The Root of All Problems – Ep. 235

Most analytics problems don’t announce themselves as “the real issue.” They show up as slow reports, disputed numbers, and teams burning hours on fixes that feel productive but don’t improve the system.

In Episode 235, Mike, Tommy, and Seth lay out a root-cause way of thinking you can apply to BI work (people + process + data), then tie it back to Microsoft Fabric realities like Direct Lake performance behavior, capacity pressure, and why monitoring matters.

News & Announcements

• Submit a topic idea — Share what you want the team to debate next (a pattern, a product change, or a thorny real-world problem).

• Subscribe to the Explicit Measures podcast — Browse the full episode archive and grab links for Spotify/Apple.

• Tips+ Theme Generator — Generate a consistent Power BI theme quickly so your reports stay visually coherent as they scale.

Main Discussion

Two themes weave together in this episode: how performance really behaves in Fabric, and how to stop treating symptoms as if they’re root causes.

They start with Chris Webb’s performance testing work, which highlights the idea of data hotness — Power BI paging just what it needs into memory — and a key gotcha: even with a Direct Lake semantic model, some scenarios can push a query into a DirectQuery fallback path (especially under capacity pressure or large-volume asks).

From there, the conversation broadens into a system-thinking framework (Theory of Constraints + cause/effect mapping) for diagnosing analytics problems. The punchline is simple: if you don’t find the constraint (the weakest link), you’ll keep shipping local improvements that don’t improve outcomes.

Key takeaways:

• Direct Lake performance depends heavily on what’s already “hot” in memory; the same report can feel radically different depending on cache state and access patterns.
• Make “safe by default” report experiences: sensible landing filters, constrained date ranges, and guided selections reduce the odds of users accidentally triggering massive scans.
• Expect capacity pressure to change behavior. If queries can fall back to DirectQuery, you need monitoring and a playbook so the first signal isn’t an executive saying “it’s slow again.”
• When troubleshooting, optimize the constraint first — strengthening other links won’t help if the weakest link stays weak.
• Avoid local optima (isolated tweaks) when the real bottleneck is upstream (model design, data layout, governance, or process).
• Treat most pain points as undesirable effects; map cause/effect chains instead of guessing, and you’ll surface fewer (but more actionable) root causes.
• Define the goal of the system before you optimize. Otherwise you can build very efficient pipelines and still fail to deliver the outcome the business actually wants.

Looking Forward

This week, pick one recurring BI issue and build a quick cause/effect chain until you can name the constraint — then fix the constraint, not the symptom.

Episode Transcript

0:35 podcast with Tommy Seth and Mike hello everyone oh yeah it’s been a while good morning did a couple recorded episodes it feels like it’s been forever since we’ve talked again again it’s gonna be even longer too because I know we have more vacations it is good to see you guys I have missed you I was in Milwaukee guys I’ve seen the light so I just got back from vacation Milwaukee’s really nice right I’ve heard one Chicago yeah well one with my family people from New York and I have been

1:05 people from New York and I have been convincing them for two years to go to Wisconsin for vacation okay I had no idea why I didn’t know why I just know that everyone in Chicago does it and oh my goodness we are in Door County oh wonderful it’s nice places great area I’ve seen the light of the best beaches so I’m amazed have you been to Northern Wisconsin before never oh my gosh I’ve never best kept secret ever anywhere man seriously and the beaches on we were in like Jackson ports like white sand a

1:36 in like Jackson ports like white sand a little cold water I’m like this is better than Florida because it’s not as busy and then I also got the spotted cows or maybe 72 of them because so but it’s good to be back the anxiety set in as soon as I got back on what I miss miss well a happy Tuesday to you gentlemen and I’m glad we’re back live Tommy love love the fact that you explored the northern Wisconsin area especially up on the peninsula there it is they do have beautiful

2:06 there it is they do have beautiful beaches for about a month or two a year yeah right and then you just don’t want to hang out around the beach it gets cold probably or yeah or maybe you do and that’s why we have the polar bears right in the freezing water that’s true a great time this has been the year to go do that because it’s been very hot in our area and it’s been very dry so it’s been like nice warm weather good time to be at beaches and hang out in near an area where there’s water and stuff like that as well as well we are also going up north as well here

2:37 we are also going up north as well here in the next couple days so we’ll go up to Sheboygan for some family stuff and our kids are extremely excited they have found someone on the internet in YouTube Oh surprise something on YouTube that my kids want to watch and they found this gentleman who goes around and searches for things like treasures and has a metal detector yeah so the kids pulled their money together and purchased a metal detector off Amazon and we’re going to the beach and so they’re gonna go roaming around with a little tiny metal a little handheld metal detector and trying to find things so

3:07 and trying to find things so they’re getting like where’s where’s my snorkel where’s my goggles like all this stuff so they can go hopefully find a whole bunch of treasure in the in the ocean or in the lake I guess so we’ll see what happens they’re like we have to go where a lot of people are so they drop their wedding rings or Rings or whatever I’m like oh okay kids sure we’ll try this whatever you want us yeah so if I come back and I’m retired they found something good sweet anyways that’ll be super fun good fun times Explorations in the summer so it should be fun

3:37 it should be fun yeah yeah Power Bia Park says in the chat make sure to bury some small coins for them throw a handful of coins that’s a great idea yeah what I’ll do that you stay here let me make sure that the beach is okay first yes yes I will make sure that there is no I’m pretty sure they won’t watch the podcast so we’re pretty safe here we can talk about it it [Laughter]

4:07 exactly right I like it that’s cool excellent okay let’s talk about we an interesting opener today today we have an article that came from Chris Webb again his stuff is always very detailed and Incredibly informative I guess I’ll call it it’s just there’s always so much the amount of times I go back to Chris Webb’s blog and research and and review what he’s done he fits very well with our topic today

4:38 he fits very well with our topic today but today we’re going to talk about Chris’s topic around performance testing the power bi data Lake and data models inside fabric that’s what his his article is so again not it’s a bit of a mouthful a lot going on there but he starts really going in deep on the the capabilities of what Microsoft fabric does as it relates to Microsoft Fabric and I really there’s a couple things that point that stuck out to me in this article there’s this concept of what he’s calling data hotness the ability of being able to see if the

5:09 the ability of being able to see if the data is recently been read and there’s actually a scoring system inside that this might be one of his other articles which is a scoring system inside power bi that as a column gets read and if it gets read multiple times it basically gets some score and a really interesting note here when he starts talking about this concept of data hotness it says using the direct leak as a data set Power bi can page individual column segments dictionaries and join indexes into memory on demand

5:41 memory on demand which this is really cool I really like the what’s going on here like this is an amazing amazing feature I think like it’s literally picking out the individual columns it needs to run the pages or queries it’s not having to load the entire data model all at once which I think is really cool that’s a t-shirt that’s for sure really data hotness oh data hotness that’s right up your alley that’s right fabricator I’m all five levels of data hotness whatever he was talking about the article

6:11 the article leaving me a 10 out of 10 indeed hotness without it then did analysts agree something like that oh that’s great I can’t believe how Chris even put together the test itself much less running it what because like we’re going through fabric where would you start in terms of like what am I going to load first how am I going to evaluate it but it’s incredible this I’m more impressed and I think Blown Away with the structure in the in the sense the

6:42 structure in the in the sense the I guess is set up to go about it than anything else here yeah I I like how it is to be close to the product I guess that’s true that’s true but he’s also he’s doing stuff though like he’s going in here and he’s actually writing like Dax studio and and little little Snippets here to help you out yeah so overall his his explanation of things is he’s always been spot on Amazing in this this article is the same as all these others so like

7:12 as all these others so like foundationally it’s fantastic it’s a great read go go read these levels of like hotness right because they’re his whole point is they’re going to impact how like the performance that you’re going to get out of a directly connection in pure crisp fashion he tells you exactly how to go do this or start testing it through deck studio what I found particularly interesting though is there there is this use case where you

7:43 this use case where you you can fall back to yes right so I think the whole like there’s two parts of the article that stuck out to me one is the very final thing direct lake is still in preview right so everything he’s telling you to do is is in here like it is going to rapidly change because they’re gonna make improvements to this entire system it’s it’s like just be aware as you go along if you catch this in six months to a year it’s probably not going to be relevant as it is today however one of

8:16 relevant as it is today however one of the main sections he calls out is fallback to direct query and I just want to read through that one real quick because yeah I like it even with a direct Lake data set there is no guarantee that your query will be answered in direct like mode interesting in as yet not fully documented scenarios but basically if your data volumes are too large for the capacity you’re using power bi will switch to using direct query mode and now like that one catches me because

8:49 catches me because hopefully in this analysis or like I will have to have some monitoring to validate and make sure that whatever capacity I’m at like am I going to get a warning that all of a sudden I just flipped into direct query mode because like that having having issues like come up out of nowhere and Reporting this is this is a new one this one would be like another big check in the list of like why are my reports running slow yep and and if you’re you running slow yep and and if you’re using direct link in the future

9:19 know using direct link in the future like this would be one of those that like checks my top box as far as like did I just hit some threshold with my because using large data volumes right if you instantly go from like oh you’re out of capacity now you’re in direct query oops okay like can I how do I limit that how do I stay under that threshold or just yeah do I expand exponentially anyway that was the one that was a warning Bell for me and my future endeavors along the way here

9:49 along the way here that’s crazy that I switched at our query you never know both I need to know my point is I’ve got to figure out how I need how I I know find out like how do when it’s doing that switch over so I so what he does he does also note here that this is like for very large data sets so yeah depending on what fabric capacity you buy these things could be quite large in size but this is actually going to be more important to your points out there you need now more monitoring as when

10:20 you need now more monitoring as when this model runs or when this report is running running what is it trying what are the queries

10:25 what is it trying what are the queries doing on the report itself and how are those queries impacting the larger effort of the partitions the model The Columns that you have together so it’s it’s it’s tying together a more direct link between the model and what’s being rendered on the report page because if you have a a smart way of rendering that report for example you’re you’re pre-applying a filter you’re not giving everyone a large date range day one or when you open the report you can actually do some things to cut that out a little bit so that way users are forced to

10:56 bit so that way users are forced to select something to get larger volumes of data of data it’s I I just yeah my mind goes to it’d be nice to have an indicator when the report switches from this direct Lake to direct query mode and then have it at least let us know so we can go figure out which queries on the page are doing that I’m going to assume if we’re in desktop and we’re doing something like this with performance analyzer there’d be something in there that would start showing me some different statistics but

11:27 showing me some different statistics but again it might be a bit tricky as to when it actually turns on or not I don’t know know if you had a data refresh in the background I probably could impact it if you have a notebook running somewhere in fabric it’s taking away capacity in the course you might run into the issue right it could be something along those lines be interesting to see how this pans out on the plus side thinking conversely right if if I have some user that’s just like ah actually I somehow figure out how to like give me everything and then it just

11:58 like give me everything and then it just pops into direct query like as opposed to not like throwing out a memory and not having access to the data like okay well maybe maybe that is a benefit but yeah I don’t know either way to it’s an interesting one I’ll have to think through you bring up an interesting point there Seth because I think they’re at some level there is a behavior modification step here that’s occurring as well right if you try to ask for data in an inappropriate way I want it to take a little bit longer I don’t want you to do that right I want

12:28 want you to do that right I want you to like I’ll do it for you but yes it’s gonna cost you right there’s an expense to it so anyways all right let’s move so I think this also so Chris Webb does a great job in the article thank you Chris for the

12:58 the article thank you Chris for the article very well written definitely gives some insights here around what’s going on in this direct Lake versus direct query mode I also really thought here this dovetails very well with our article for today that we’re talking about is because Chris actually goes through a very detailed way of here’s my method for testing this this pattern he talks about preparation he talks about okay to test State one I did these things to test state two I did these things to test

13:30 state two I did these things to test state three I did these things and he progresses his way through as he continued to walk through his testing procedure on how this works well that brings us to our article for today which again being that I’m a mechanical engineer by trade this article really makes me happy it’s called data patterns well actually it’s the data patterns here but the article name is the root cause of all problems in Analytics very interesting article and starts

14:02 very interesting article and starts talking about initially some methods by which you can look at your data problems and start analyzing it down till the root cause let’s just start there we’ll maybe just give a quick introduction around the article Tom you want to give us some initial thought points some key elements here of what this article is starting to talk about yeah and who it comes from oh yeah let’s go who it comes from too yes I will do my best with the name so the the name of the author of this article is

14:32 the author of this article is first first [Music] [Music] and yeah like I said we’re gonna do our best here e-r-g-e-s-t burgers yeah sell a body maybe Excel that’s a great that’s a great great try at that Seth thank you it’s closer than I was gonna get he’s basically wrote this article and it’s on the on the foundations or on the shoulders of two other call them systems

15:03 shoulders of two other call them systems Architects but really going through the idea if anyone’s ever heard of the theory of constraints or The Logical thinking process which is not just for what you think it’s really based on systems but you if you initially read it you’ll think okay this is all engineering this is all I T or computer science absolutely not it’s actually really based on anything that’s a system which basically is defined as it has constraints it has a leader it has an

15:33 constraints it has a leader it has an objective objective and this whole in a sense methodology D principle goes through on how do you set a goal and Achieve something but more importantly how do you actually find when something’s wrong with a system whether it’s an organization whether it’s the goal that you’re trying to reach to reach and how do you identify it and how do you attack it so this is can be applied obviously from an I. T again engineering point of view but the majority of the examples actually from the

16:03 examples actually from the the original author is business based culture-based or like people based based so I like in the very beginning the article I think he does a really good job of saying hey before we really begin talking about all the idea here all the ideas that’s going on with this article right it says before we begin here are some details that are important to understand or key principles around he’s calling it Toc theory of constraints yeah yeah okay I defined

16:34 constraints yeah yeah okay I defined that one they’re correctly there number one if a system functions as a chain of interdependent components like step one step two step three like what Chris was doing his testing method you can find you can find and strengthen the weakest link that is that is in that system the constraint is the section of the links that are the weakest and I just listened to a a talk from Malcolm Gladwell who was talking about weak link structures and strong link structures and this also

17:04 and strong link structures and this also fits very well with that right if you if you want to improve something find where it’s the weakest in that entire system and work on that weakest link that will then raise the bar for everything in the system system you cannot number two is you you cannot improve a system a system by improving the components individually aka the local Optima because you’re limited by the weakest link if a chain has a weakest link it doesn’t matter how much you strengthen the other links you’re always dependent upon that

17:35 you’re always dependent upon that weakest link I think that’s another really good one as well the other assumption here is all systems operate in an environment where there is a cause and effect and then in those more complex systems it’s almost nearly impossible to discern due to different feedback loops and actions of independent agents outside sources so you’re trying to think of you sources so you’re trying to think of we’re assuming these are know we’re assuming these are cause and effect based situations and then lastly number four he says almost all problems you encounter in

18:05 almost all problems you encounter in Daily interactions are most systems with not actual problems rather than symptoms of a few root causes and I’m gonna I like the word here root cause because whenever I’m working around my house or doing things around the house I’m always like I’m always trying to find what was the cause of the problem and fixing that and and as you have kids you’ll find that there are a lot of weird behaviors that come out of them and you’re always trying to go about okay what was the root cause of this why why did why did you trip over and fall over on your bike well that was

18:35 and fall over on your bike well that was because you’re running sandals on your bike pedals this is a very real occurrence recently they are very slippery or you’re gonna fall off or they’re not very stable when you’re riding on a scooter or a bike so let’s not do that next time let’s put on real shoes and socks and go ride our bikes and scooters then so there’s all these like my mind like I always think to root causes and much to the I guess Chagrin of my family they don’t like it when I always do root cause and Analysis on everything but at the same time like the final sentence of that thing is in Toc nomenclature they’re

19:07 thing is in Toc nomenclature they’re called undesirable effects right like yes Hey kid hey like this this is what happens there’s a there’s a breakdown and an undesirable effect it happens face on pavement undesirable effect let’s figure how to fix that like that’s not a good idea well what I love there’s really two systems I think that will set up today there is the gold tree and this idea of the current reality tree so with any system or any problem and every I think every developer who’s going to love the next phrase I’m about to say

19:37 love the next phrase I’m about to say but it’s an appreciation for systems and there’s no way to identify the problem to identify what can be improved unless you identify the system gold and the idea of the goal here’s a little different than what I think we’ve ever used in terms of when we’ve talked about goals and objectives but really the goal is is just one objective that it’s trying to achieve it’s not an activity it’s kind achieve it’s not an activity it’s like an end point

20:07 of like an end point so it’s not so much like a process that’s going on but it’s actually somewhere that someone’s trying to go that makes sense that’s so that’s a little different than what we always say like like like our goals are very specific goals right like this is going to get into the weeds a bit and I know we’re already there but like I I just I had one thought around like the Four Points here that even outside of us delving deep into the analytics analyst

20:37 delving deep into the analytics analyst space right data space there there are two parts of this on a daily basis that I think all business users should should pay attention to one is this weakest link mentality right yes because this

20:51 link mentality right yes because this this chain of how business processes work in general whether it’s a manual process whether it’s automated and data people interact with this I think more so but like the the number four is where if it is most valuable if you start thinking about why is it that like all of a sudden my team is dealing with like 30 bugs a day on something rather than dealing with just the 30 bucks a day like thinking through what could be the

21:23 like thinking through what could be the cause of something right is really important to solving the right problems in a business and making sure you get to spend the time where you need it and remediate the the challenges that pop up and this weakest link thought and finding root cause problems and trying to solve those is something that I don’t see enough of in business right I see a lot of what we would term like what virtuous waste right I guess

21:54 like what virtuous waste right I guess right where you just yeah you’re working hard but do you need to right like can you work hard and then solve the something Downstream that is going to alleviate all of this pressure that you’re feeling or all these these pieces of work that have to get out the door and I think just having this Clarity around that like in the back your head like is this a root cause problem or is this just one of the things that we have to deal with and I think and also no I really I want to

22:24 I think and also no I really I want to hang on that point there just a minute Seth because I think I also agree very much with that One initial analogy that came to mind when you when I started reading these the first time I read them which was in in those four sections of framework elements elements the weakest link always felt like to me whenever I’m connecting to power bi with an Excel file or or I’m going from I’m using Excel file as a data source to RBI to me that feels like if you think about your data sources and how

22:54 think about your data sources and how they are reliable or not reliable at getting data into your system right there’s different I would say the quality of the data coming out of the different sources will vary very much depend on what controls you have inside that system to make sure that it’s accurate and up to date and Excel for me has been one of those it’s just been a boom it’s just it’s challenging just a different like someone has a column they put data where I wasn’t expecting it to a number column now turned into a text column because someone added some numbers and text together in one field

23:25 numbers and text together in one field like it just always kept changing just became harder for me to work with so I felt like yeah I can really resonate with the concept of the weakest link and especially when you’re talking about like getting data sets to refresh regularly right or you’re doing a lot of data cleaning or cleansing to make sure that works it’s a really good good practical example yeah I thought that’s what kind example yeah I thought that’s what first came to mind when I started of first came to mind when I started thinking about that one or that portion so Tommy you sorry we’re gonna go we’re gonna jump back now to Tommy’s

23:55 go we’re gonna jump back now to Tommy’s Point here talking about the the gold tree and how that works as well I will say there’s an image there about halfway down the page I’m not going to share it here on the screen but if you’re reading the article or following along in the article that’s in the chat window I’d highly recommend looking at the image that’s represented there it’s very well done and it really helps you talking about there are some things that are necessary conditions they were critical situations or critical success failures that you would

24:25 critical success failures that you would observe here factors factors sorry I think I read that wrong critical success factors and then there’s like the final goal or objective and this is interesting because this really frames out I think a lot of what I see happening inside organizations today where there are weaknesses in certain blocks of this diagram and therefore we have mistrust with data things don’t load correctly we start having like you load correctly we start having like silos of data and Pockets occurring know silos of data and Pockets occurring because this methodology or at least this diagram isn’t really helping that

24:56 this diagram isn’t really helping that broader broader right the goal is analytics is used effectively in the it by everyone in the organization that’s that is the main goal and everything that we’re talking about here in this diagram builds upon this I really like this diagram I thought this was really cool it’s actually a little like the okr framework because the the actual definition of a system’s goal is the result or the achievement toward what the effort is directed so it’s not the effort itself it’s not

25:28 it’s not the effort itself it’s not in a sense going to be any activity but the underneath that you could basically have in a sense rooted underneath it or those critical success factors those are little high level items that each one has to be a critical critical imperative work and progress towards because if it’s not then the goal won’t get done so they’re each in an integral link towards the goal actually become getting completed I really like that idea

25:58 completed I really like that idea because what what hmm the gold tree defines the goals of each each particular area right like you have the main goal or objective and then if we take down the step further to the critical success factors the reason he’s talking about those four points above is because if one of these five factors is is not done right is the link it’s not going to work correct right so by creating this hierarchy of like if we want to achieve

26:28 hierarchy of like if we want to achieve this thing here are the goals we set or objectives before us and then while not necessarily like in this gold tree Tommy you bring up almost we could extend the gold tree further and say hey if these are objectives that we have what are our what are our key results we need to tie into these to make this objective a reality right how do I flip this thing Green in terms of this overall objective within the organization to say

26:59 objective within the organization to say hey at the end of the day we’re moving the needle in driving for what what his top objective here is analytics is used effectively by everyone in your organization I’m hearing a lot of our past conversations in reading the bubbles on this chart yeah so bubbles here that that bring true in my mind is you that that bring true in my mind is does does the value of anal is the know does does the value of anal is the value of analytics clear and understood by everyone right do we do we all understand take a little extra effort do

27:32 understand take a little extra effort do analytics we’re gonna we’re gonna build some extra effort around manipulating our data Tommy I hear you saying I literally hear your voice reading this other one to me metric definitions are agreed upon everyone in the org right and that then funnels into key metrics are consistent everywhere and trusted no matter what report we look at like if we talk about sales we’re talking about sales at the same way we just we actually just ran into this major issue with a client where there are there is a definition coming from Finance around

28:05 definition coming from Finance around metrics in an organization this is how we want to count things but then the different business units will say well we we can’t with active items we we don’t count it without we include an active and inactive or we include canceled and active so everyone had a different definition a slightly different definition on what they were looking for and again it served their business units that was needed but we had to change the main definition of these phrases and say okay you weren’t looking at total sales you’re now doing total sales excluding this

28:35 now doing total sales excluding this this and this right defining additional Rich enrichment of a new measure or new calculation so right we were we were doing a bit more to enhance the conversation around communication around what that key metric was this is this is really good yeah and the the slight difference for us when we think when we normally think about a goal is we always think the it’s always gets muddled I think whenever people say they need goals but I think this is probably the most direct way to say it is goals should be stated as a condition not

29:05 should be stated as a condition not an outcome of activity not as an action or an activity itself so outcome of an activity yes I agree with that or no not as the action itself yes it should be stated as a so I think there’s two examples here we’ll say non-data examples one goal for a company increasing profitability now in the future or a cost-effective Improvement of the overall health of the community so those don’t have in a sense themselves inherently a number or an action tied to

29:37 inherently a number or an action tied to it but then that goes to and again the analytics is used effectively by everyone in the organization yeah yeah to your point right like when we flippantly throw around like ogsms and okrs within organization organization like okrs there are there’s a framework for how you develop those objectives and those key results right like they’re just as gnarly and in the weeds as this gets like as Leaders building objectives

30:09 gets like as Leaders building objectives within an organization yeah they are very similar to the types of goals that you see on this chart right like we need to yeah tier points they’re they’re more they’re not nebulous they’re directional right and in many cases they the challenge is how do you take that objective and break it down into the key result like into the the facets that you need to go Implement within the organization because those are the things that are measurable in some way

30:40 things that are measurable in some way shape or form they’re the actions that have to drive this larger objective and then you you see whether or not you’re successful I think another thing that strikes me as I was reading this and and looking at the gold tree is is it it I kept thinking of like a lot of the data maturity models that are out there there where like you have a description of the the certain area of business and where do you measure up and then like the

31:10 do you measure up and then like the actions you need to take out of that this seems to be much more

31:15 this seems to be much more specific in the objectives in like the hierarchy of like I’m building this I’m building all the components I would need to reach my goal versus an understanding like a new company trying to understand like where do I land in this data world and I think the maturity models that I’ve seen out there give some of that like first step in the direction towards building out something like this gold tree

31:46 what I really like to and then I think you guys don’t know what I’m talking about I have no idea I was above my intelligence so Mike I think it’s about my pay grade I think is that what you’re saying all right let me see if I can find it quickly no I I’ve been really appreciated I’m getting hung up on to me right now like look at this I’m getting hung up on I really like this diagram and when I read through the blocks and how they interdependent upon each other to build the tree structure down to the main objective I’m thinking how would I have ever gotten to this on

32:16 how would I have ever gotten to this on my own would I would I ever like I could articulate parts of this but I don’t think I would have been able to put together such a well-crafted like you together such a well-crafted like here’s the main objective analytics know here’s the main objective analytics is used by everyone here’s the five key pillars that we would think we would need to lean on everything in here we’ve talked about like there’s this whole series of bubbles around data quality and I’m like yeah 100 right data quality has to be monitored step one yep once data quality is monitored you’ll be able to find issues okay issues are resolved

32:46 to find issues okay issues are resolved quickly great yeah that makes sense okay once issues are resolved quickly and there are policies and procedures in place to help you enhance and grow the quality of the data you now have the outcome of data quality now remains high now I I talk about these things in a slightly different way but like this whole area or bubble around data quality supports supports the idea that key metrics can be trusted and and I really like the idea of like the stacking of ideas to get to a a

33:18 the stacking of ideas to get to a a different like the outcome based of what that means this this is a very logical way to go about I think things that we talk about all the time yes and and why as as I’m sitting here I’m like man I should I should have done something like this a long time ago it’s a good thing that’s what I’m thinking about this article because there is a cause and effect for everything yes right and and the longer your business the longer you understand this right like if you if if an organization chooses not to invest in

33:49 an organization chooses not to invest in these areas right yes these are not objectives to you right right and the then the effect is going to be you’re not going to have this thing yeah right or it’s a sub it’s like a partial it’s like not quite easily baked if this is important yes and this is where we need to focus some time because we want to reach this level if we don’t reach this level then we can’t get to the next level right and and to lay it out in such a straightforward way like regardless of using this specifically I

34:20 regardless of using this specifically I can see using something like a gold tree foreign like right here here’s progressively how you have to build upon the successes of something else and with data it is absolutely that way right Mike to your to your point one of our goals is to ensure that there is data quality and governance around data sets that come out of this business unit okay well like if you’re looking at a

34:51 okay well like if you’re looking at a power bi file that has Excel documents all over the place like that’s and it’s manually refreshed and you’re doing all these things like those are are things that are not accepted in that that one of these boxes right we have to get rid of that we have to get to an automated standardized way for the data yes and then we move on to the next step right then we can put qualities around it and fresh now freshness timings and all all these other things but I can’t get to step two without doing step one and and so much of getting to this point

35:22 and and so much of getting to this point where we can declare Victory around like data quality and governance or reach higher levels of satisfaction throughout the organization related to all these high level objectives we talk about like in data literacy adoption all these things things it’s it’s one of those things that’s very easy to talk about and and business I think I think s their data quality is a lot better than it is but that’s so true in the

35:52 than it is but that’s so true in the spectrum of all of the business it’s really not that great so that’s a really good point because I think a lot of times we always see it goals whether it’s for the organization our team or for ourselves on a chronological framework where it’s like we all get here in February and we ought to do those things in order to get there in this order like like a scrum where this framework and this last piece of that with those critical success factors is finally the things we’re going to

36:22 finally the things we’re going to measure right where it’s the unnecessary conditions they’re either yes or no or there’s something that’s actually quantifiable which comes us but think about something like you said like how do you we always get tasks like we need to improve our data culture or we need to increase we’re paying all this for power bi and we’re always starting with these branches where if we could actually put in this framework where the first two items here are not necessarily directly measurable and nor do they have in a sense a a path right where it’s

36:54 in a sense a a path right where it’s like we have to do step one that’s a two step three all of our critical success factors so getting people to use power bi bi getting in usage at a center of excellence are essential to it yes yeah that’s a really good point there because this is one thing that I’m finding very recently in my experience right now is when I look at different organizations and what they’re trying to manage everyone wants good quality data everyone’s reports serving the needs everyone wants metrics and information

37:24 everyone wants metrics and information that’s consistent across the team but what we’re not doing is we’re not looking at okay well we’ve built we’ve spent a lot of money and or effort on these reports and or objects but we’re not able to accurately monitor the usage the consumption the activities that are occurring inside our environment we don’t know what people are doing and more and more I’m recommending when you start doing a power bi project if you’re going to be serious about it first step start collecting data start collecting what people are doing inside your

37:54 what people are doing inside your environment because you do need to know which which reports workspaces and artifacts that are being generated in power bi are being used the most would would the the one of the things that is just striking me here is this this is one of those articles so thank you I guess for for writing it that is why I love this conversation with you guys and the audience ever every Tuesday and Thursday is it brings it brings to like different

38:24 is it brings it brings to like different approaches to enhance our daily work and one of those that like it’s too often I think that we aren’t thinking strategically and recently one of the things I’ll share is I always take like I always have an attack list for the day and I use that in one note all right and I think we talked about this in the past but it recently I’ve I’ve altered it a little bit to be three parts I always haven’t execute those are the things that need to get need to happen today or within a certain time

38:54 today or within a certain time frame frame I have a planning section so I have a bunch of planning tasks and those are inter like not inter related but a different part of like what I have to do as an organ but the other third part is strategy and where this one comes in is this is a strategy idea like this is how do you continually push the needle as anyone in the org right but particularly people who are responsible for objectives or hitting goals you always have to keep in mind are we are we

39:25 have to keep in mind are we are we working towards one of these objectives are we in this goal tree I think is something that I’m absolutely going to start to implement or at least figure out how this logical flow of like I should be able to tie in results to these things and this rolls up to my objective or whatever the case may be within the organization but it just strikes me as that like this is one of those great like hey maybe maybe I spend more time more time just flushing out what the goal is in

39:55 just flushing out what the goal is in these these particular areas related to obviously the world we live in which is data data so that’s you’re actually going on to that current reality Theory right where when we’re trying to identify a problem or something that not necessarily undesirable effect but we have a system and one of the things that’s really talked about is any process or system we put in place Center of Excellence access the data is going to have two types of effects and or potentially three the ones that we’re

40:26 or potentially three the ones that we’re intending but there’s always going to be the unintended which are usually the undesirable effects because they are just they always grow the processes that grow out of a system so to speak which to your point like when we’re trying to attack a lot of times we’re trying to attack the root like oh you want more usage okay we’ll do that or oh these reports the data is wrong well rather than just going to that report and editing it or telling people to put these filters trying to take that step back and say okay what system or is in place right now

40:56 what system or is in place right now that’s causing this to happen repetitive like but yeah I think I think that’s that that’s what I’m also referring to is not this isn’t just net new stuff yeah right is taking this methodology of like okay what what is my ideal how do I want my my area to grow how do I want this project to look versus and you can do that against existing implementations too right to figure out

41:27 implementations too right to figure out where the breakdown is and I think he goes into that in the next triav tree specifically where it’s like talking through like the if the if if the gold tree is the

41:41 tree is the the ideal right what is what is current reality right and this the current reality tree is going to be just as I think granular from a logical perspective but in the negative right if that’s how I read through things yeah It’s it’s in a negative but this is mean it’s it’s in a negative but this is also I think of helping you identify what is the root cause of why one of your your gold tree elements are where when we when we say we don’t have quality of data you hear that word right I can’t trust this data

42:12 word right I can’t trust this data there’s no quality in it right the current reality tree is that opposite thinking you start at the bottom of this tree and work up with the what is causing the problem right so you if if I go into this tree thinking the mindset is I have poor data quality and then you read from the bottom of the tree and work your way up okay is my is the reason we have poor data quality is because everyone does not know how to use the analytics no we we’ve done the trading it it should be it should be clear right

42:42 it should be clear right the analytics the value of analytics is opaque and I understood no we have it we have a charter okay so it looks like we understand that part Murphy’s Law causes systems to break okay it’s just part of things happen right people enter in weird data it’s going to break or maybe we find okay maybe the engineers in our team are making frequent changes to systems maybe those changes are causing problems so it actually it walks you through like okay like okay thinking through these steps these are the reasons why

43:13 these are the reasons why halfway up you see there’s a quality in issue data quality issues go unnoticed and unfixed maybe that’s part of the problem right we we have some process in place people are doing things but we’re not actually monitoring and we’re not fixing data quality problems therefore the data quality deteriorates we lose trust and then all the way at the very end it goes all the way up to the very top of this the same way that the gold tree worked it basically says at the very final step if if this stuff starts failing at any point in time here

43:43 starts failing at any point in time here right if any One of These Arms of this tree fail we start eroding trust in our data and eventually the organization says I I’m losing interest in having good data quality I’m losing interest in being data driven because it’s so hard for us because we don’t know but it’s we’re not making a clear analysis output does that make sense it absolutely does because this is also whether whether it’s these these dependency trees right are a good like logical way to build these things

44:15 like logical way to build these things out so that they’re a visual the same way it’s having an impact on us in in how we’re like oh yeah like like a lot of the things we’ve talked about like aligned with this is bringing clarity through a visualization like this to a business that says hey how if you’re stuck in this state if you’re starting to lose interest why what are the what are these main things that are like all of the sudden you’re you’re saying our our quality is is not good it’s deteriorating why like understanding is

44:47 deteriorating why like understanding is 10 different things that you could do to start to fix those and then the the value of fixing those comes up to a solid point that you can move on because I think a lot of times with going into people it’s always hard with people but I think it will say you go into the office tomorrow and you’ve completely like digested this like kind completely like digested this like framework and they say hey of framework and they say hey the we’re looking at the data on our

45:17 the we’re looking at the data on our sales report it’s wrong it’s been wrong every other week for whatever reason fix it and you start going in going well let’s try to identify the root cause here let’s start building this up you do need other people throughout this process to either agree that you’re not going crazy that these are actually true effects that are occurring but then it’s actually helping you break down like is this just occurring because the report’s bad is it bad because of someone not inputting their information is it a

45:47 inputting their information is it a techno technology thing and really breaking down but also for the users too to break down why are you guys getting faulty data sometimes or why are our growth not go increasing but you do need the team users Department to at least agree on okay this is the problem we’re seeing it it I really do love I almost I think I almost love that this current reality especially what you’ve been saying is in terms of you

46:17 you’ve been saying is in terms of you you’ve been saying is in terms of let’s identify together let’s in know let’s identify together let’s in that let’s identify the root cause here so I look at this current reality tree and think wow this is very well put together but I’m also thinking about okay what if I had to just start this over from scratch where would I start from like how would I build my own version of this or for whatever else I may need again this one really focuses on like data quality metrics and making sure people understand the value of analytics

46:47 people understand the value of analytics right those are the key areas this thing’s addressing so I’m trying to take this example and or the goals tree and overlay this against something that I’ve actually done right how would I have written this when do I find something is a root cause and how would I work my way back from doing my analysis and documenting it in a way that communicates that back to leadership here’s what we did here’s a failure or a part where we have identified weakness and then basically being able to

47:17 and then basically being able to rebuild this chart specific for my situation let me give you an example here yeah let’s let’s think about this Excel file causing problems in my data in my data system right so what typically would happen for me which would be is I would build something on top of excel I would publish the report it would get shared via whatever and then what typically would happen is I’ll get a message from someone or a team and say hey this report doesn’t refresh so immediately I have an input I have

47:47 so immediately I have an input I have some element that says okay I need to start investigating why is this report not refreshing or the numbers don’t look right I’m listening for that notice right so already we’ve already had a a refresh issue the metrics are not right there’s there’s already a higher level item here but that’s not the root cause the root cause is not it didn’t refresh we had to go one layers deeper okay so what recent changes did we do where which of the sources of data did I have that it actually failed on

48:18 did I have that it actually failed on which table did it actually fail on so you go I like walk my way down until I find okay this data set is not refreshing because the credentials timed out great now I have a step here to say okay the credentials timed out on this data source or someone left the company and they’re no longer there those credentials now to be need to be re-entered how do we solve that moving forward to in to basically say enabling that to not be a Blocker in the future or we go other we get down even further and say okay we’re our we were

48:49 further and say okay we’re our we were relying upon an Excel source and the data changed therefore it broke some of my data engineering steps in power query that then caused the data set to fail we had a duplicate number in a column so all these things can be broken down to like what is the root cause of these things and I’m trying to in my mind think through as I’m walking through that geek that root cause analysis to me it starts with I could identify on this

49:19 identify on this current reality tree any bubble in the middle that’s where I get involved I get involved at a bubble somewhere in the middle of that of that chart and I had to work my way from that middle area further to I’ll until I actually find the root cause so in my mind I’m thinking like if I’m going to build these for other people or customers or for me I would have to start with a what are common known patterns that I understand today that make things fail start there and work my way back to the different middle tiered element and build out the

49:50 middle tiered element and build out the tree beneath of it the roots essentially does that make sense yeah I think you could you could definitely do this on the granular level with specific tools or like the specific connections or specific ways you’re implementing things because they’re there likely is that chain that we talked about right like this this dependencies that we we would deem unacceptable in a in in a system that would we would call this right so I what I don’t know is

50:20 this right so I what I don’t know is like a lot of This Is A Step Above it’s a little higher level right that that speaks to larger swaths of hey we need to we need to increase data quality or or the quality of data that comes through so therefore we have 10 different third-party systems that we ingest data on we want to make sure we understand the business logic and the rules and identify whether or not there’s bad data coming into the system right that’s a it’s a huge task right versus what you’re defining would be the

50:52 versus what you’re defining would be the same thing in if I want to reach a goal maybe this is a thing it depends on the goal right so if your goal is I like all of these reports should reach a certain level if they’re going to be deployed to this workspace we call them certified in order to do that here’s the dependency tree underneath that and I think that’s where your comments about knowing the the good or bad parts about how something is implemented you could have a gold tree and say this is what it looks like and then several of these

51:24 looks like and then several of these current reality trees are like something that is a placeholder that goes along that hey if you build a report in this way this is your current reality if you build a report in this way this is your current reality how do you get out of your current reality into this goal framework I think is the piece that’s that’s I don’t say missing but I didn’t find in here it’s like hey if I have these goals how does that butt up against what my world looks like right now because I would need to plug into one of these objectives in order to fix that

51:57 objectives in order to fix that current reality and move towards a larger goal so I think a big thing is they talk about the weakest link so I

52:04 they talk about the weakest link so I don’t know in in terms of if you were to identify the root causes or the things that you can control because Mike I think a lot of things you’re saying is it’s not I don’t want to say personal but that’s like okay my development my reporting and I think I honestly reading through this I was thinking more of this on like an Enterprise or adoption mindset it’s like oh no I can apply this to my just my daily life or I was taking it personal initially to kind was taking it personal initially to relate it to something that I do of relate it to something that I do right now but I think you’re right Tommy like I think the the intent here is to

52:34 like I think the the intent here is to take something that you understand personally internalize the learning of like the value these things can bring both the gold tree and the current reality tree and then saying okay now that I understand how do you how I would build and or utilize these things for what I’m doing let’s extrapolate into what the organization is doing right the goals will be bigger to your point Seth right the goals are bigger the objectives are larger we’re talking about bigger outcomes across many team members and many different teams but I think the bigger thing when it

53:05 but I think the bigger thing when it comes to if when you now if you’re let’s say tackle this with a group of people whether it’s your team or stakeholders one of the biggest things that you have to have is obviously that agreement but also there’s that realization that there are some things that are just out of your control even if you’ve identified a root problem it’s again we’re you’re not talking to computers where like you are the error thing thing and trying to understand okay what’s in our what can we do now and at least

53:36 in our what can we do now and at least try to chop down this idea of the weakest link I the the idea is if you were to look at five issues five causes the idea is you’re only as good as your weakest link we know that cliche but that is like hammered in in the idea of the undesirable effects where you will never get to your goal you’re gonna fix the problem unless you at least start with the lowest hanging fruit or the like the lowest constraint that’s it within your power

54:06 constraint that’s it within your power which is another way to tackle I think the things that we do both from the development and then when we are trying to do the soft skills and the the gray area of what we do is rather than trying to tackle everything now which is usually my tendency where it’s like well we do need more adoption I think we need more support it’s like well what’s the biggest issue let’s identify the issue that my team in sales agrees on and let’s say okay we need to focus on improving this

54:36 improving this first because this little it is the worst of all the links so far so we need to strengthen this then we’ll move on to whatever the weakest is after that which I think is a good way too with yes it’s so funny because like even what we do in in bi how many times have we said this like we don’t have a lot of our own internal reporting reporting in terms of like looking at our own success factors Mike you and I were talking a few days ago on the podcast

55:06 talking a few days ago on the podcast itself and I was like man I think there’s a maybe we could do something here it’s like show me a number thing and it’s that to me this goes back to something I’ve said on you goes back to something I’ve said on I think part of our role when I’m know I think part of our role when I’m with a fabric coming in and AI is that ability to be a leader in this framework too because this all comes back to some numbers that we’re going to quantify and those necessary components or those NCS and that’s a great report that in terms of

55:36 great report that in terms of building report off of these goals but everything becomes something off your measuring measuring yeah and I think a lot of things here we’re talking about here is there’s a lot of behavioral movements here that we’re talking about and again we’re talking about system architecture and the neat thing about what’s being applied here I think I like these Frameworks these mental Frameworks that we’re talking about here work very well for what and again the author here is really talking about analytical framework which resonates very well with us however these Frameworks can be applied anywhere it’s

56:06 Frameworks can be applied anywhere it’s a framework it could be applied in any at home other situations different business units it doesn’t have to be focused only on analytics so really really like this article and really found that this was very impactful for me and have to figure out how to incorporate some language or learnings around this in other exercises that I do with with clients around this as well as well well I think right now it’s a perfectly good time to go ask chat GPT the real knowledge thing of all this figuring out what it really knows here about

56:37 out what it really knows here about understanding a gold tree and a current reality tree so I asked it basically what are the advantages of the gold tree and current reality tree and so here’s the response the initial setup was here is defining what is the gold tree the gold tree is a diagram that shows the necessary and sufficient conditions to achieve a desired goal I thought that summarized that very well the current reality tree is a problem-solving tool that shows the different root causes and undesirable effects or symptoms that prevent you

57:09 effects or symptoms that prevent you from reaching your final goal so yeah I like that and definition there because the current reality to read literally is these are the things that have stopped you these are the things that are going to block you from making your final goal a reality and I like that that’s a good definition there and then lastly it says at the bottom here the advantage of these tooltars are the advantages of these two two tools are wow what a tongue twister they help you think systematically and logically about your situation and help you think about your goals

57:39 you think about your goals they help you communicate clearly and effectively to others who are involved or affected by your process or what you’re trying to reach towards your goal which I think is good too right we got to get people on board Tommy you’re just talking about the people part of this is a challenge and then and then this this these tools also help you prioritize which actions and monitor your progress towards a goal right when we have metric issues or if we have data quality issues we could actually stand back and say okay how do we measure that

58:10 back and say okay how do we measure that what is the measurement of data quality what does that look like and then you can quantify that and you can then work towards getting a better number or a better measurement in that area which I think I really like that idea as well and then lastly here they help you adapt to changes in your environment or in your organization by updating your analysis and or your Solutions I don’t think the article talked about changing Dynamics or or things changing over time but I think this does this I think this helps you get closer to

58:42 think this helps you get closer to the the data landscape is at the data landscape is Ever Changing I think we see this even more now with fabric being introduced we just saw a whole bunch of Technology get locked in our laps or dropped in our laps now and now this is changing how we need to potentially think about our goals goals does this change what we think about now because now our team can do more with fabric our team can do more did engineering as opposed to before work couldn’t couldn’t if we can incorporate more of these

59:12 if we can incorporate more of these goals I think this is interesting and I like that idea because but at the same time like that’s where that’s why sticking keeping up to date doing r d right like things like that could influence or impact how fast you can solve some of these higher level objectives right like and that’s where the strategy component of this really strikes me because these aren’t things that you’re investing tons of time in every single day but if you’re if you’re always keeping that high level strategy in your mind right you’re

59:44 strategy in your mind right you’re plugging into those Key Parts right new technologies new methods to perform something et cetera Etc it just feeds that long-term vision and helps you achieve those long-term goals because guarantee like your focus always wants to come in like oh my gosh what do we gotta get done today or this yes and without without aligning a big picture yeah yeah because you start missing them you can you can get focus in the weeds and just fixing the problem and not actually fixing the in map

60:14 and not actually fixing the in map making the solution to fix the systemic problem that is occurring over and over again ironically the older you get time seems to just fly by faster and you’re like oh wait I have a major objective I gotta hit in two weeks exactly well that we we definitely appreciate your time today this one actually incredibly fast really good conversation this was a great article so thank you very much I’m gonna try and say the name again help me with the name guys thank you very much for saying

60:45 guys thank you very much for saying erst erst erst erst might be a good word okay thank you Ernst for writing a great article we really appreciate you this was a great article to read highly recommend you read it on your own internalize us a bit see where this maybe would apply in your organization bar of the diagram there’s actually a couple great diagrams or images here that help you describe some key objectives or main goals of your analytics Department which you may want to borrow and reuse so thank you very much with that we only ask you if

61:15 very much with that we only ask you if you enjoy this conversation if you like we were talking about here if you learned a couple nuggets of information here that you can take away from your daily job we really would appreciate you sharing and or subscribing to our Channel it really helps us grow the audience and helps us communicate this out to more people with that tell me where else can you find the podcast you can find the podcast anywhere it’s available Google Spotify make sure to subscribe and tell your friends if you have a topic an article or question you want us to talk about you can ask us go to powerbi. tips

61:46 about you can ask us go to powerbi. tips the podcast and we have a mailbag for you to submit your questions and finally join us live every Tuesdays and Thursday 7 30 a. m Central I keep thinking there’s got to be some way of like doing this analyst is there any data patterns or things that I’m missing inside the context of the podcast or maybe there’s some maybe there’s some major issues that we’ve been having here some systemic problems that we need to address here on the podcast like Michael getting up early enough to be able to get on the video and get everything started on time maybe there’s some systemic problems I need to address

62:17 systemic problems I need to address analysis of like why we have challenges oh oh thinking I’m thinking I need to apply this in other places anyways thank you all very much we really appreciate your time have a great day and we’ll catch you next time bye

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/empodcast and submit your topic ideas.

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