that SSRS or legacy environment. And and my selling point is not the rich interactive reports, but if you’ve ever built an SSRS report, they are every single one is its own definition and its own in a sense query. It’s not off a single model. Yeah. It’s a black table.

35:30 Exactly. Exactly. White. So every time you need to create a new one and maybe it’s built off it, it’s its own definition and you can make easily the argument to me if you want to talk about saving cost all the investment in time you’re currently spending to make updates to any of your existing SSRS reports or build a new one which each one of those again is it’s a start from scratch query querying individual tables without any structure definitions between them

36:00 And yeah, yes and nos, Tommy. Like I’m I’m going to say yes in the fact that if you’re going if you’re going from like a current SQL server where you’re doing that, I think you’re right. Like you’re kind of starting from scratch every single time. But I would argue like when I go into like the pagion report experience, like if I’m if I’m using that now now I have a little bit more it’s a little bit less friction for me, right? Because the semantic model already exists with tables and relationships and calculations and I’m able to reuse that. So me personally, if I’m going straight SSRS reports from a

36:31 SQL server, it’s you’re right, it’s laborious. It’s kind of a pain. There’s not a lot of help there, right? If I’m trying to build parameters in the report, it’s it’s kind of difficult to build it straight in. Like you got to know what you’re doing. Oh yeah. But I would argue when you have the semantic model backing it and then you’re adding parameters, that’s where I’m going with it. It’s a lot less friction. So I agree with you here. Yeah. No, no, no. You’re this is exactly where I’m going with this. To me, the selling point of just that investment in just PowerBI is what the semantic model provides because if

37:03 Again you have 400 SSRS reports, those are all 400 individual queries that if you need to update one definition, you got to go through all of them in terms of a calculation or parameter. with PowerBI just the idea of the semant the semantic model alone that source of truth is reus can be reused to your point in a pageionated report and other reports we already have that definition and this that’s where I’m always lost on the ideas like well we have ssrs we don’t want to pay for the rep just

37:35 The pretty reports like the honestly the rich reports the interactive feature that’s just the cherry on top of what powerbi really is and I I think the misconception is they’re thinking PowerBI is a reporting based tool. Yeah. And that is where for me that’s the trade-off. You want to not have the licensing, you don’t want to deal with, not being more in the service. Fine. Well, you’re still dealing with you’re way in the heavy code area. you

38:06 Have no and again reusable structure or semantic definitions and with PowerBI it’s that easier deployment. So yeah, you can easily make the case I can save a ton of money just by having 20 SS SSRS reports going from a single model and and for me that’s where the big thing is here with a lot of organizations they don’t understand and maybe this is what the guy or the

38:37 Person was missing where the conception is if we’re going to go to PowerBI it’s because we want reports that look better than SSRS SSRS And I think that’s missing the point or really missing the u what PowerBI actually provides. I don’t I don’t read the question that way. I don’t read the question as a we’re looking for a better experience. It it sounds to me like this just the way I read this question is there’s already a SQL server reporting some onrem some in cloud there’s usually

39:07 An SSRS report on top of it. So like that’s kind of like the word that they’re using. Yeah. And so they’re trying they’re trying to move forward with like how do we share the PBX files that we’re trying to produce. So it sounds like we’re starting to like just on the edge of we’re going to begin to use PowerBI, but the only way we can do it right now is sharing PBX files directly to everyone who needs to have access to data. And that’s not I really don’t love that whole PBX sharing of files. The service is really where I

39:38 Think things shine. So to your point there Tommy though around this I really want to focus on almost taking like a ven diagram or or looking at like so let’s take every person in my organization and let’s just kind of look at them and say okay of all the users in here which users need the certain certain kinds of experiences and what I mean by that is let’s think let’s talk about experiences with I’m a user and these users like we have let’s let’s just call it a hundred users

40:09 Right? 50 of them just need view only access to SSRS reports. Okay, great. We think another, 10 of them would want to create their own SSRS reports. So, do we have that experience today for them? How do we handle that request or that requirement? okay. Now, what of what percentage of our users now are also needing to create reports like build a report on top of an existing semantic model? How many users do we think would actually use that

40:39 Feature? And could we go from like a small number 10 or 12 users up to like 20 or 30 users? Would that add value? Like do our does our audience as our company understand that capability? And then going even further, okay, how many users do we think would need to create their own data? How many would pull down data from different data sources and then actually need desktop and actually do imports and and other things? like are we giving out are we actually giving out SQL server access to PowerBI report builders so they could go get our

41:11 Existing tables and then use that to build their own semantic models. And then the last kind of audience I would look at here is which number of users in our business would want to write their own SQL statements and use that to build their own tables of data. not necessarily SSRS, but do we have users in our company that are actually writing SQL that we can use to go build things? And the reason I ask these questions is because I worked in a company where there’s a couple thousand people, but

41:42 They had over 12,000 access databases. And okay, yes, you can throw up if you want. Fine. Wolf is right. But the reason I bring this up is because one of the requirements for their company was that everyone who came into the company always had the need or the opportunity to go get a SQL course. Everyone was run through a rigorous you will learn SQL. We write SQL in this company. This is how we access our data. So they had a

42:13 Culture in their company that required everyone to get to a certain level where they could all write SQL everyone in the company. And that’s why we had so many access databases, right? We’re doing access because the big data warehouse couldn’t handle all the queries for Excel. So we made access in the middle to write the SQL queries, pair down, aggregate the data that we could then connect directly to Excel and then we could go make reports and build things like it was I understand the pattern. It made a lot of sense. They went from big

42:44 Databases to medium-sized databases into PowerBI because they had to agg or or basically Excel because they had to aggregate data up fine. that that is the pattern. But walking into that culture, we could have easily turned on fabric. We could have easily added, lakehouses, warehouses, and then just virtualized a bunch of data for people that they could get access to the full big table and then they could build their own queries and their own analysis. So if that’s if that’s a cultural thing, if that’s what’s in your organization already, you can use that

43:15 With fabric. And so going back here to this question, right, credit unions, small businesses, there’s a price point for fabric that would make sense to you. There there is there is I mean, you can’t tell me that your business is so small that you couldn’t afford an F2, 160, res reser reservation for a license for fabric for a year. I mean, my business, if your if your if your business can’t make an additional $150,

43:46 $160 a month for like that small level of like, can we try this out? Does it add value? Does it save time somewhere? then then I you shouldn’t be even touching PowerBI at all like you should just kind of walk away from the whole thing because I think 150 that’s 15 users not even that’s 10 users at like a little bit more like 16 users at a PowerBI pro level and you get the full fabric that means you could add free users you can add PowerBI embedding

44:16 There’s accelerator solutions around that as well like there’s options to get the price per user down to a certain point and I would also argue Here’s another question you want to ask yourself. Sorry, I’m making lots of points here. I apologize. This is my last point. I promise. Think about what’s the price points of what you need for users to go do in your application. Right? We have PowerBI Pro users. We have premium per users. So, we have 14 and $24 for those two licenses. If you say that Fabric’s too expensive, tell me what is the right price point

44:46 For a free viewer of those reports? Is there is there a price point that you would even care about? Like would it be a dollar per user, $3 per user, $4 per user? What’s the right threshold of spend on giving those users access to either SSRS or PowerB Reports? Because then I think you can have a better conversation around building your own embedded stuff or going to look off the looking out the store, go find if there’s an offtheshelf solution that would help you get to that answer a lot

45:16 Faster. Anyways, a lot of considerations. I’m sorry there. I I do want to go I want to go back to your first point because I think this was I want to flip it on its head a little because your example about the organization where all those access databases was because everyone was coursed and skilled up in SQL to use the access database so more access databases are being created etc. To me, I think that’s a little you got to flip that on its head when you look at PowerBI

45:46 Because you talked about the different users you wanted to look at who needs to do SQL queries, who needs to build their own reports. Well, with PowerBI, this the core experts so to speak, the ones kind of building the foundation, it’s a small number just like it is probably on an organization right now with SSRS, the hyperskilled people who are building those SSRS reports. Well, the problem is they are the only keepers of that and if anyone wants to make an edit or do anything else again they have to have a certain some skill or

46:18 Experience in that coding or language. But with PowerBI, it’s different. And I’m I don’t think there’s a need there either, though. Like, well, let me get to the point with PowerBI here because Okay. Yep. To me, it’s the technology of what PowerBI does changes that argument because now I don’t need people to even worry about what skill that they have is if they can connect to a semantic model, right? Where all do they know how to drag and drop? we already had the definition created. So the amount of people can

46:50 Actually interact with that goes up exponentially with PowerBI compared to let’s say you’re in an environment of access databases or something like that where it’s more kind of in a sense core skill. PowerBI allows a few people to create the models that can be reused and recreated across the ecosystem without a ton with skill at all. you don’t need to know about the tagler model. you can easily connect to the same model that again those core

47:20 Small group of people who are doing all the development in general. So I would yeah I would I would agree with you in that regard. So I think I mean the semantic model is key here right? So right now the question really is aimed towards look we’re trying to share PBX files. Well you actually don’t want to share those. That’s not that’s not the intent. The PBX file is a semantic model and report all bundled into one piece. So what we’re really trying to share here I think is what we actually want to share is we want to share the not

47:51 Semantic models we want to share the sorry we don’t want to share the reports themselves we actually want to start sharing the semantic models the semantic model share is what I think we care about sharing so to me I’m looking this going at the question here what are we trying to share are you trying to share a semantic model or you just trying to give out reports to people so if you’re trying to replace SSR SSRS with just pure PowerBI reports, not the semantic model I want people to build.

48:22 Now the question I think becomes how much self-service do you want to have to do here, right? Do you let people build their own SSRS reports today on top of servers? If the answer is yes, then I that so small I bet. Well, I I don’t I can’t guarantee you one way or the other on that one. I again I would have a harder time reading this one. My my assumption here, Tommy, is what’s probably happening is you’re right. The SSRS reporting thing is being built by a central team. There’s a system in place that says any user who has a problem or

48:54 A question, you submit a question here. It gets into a queue. There’s a there’s a team of people who pick up the queue, adjust the report, probably make a copy, make a new report, and spit it out. Right? That’s that’s what they’re doing. That’s the workflow. So, fine, it is what it is. But I’m looking at it going, is that the most efficient way? And what you’re introducing here, Tommy, is you’re looking at it and saying, look, I’m going to build the relationships, the definitions, the semantic model, and we’re going to push that to users, and we’re going to say, hey, users, now you

49:24 Can go in and explore and create your own reports, your own tables of data. And this is what I’ve seen this a lot coming from SSRS organizations going into PowerBI. What typically happens is everyone’s like, “Wow, this is amazing. We can make our own reports much easier.” And what everyone starts doing is they make a whole bunch of pages and a lot of reports that are all tables. Not what a PowerBI report is meant to be doing. So people are not even like the data culture is not even in a place

49:54 Where we can start aggregating and pro providing visuals and trends of data over time. We’re still just trying to focus on we need people we we’re still expecting data to be emailed to us and that’s just I I think that’s the wrong mindset. I think I think that’s but it’s very difficult to change an entire organization. You have to have kind of some strong leadership at the top to help you move away from we’re not just going to focus on emailing reports anymore. We’re going to shift towards self-service. That’s what you’re doing.

50:24 And I think that culture really sets in because you think about those environments where users can’t touch the SQL tables, right? They have to be they’re probably on a VPN or network and there’s all these things. I can’t touch the SQL server. I need a this user thing. So, they kind of have to expect things to be just handed out to them because they have no control over that. And to me, once you introduce the PowerBI model, well, now they’re not talking. They don’t need any access to

50:54 The SQL server itself because everything’s stored in power band the model with his own security and easy access. I don’t have to worry about networking. Again, the idea of alleviating this this team and again most organizations that are small to medium size do not have a robust BI and data team. Yeah. So the thing is if now we can alleviate so much of their effort around modifications and building because again in those systems the only

51:26 People can do it are the ones with the access to tables and that skill that in that particular language or whatever tooling they’re using. Yeah. But now and to your point about also the sharing the PBX file. Well, yeah, that you can see your data, but you’re missing the point of that one source, one source of truth. I that and how far a single model definition goes. I don’t need everyone building the reports. And honestly, if people start putting tables in PowerBI,

51:57 But they’re using it, great. if that’s a start, it’s a start because now they they are at least in a sense being able to now interact with data they never wouldn’t be able to do so before. Everything before was a snapshot or an export or in the exact way that SS report was built. Now I can tell someone even if it’s just for them, hey, you want to see the that report and you want to see the customers by whatever sales region? Well, just

52:28 Connect to the model, put those two columns in there, and you can save that. You don’t have to build that for other people, but now you can get that quick view. That definition of the relationships and the pre-built metrics goes such a long way to really enabling users. Again, not even talking manage self-service here, just I want to see those three columns and I would like to see a certain way. I can do that. I have that ability and it’s still coming from the same source. Yeah, this your comment there also

53:00 Brings up another point that I’m thinking here too. In SSRS reports, when you’re building these custom queries that go directly back to the SE semantic model and you’re over and over again getting the same data, you are basically locking in business logic into the single semantic report that if you don’t repeat that logic for the second, SSRS report, right? you you you’ve locked in the business logic into one SS SSRS report custom aggregations custom queries custom like so you’re writing custom SQL to build a wide table of data and unless

53:30 You’re pushing that up to a view that lives on the SQL server that way when the report runs the views run and then you can so again I don’t know the data culture what’s going on here if we’re talking small business there may be a variety of options that are happening here there may not be the rigor that we’re we’re focusing on and all we’re doing with the semantic model is we’re centralizing izing more of that business logic together into the same thing. So, I think it’s another really valid point there. What I what I do want to kind of maybe like end on my final

54:01 Thought here is I just want to think about, how how do you want to use this new world for yourself, right? Are I think the price point of fabric going all the way down to 150 to maybe 260 bucks per month if you’re not on the reservation plans. I mean, I feel like that’s a very reasonable price to get started with some of this. And if nothing else, it’s a reasonable price to at least get the understanding of is there something we can do more efficiently with this new world? And I’m

54:31 Also thinking about your your user audience. Are there requests that you’re trying to handle today that your SSS SSRS reporting does not handle right now? My my inclination is saying your users are asking for things that you can’t do. That’s why you’re going with PowerBI to begin with. That’s why you’re starting to share things around with PowerBI starting. So I think the the answer here is are you able to make the business justification that says by

55:01 Doing the PowerBI things or even using fabric at these lower-end SKs are we solving enough of these hard problems are we able to take a solution that would have taken us months to develop and can we build it in now days or weeks I’ve talked to Alex Powers about this as well with Power Query. He said he’s seen organizations that have had some very complex data engineering and logic go from potentially weeks of development time down to a couple days. And so you can turn things around faster and you

55:31 Can get data into the the lakehouse into the fabric world in a very quick turnaround time. And I think there’s also something to say for being nimble like that. Is there is there a need in your organization to be fast, nimble and quick to de develop those? And that’s what fabric is doing. fabric is giving you this capability of being more data savvy. I think I think you also can move your data company to more of a culture of self-service and what I mean by self-service is building their own reports, making their own visuals, teaching them that capability. And if

56:01 You do that, if that’s what if that’s the direction the company’s going to go, I think the price will will easily be able to sit and find the right price point for your users. Anyways, that’s what I would I would argue there. I I kind of disagree with this statement. Yes, maybe we talk a lot about enterprise organizations or bigger organizations using fabric, but I work with a lot of small and medium-sized businesses that just use fabric in its form. and they just pay for cheaper SKs and they’re not a huge mega

56:31 Business, but they still find a lot of value. I mean, my business is a small business by far. I use PowerBI and I even pay for my own fabric F2 skew to run my reports on because I find it valuable enough to centralize and make it flexible for people to do their own reporting. So even already like as a small business that doesn’t even do a lot of reporting internally, I find it valuable for me. So I I think there’s the right price point. I think you have to really evaluate where is the value coming for your business and then justifying that value through the price

57:01 Point for you. No, I think you want to know when it’s what you’re missing here is if you have 15 or more SS RS reports, you’re already spending more time on that than it would take to to do what you can do in PowerBI because I think that the biggest takeaway for me is what you just talked about is what you a lot of people don’t understand is these legacy things to try to keep upkeep on similar definition hidden cost baby. So that’s right where

57:33 You want to you want to alleviate time and the cost of every time pulling your data and I have to now update 15 tables every single time because one change well I have one model one source of truth there’s where to me everything changes in the conversation but interesting to see as we keep going farther into the fabric world we still have companies like this who are still very bent on theirs rest and I I can’t wait to see more of the bridge, but

58:03 Loved the conversation today. Yeah, I’m going to I’m going to encourage you here. It’s just, keep up the good good fight. Keep thinking about what you want to be doing. And with that, we’re going to wrap. Thank you all very much for listening. We appreciate your hours of time listening to our podcast here today. If you found this one, please share with somebody else. Tommy, where else can you find the podcast? You can find us on Apple, Spotify, or wherever you get your podcast. Make sure to subscribe and leave a rating. It helps us out a ton. and please share with a friend since we do this for free. Do you

58:33 Have a question, idea, or topic that you want us to talk about in a future episode? Head over to powerbi.tipsodcast. Leave your name and a great question. And finally, join us live every Tuesday and Thursday, 7:30 a.m. Central, and join the conversation on all of PowerBI. Tips social media channels. Thank you all so much, and we’ll see you next time. Heat. Heat. N. Get out.

59:16 [Music]

Thank You

Want to catch us live? Join every Tuesday and Thursday at 7:30 AM Central on YouTube and LinkedIn.

Got a question? Head to powerbi.tips/podcast and submit your topic ideas.

Listen on Spotify, Apple Podcasts, or wherever you get your podcasts.