Customer Understanding & Fabric – Ep. 280

Customer understanding is one of those goals that sounds obvious—until you try to stitch it together. Marketing has campaign touches, sales has CRM activity, product has usage telemetry, support has tickets, and finance has renewal history. Each system is ‘right’ in isolation, but the business questions live between them.

In this episode, the team reacts to a Data Marc article on the future of customer understanding through Microsoft Fabric. The big idea: Fabric won’t magically create a customer 360 for you—but features like shortcuts, mirroring, and Direct Lake can lower the cost of connecting to distributed data so you can spend more time on the hard parts: identities, definitions, and trust.

News & Announcements

• The Future of Customer Understanding Through Microsoft Fabric — Data Marc’s perspective on why ‘Customer 360’ is fundamentally an integration + identity problem, and how Fabric’s lake-first approach can make that work more approachable.
• Submit a topic idea (Explicit Measures Podcast) — Have a real-world scenario you want the team to unpack? Send it in for a future episode.
• PowerBI.tips Podcast — Subscribe and browse the full back catalog of Explicit Measures episodes.
• Power BI Theme Generator (Tips+) — Quickly generate and manage report themes so formatting stays consistent across your org.

Main Discussion

A customer 360 isn’t a single report—it’s a repeatable way to answer questions like: Which customers are adopting? Who is at risk? Which campaigns actually convert? To do that, you have to connect data that was never designed to connect.

The discussion covers the practical reality most teams face: data lives in many systems, identifiers don’t line up cleanly, and extracting data via connectors/APIs is often the bottleneck. Fabric is interesting here because it adds more ways to virtualize and reuse data (instead of copying it into a new silo every time you want a semantic model).

Key takeaways:

• Customer data will always be distributed. Even small orgs end up with CRM + marketing + product + support tools that don’t talk to each other.
• Identity is the hard problem. If you can’t reliably link the same person/account across systems, your ‘360’ becomes multiple partial stories.
• APIs/connectors determine your ceiling. If a platform makes extraction painful (complex JSON, weak APIs, limited export), your analytics maturity stalls until you solve ingestion differently.
• Shortcuts and mirroring change the economics. Reusing data where it lives (and exposing it as tables in Fabric) can reduce the endless ‘copy/refresh/compute’ cycle BI teams have lived with for years.
• Direct Lake is a serious enabler. Being able to iterate on models against lake data without repeated big imports can speed up experimentation and deployment cycles.
• Virtualization shifts the load, it doesn’t delete it. If you’re ‘hammering’ source systems through mirroring/virtual access, performance and governance still matter—just in a different place.
• Definitions and trust come first. Before tools, teams need agreement on what ‘customer’, ‘active’, ‘churn’, and ‘conversion’ mean—otherwise you’ll just industrialize disagreement.
• Expect this to be incremental. The future is promising, but customer 360 still requires deliberate data engineering and a data culture that treats quality issues as fixable—not normal.

Looking Forward

Treat customer understanding as a product: start with shared identifiers and definitions, then use Fabric’s virtualization (shortcuts/mirroring/Direct Lake) to iterate faster without rebuilding your data estate every time.

Episode Transcript

0:27 good morning and welcome back to the explicit measures podcast with Tommy Seth and Mike hello everyone and welcome back to the show good morning Mike good morning Mike now typically we’ll do a little bit of news or articles but this is a pre-record episode so we don’t know what news is going to be out in when this episode airs so because of that we’re just going to jump into the main topic today our main topic today which was also in the description below you’ll be able to find the article that we’re talking about today an article from data Mark talking about organizations and the

0:58 Mark talking about organizations and the future of customer customer understanding through their fabric it’ll be interesting this is this is talking through some of those customer 360s do you really know your customer what are you doing and how that relates to what fabric is going to portray there as well anyways let’s jump in and we’ll start kicking off some kicking the tires here on part of the article prob want to give us kind of the article prob want to give us maybe an initial impression and start of maybe an initial impression and start us off a little bit with the topic here I I love this article it’s a bit near

1:28 I I love this article it’s a bit near and dear to my heart data marks talking about how do we look at customers from a 360 holistic view from how do we Market to them how do we sell to them and typically this has been a lot of systems you think CRM you think Dynamics you think think Salesforce with fabric becoming more of a feature in what we do can the features of fabric mirroring shortcuts play a bigger role to provide an easier and maybe a finally a holistic view of the

1:58 maybe a finally a holistic view of the 360 customer analytics so this is this is a little bit of a let’s clarify a topic here this is a spin on like fabric is the technology a customer 360 is what does your customer do who do you sell to how do you understand you you sell to how do you understand what what makes their buying habits know what what makes their buying habits occur how does that relate to your company and I think I think there’s a couple points here in the very beginning of the article talking about

2:28 beginning of the article talking about this data doesn’t always live in one single place all the time and I think maybe the argument here initially thinking about this article is hey there’s going to be data all over the place different systems but you got to pull it together fabric could be one of those places where you pull together multiple metrics for a customer to understand what they’re doing yeah I agree I think what i’ I’d love to do is brainstorm a little bit here right at the beginning before we get into like the technology aspects because one I I

2:58 the technology aspects because one I I agree with Tommy from the St in both of you I think one of the greatest examples of probably one of the the highest value propositions of business intelligence is this right because for the vast majority of companies that I’ve interacted with to to Mark to data Mark’s Point here and to what Tom he’s alluding to right away like you you very rarely have a single system that tracks

3:28 rarely have a single system that tracks all this MH and regardless of what type of business you’re in there’s a a large spectrum of where you store customer information in how you interact with them like or your own customers where you’re trying to Market new customers like how are those marketing plans panning out are they resulting in new opportunities what are how do customers like use your system are they happy you

3:58 like use your system are they happy you like use your system are they happy Etc like there’s so much across the know Etc like there’s so much across the board here like what are your guys’ some of your experiences as far as like the different points if somebody’s going to plug into this topic of the customer 360 view I listed off a couple but what are what are some that you see and maybe like some of the systems that we see we need to pull data from well I’m going to point out one that’s very real and near and dear to our hearts we broadcast the podcast on many different media platforms so one

4:28 on many different media platforms so one that we are constantly trying to figure out is what topics are the ones that are resonating with listeners very plain and simple so this is you you listening to this I’m talking about you so but we do this on Twitter we do this on LinkedIn we also do this on YouTube and so each of those platforms has a system that captures data about who’s listening when they’re listening to it how many listens are occurring and so none of those systems talk together so even just for us even at a simple

4:59 so even just for us even at a simple level you just things you’re making data you’re making content you’re making something that’s getting distributed everywhere and not every tool has an easy method to get it out like the Spotify API is horrible there is no really easy way to suck the data out of that system and YouTube is okay but it’s not as easy as Google analytics so there’s a whole bunch of like challenges with whenever you’re let me let me step back one moment here so that’s an that’s a specific example

5:30 specific example in companies I think they struggle to find that software that enables the ability to handle customers well but then also blend that data with other customer metrics and data like your purchase history probably doesn’t live inside Salesforce it it probably lives in some on-prem system where you identify that user and what they’ve purchased over time you will need to join some of that Salesforce type data

6:01 join some of that Salesforce type data with other data sources and so that the challenge I think here is that we’re alluding to is the better 360 view of things turns out when you have all the sources coming together and you can link users together and with Google’s change in me mentality here around websites and things it’s actually gotten harder to figure out where that user is and what is that user across multiple platforms yeah and and you’re just speaking from the marketing side too there’s honestly there’s two main areas of barriers that make this

6:33 main areas of barriers that make this hard at most organizations you have the technological side because there’s three areas we’re focusing on the audience how do I track an audience how do I track opportunities and how do I track accounts from a pure technological side you have to put in ample tracking and none of these systems talk to each other if I have audience well there’s no customer ID yeah typically typically speaking whether you’re B2B or B Toc and again audience opportunity and accounts are all different platforms which need different systems which need

7:05 which need different systems which need tracking already put in place for that to work which usually don’t talk to each other the other side is the business which usually is a barrier how do I know I’m spending money the right way if I’m marketing towards someone am I getting banged for my buck are these campaigns effective if we’re doing opportunities there’s all these sales funnels that we have to try to track okay how long should we talk to someone is this an ample opportunity again so those two main areas make this very hard for us to track audience

7:37 very hard for us to track audience opportunity accounts from a technological side and the same area from the business which are really the hardest part of getting this from an analytical point of view the last thing too is we’re also dealing usually with the row-based databases right we’re not dealing with analytical designed t database designs they’re meant to look at a single account so this makes it already harder for us so we have a lot of barriers just dealing with this from a business intelligence

8:09 side yeah one of there’s one yeah the only other one I I I’ve experienced also is there there’s typically an opportunity depending on levels of area or business like it’s not just Revenue generation but it also could highlight cost savings for the organization as well right ensuring that all of your customers are using a platform in the right way are on the right plan are purchasing the things the right way etc etc so it’s it’s a

8:39 the right way etc etc so it’s it’s a really large scope that produces many different like I think types of reports that are extremely valuable to a business yeah and I think the the real important part here is not just the tracking but especially in the article it’s getting excuse me it’s getting right the 360 View from acquisition to opportunity to account that is the hardest part that a lot of organizations don’t have the design in place of

9:09 don’t have the design in place of getting that holistic view can I see the person I brought in through this campaign to do X Y and Z and now how are they fairing now they have all the pieces in place but they can’t connect all the all the systems and and I would I feel like there’s yes I think this is like the data challenge right the why the data is an issue to get it out of the systems and get it into a central system I also think there’s a a business you you mentioned something about the business Tomy and I think I think not

9:39 business Tomy and I think I think not every tool is purchased by the same team all the time that’s a good point yeah and so I think I’ve observed this a couple times in talking with organizations and at the end of the day like every tool adds some level of value

9:55 like every tool adds some level of value to an organization Baron it works y however ever whether or not those tools talk to each other and I think there’s and I don’t know how to I don’t know how to correctly articulate this but like where is the real value in these tools coming from so you may have a customer relationship management system a CRM system that’s sitting out there that system may have a good amount of value it may have some AI built into it it may be this really easy weight

10:25 it it may be this really easy weight thing that integrates with your emails it may even be able to listen to to like your phone calls and track those and put a summary together what you talked about to the client and things all that data is fine and dandy but it lives in that system so like the challenge I find here is organizations find these really rich tools that are third party tools that do a lot of great things but when it comes back to okay now we need to figure out the plan to get the data out and suck it into or aggregate it or pull it out and

10:57 into or aggregate it or pull it out and then think about okay well how how does that data even just a simple metric like number of emails sent to customer how does that correspond to does this customer stick with us long term do we know any more about them because of this so just because you’re collecting the data just because you’re going to go out and buy that tool it doesn’t necessarily mean it’s going to integrate very well with your rest of your system and so in the article here from data Mark he’s talking a lot about like the Microsoft ecosystem right its

11:27 like the Microsoft ecosystem right its Dynamics it’s hang there for a second fa okay sorry yep go ahead no I I agree with you right but like why do business why do business areas like buy into these tools because it’s solving a business problem it does I agree right the I I think I think the other touch point that that you’re you’re scratching here is something that we’ve talked a lot about where organizational units typically operate like in silos right

11:59 typically operate like in silos right like age I’m I’m this group here’s our problem we’re going to go solve it how do we solve that oh hey someone’s used this tool before or budget hey we’re going to we’re going to review these tools to solve these problems as opposed to custom devit because we can just plug in then we’re going to get taught how to use the tool and it’s going to solve this problem I think all of the questions after that so it’s solving a problem for them obviously right the next steps to that though and where you’re going to go with microp soft

12:29 you’re going to go with microp soft ecosystem I think speak largely to why we are pushing these these narratives or conversations around data culture and especially data literacy because this that is where you sol start solving the silos of this stuff when business people in those initial conversations where they’re choosing platforms to solve their problems

12:59 their problems know that yeah it’s a third party system I’m G to I’m going to have a need to plug this into our data how do I how do you guys help us facilitate that aspect and that’s where also at the same time like we come at this I think differently than any other part of the organization as business intelligence people because we have a foot in it and a foot in business and these are the data challenges and things we struggle with on a daily basis but if you think about holistically we probably one of the best

13:29 holistically we probably one of the best views or understandings of like the business Health from a data perspective and that’s exactly what we’re trying to bring or why you would go down those data culture initiatives so that everybody starts to understand at least the basics of plugging stuff together and and and this is really heavy what Seth is saying because I I don’t think you can really deemphasize how important is to refine your statement Mike you said the business is solving problems

13:59 said the business is solving problems with these tools I’m going to refine that to say the Departments are solving problems because typically agree it’s great utilization and great value for that department but it’s not thought of in a holistic view of how does this tie in to the overall organization correct I buy an organ a marketing tool or say Google analytics and it serves a great purpose for us but as soon as the conversation comes up of how do we tie this and everything else that’s never been conceived that’s never been thought

14:30 been conceived that’s never been thought of okay what’s the next step all I know is my social media campaigns are great because I bought Sprout or I bought buffer or whatever the product is yeah or and that’s usually the problem this was my life we’re explaining the first years of my life in bi was trying to tie in all these different technological tools to get this holistic view because that’s eventually what everyone wanted to see yes and I want to jump into your I love the comment Tommy and that’s a really good distinction there on

15:00 really good distinction there on Department level analytics it’s solving Department level problems which is great again I’m not the idea here is I’m not trying to poo poo this idea of like hey don’t go buy third party tools that’s not what I’m trying to say at all I I think Seth you hit the nail on the head here when you started talking about when you took you knew where I was going with things and and you 100% read my mind on the way through this meaning there is there is a level of hey we need to at least holistically think about the integration of that data coming back

15:30 integration of that data coming back together and so whatever third party tool you’re going to go get fine go get it not a big deal but there is a moment in time where you’re like if I’m going to use these other tools What is the characteristic I care about in those tools that I need to bring back to either tie the data together a customer ID something that’s important because you have Central data somewhere there needs to be at least a conversation around okay you’re you’re going to go byy that tool is your intent to keep

16:01 byy that tool is your intent to keep that tool in a vacuum and never join it back to comp company data and they may say oh yes it is we’re going to do our own thing and that might be just a naive lens on thinking through how this data looks but we know in a lot of these projects there’s always some need to come back and say well okay how do I know the user engaged with our product on Twitter as well as watched our YouTube tutorial and then did something else or engaged with some AD we spent money on at the end of the day we want to make sure that whatever we’re

16:32 want to make sure that whatever we’re spending our money on is producing the most effective return to to get customers in the door our cost to acquire customers we want to bring those out as low as possible and that’s what these tools are trying to do is to bring that customer acquisition cost or the retention of existing customers as low as possible yeah and and I would I would not argue but I believe like this is where and why organizations have start to create rigor around who can purchase what because if if you don’t interject

17:06 what because if if you don’t interject like this is where does the tool have an output that we can extract data for the core information that we would want to know that you’re solving the problem for in that application if the answer is no like in my mind you immediately throw it out because I you guys I’m sure have also dealt with this where it’s like e they there’s a little lip service but then the minute you like you ask for something or try to engage and do you have an API do you have access to this or oh yeah we have an API but when

17:36 this or oh yeah we have an API but when it comes down to like the one table you need they don’t have access to that one you would need to purchase the next level the premium the the the premium for data right I’m all about premium for aggregated if you went through all the exercise of taking the system and modeling it and simplifying it and and giving me value yep out of that Prem me up man right but if it’s just the data big problem there bud like uhuh not not going to sign on with you well it’s

18:07 not going to sign on with you well it’s not enough just for the data to have the extraction and I think the unintended consequence that we haven’t even discussed is each of these systems they may output data but what becomes is that data becomes siloed and the longer you wait that becomes that Department’s metrics and goals based on that system based on how that system measures it it’s not enough for just the data to extract it I need that data to talk to the other systems and I need to have that plan there and I I’ve seen this where it’s like well we’ve been using

18:38 where it’s like well we’ve been using this system for this long this is where our metrics are coming from as soon as bi starts integrating it and the metrics change because now it’s evaluated based on other other filtering or how the organization wants to see it they’re like this doesn’t match our systems tools this doesn’t match what the output of our our what we see in the UI so therefore the source of Truth is going to stay with us in what we see and the longer you wait you get that unattended silos of how people measure but more

19:09 silos of how people measure but more importantly it’s about can the systems and the tools talk to each other have you created way a plan not just to get the data into tables but to actually have that okay I do I have a customer ID that I can bounce between each tool yes and I agree with that but that’s outside I would that’s outside the scope of like business intelligence right like that is that is platform integration okay right because I would I would argue it would be

19:39 I would I would argue it would be fantastic if all like the core thirdparty systems that you’re using carry the same account ID right that

19:48 carry the same account ID right that would have been great to know back in the day yeah because there are they have we aren’t the only ones using apis we just use them to extract data right the vast major majority of these tools have application or apis that you can interject or insert data into as well right and for this very purpose if I want to if I’m going to create something in your system the reference point for my other ones is this one and yeah you yeah I think that comes to so the API

20:18 know I think that comes to so the API message here so I’m going to maybe change the conversation slightly towards fabric a little bit here but API stuff anything that’s is amazing I API stuff anything that’s is amazing and based on what third party tool mean and based on what third party tool you buy that API mileage may vary right some some tools are going to have a really rich API and we let you get down to all the detail you need other systems don’t have a really rich API and I find the systems that don’t have a very solid API are ones that are built on more I’ll

20:49 API are ones that are built on more I’ll call it Legacy technology they’re not modern web apps they’re not modern infrastructure it’s it again this is my impression it feels more older type web technology it’s a company that has built an application that has a full backend that’s a server it’s not forward thinking like a static web app or or more modern solutions I think more modern solutions require that there are more apis to make the the app performant but that being said fabric I think when

21:21 but that being said fabric I think when you talk about powerbi previously right so before frbc showed up powerbi had a hard time talking to apis in General because you had to do all this complex work inside power query and power query is not has not been the easiest tool in my experience to just go suck out a bunch of data and oh by the way we’re going to return to you this really complex Json object you now need to parse through this thing and flatten it out to something that you can actually understand like it’s it’s just not good at that now however fabric

21:51 not good at that now however fabric shows up notebooks do a much better job of being able to go grab data from apis it’s it’s more code based it’s it’s python it’s there are libraries you can use off the shelf that make it much easier for you to communicate to these apis so I I feel like because RBI was purely data flows semantic models and reports previously what we’ve added in addition to data flows is we’ve added a whole

22:22 to data flows is we’ve added a whole bunch of other really rich data engineering tools now we have things that are like real-time analytics we we have streaming analytics we have re reflex and now we have spark with notebooks that gives again another option to go collect that data so I feel like the API layer of things is getting easier to integrate with and becoming less of a headache because we have new technology now it’s not simple you still the right code so it’s not like a a UI where you click through buttons and

22:52 where you click through buttons and things and even data flows is getting better in general I think they’ve got like an API an open API endpoint feature now but an ooth token request going through data flows not easy I would not recommend it it’s just not a simple experience so another situation here that I think is also evolving with fabric is these thirdparty companies realize Fabric or

23:24 thirdparty companies realize Fabric or Microsoft powerbi is a very large driving force for organizations to get data out and so I’m also seeing while Microsoft is building a richer tool to collect the data other companies are also acknowledging wow our customers ask us for hey I got to go connect to this data in powerbi I want to visualize it outside of our program we should provide a series of apis that are going to make it easier for you to connect to it via powerbi or go copy the data out or collect the data so I think it’s a

23:54 collect the data so I think it’s a double it’s it’s like coming attacking the problem from two directions Microsoft’s doing a better job and other companies are hearing Microsoft is doing a better job and customers are asking for those features so more companies are building into their products here’s a button where you can just go connected to powerbi here’s a button where you can go here’s the instructions on how to connect to that system and get data out of it it’s big point you said too about prefabric and where we are today because even if we had all that

24:24 today because even if we had all that integration with what powerbi was Prem May 20 May 20 23 it lived and if it was finalized in the semantic model or the data set back in also May right so even if I had all this neat Nifty integration the best I could do was create data flows to integrate these tables but it was still going to only live in a semantic model that was the end point to your point with notebooks in libraries we can install by the way you said something

24:54 install by the way you said something that Microsoft if you’re listening please allow custom libraries for power query that sounds incredible like a ppie yeah like having my gosh but we have all these Integrations that have already been available to allow us now to not just have the endpoint B A semantic model but that endpoint B A Lakehouse for all these Integrations utilizing the same knowledge and expand expanded knowledge of notebooks of python of data flows and pipelines and still also power

25:25 flows and pipelines and still also power query if that was my jam I can utilize all this and it doesn’t have to have a end of Route with the semantic model that’s the that’s really where I think data mark is really talking about this becomes more realistic for this to be a 360 view of fabric being that location yeah because you’re going to have just piles of third party specific data for customers that are going to need to be landed somewhere right you’re going to need to investigate what those piles of data

25:55 investigate what those piles of data look like and then figure out okay what do we do to stitch them together right and then so again it’s really it’s just really this true semantic models were like too refined for what we need here or the customer 360 you really need like the data warehouse I just need a place to just land data figure out what to do with it and then shape it into what we want to do and in every project I’ve ever been on nothing ever comes out of

26:25 ever been on nothing ever comes out of those third party tools or Source system correctly the right format so you’re always doing some shaping no matter what have those models in powerbi desktops and those are the ones I did not want to open because like I can’t imagine trying to do that without the Enterprise tooling oh 100% oh man I have stories you got to close every application and that’s that’s that’s where my my head went immediately with this is it’s not like these are solving the customer 360 problem without a landing

26:55 customer 360 problem without a landing Zone prior to model yep all also like and we didn’t talk about this I think also is one of the most important things when you’re doing a customer 360 because like what you’re presenting is if you went straight from those systems to a model that’s current state only ever just like those thirdparty systems and one of the largest thing one of the biggest value ads from a customer 360 perspective is there are tons of data points within here that if you’re

27:25 points within here that if you’re tracking on on a cap captured state which you’re doing with Enterprise tooling where you’re plugging this in and pushing the data somewhere you can you can show Trends you can show what’s happening in those third party systems over time because you’re ingesting the data and that’s massively valuable especially when it comes down to like the opportunity part of it I I will tell you powerbi you can be very creative I have schemas and tables because we wanted to do the customer journey and

27:55 wanted to do the customer journey and they wanted to powerbi I it can it can handle a lot powerbi can do so much more than just obviously we want to follow the best practices but oh there it is there it is obviously we don’t well yes does it provide the flexibility to go way off the reservation uhhuh it can be can be I you stretched it you stretched it Timmy okay yeah Tommy Tommy Timmy what was I gonna say what wait hold on what was I saying what that was a disagreement no you’re not like you you

28:27 disagreement no you’re not like you you you you did things let’s just not talk about we did yeah I am I still am proud of this was 2017 and they want a customer every like the customer Journey every touch Point without actually having those tables but we did it and it worked very slowly but it worked and and these are these in any harb developer’s life you go back and look one year two years three years and you’ll be like what was I doing I’ve learned so much since then

28:58 I doing I’ve learned so much since then and this is where I think you and this is where I think this is why I think it’s very know this is why I think it’s very important for organizations to think about what does their to we’re talking a little bit about data culture here but that really alludes to you really do need a plan to continually invest in your team anyone who’s touching or building or creating with PBI it’s very relevant for you to like continue investing in them through a team get them in front of people who are doing better stuff get them in front of real modeling tool like train and programs

29:28 modeling tool like train and programs because that helps everyone Elevate their their skills and gets better over time I think one that’s one of the opportunities here coming right like so if we’re going from the powerbi bent where like there there are some

29:42 where like there there are some deficiencies in how we connect to extract Andor stage data prior to a model that fabric opens the doors to that so I I I do agree with that stance as far as like taking the view into this into this article and part of that conversation too with fabric is I think no matter what we had with customer Journeys or customer view in a semantic model the life cycle or the lifespan of that model

30:14 life cycle or the lifespan of that model was always going to be short just because again that was the end point and we live in this area now where if I have this higher level view where the data lives farther Upstream in a RBI in a sense interface fabric we’re really allowed more to I don’t want to say control this the flow of data or control the intelligence as much but we’re not at the whims of all the problems we talked about at the beginning of this episode we were talking about the unability to integrate

30:45 talking about the unability to integrate with the data we’re we’re at the we’re at the hands of if the data is already integrated or not like we have that ability Now to control that more and we have have more of the ability to have the conversation what systems are being applied or at least I’m spitballing here but if we have more control Upstream in fabric from the business side not just and the IT side and we know the data that’s coming in we’re trying to connect it we have a better argument to say that

31:18 it we have a better argument to say that hey what systems are you using let’s have a better conversation on how we’re integrating this because we’re not just dealing with a semantic model we’re dealing with something that we’re trying to we’re almost be becoming a bit of the systems integrator here no yes and I and I think your point there I think the business I don’t think the business understands always the challenges of why data won’t just connect together there’s there’s a there’s potentially a naivity around like they’re naive to like well why

31:51 like they’re naive to like well why can’t the customer be found in YouTube versus here versus here Well here here’s why here’s why these different systems and I think until you can really articulate look we have we can we can join the data based on these three characteristics but it’s just a guess right and it also determines like if the customer entered their name correctly or their email correctly or if they use the same email then yes we could track them across these systems but there’s not an identifier that we know for sure that that person is the same person so I think what you find is to your point

32:23 is to your point earlier the the initial semantic model will likely have to change as the business understands more of how the different systems need to integrate so there it’s going to have to evolve and change and shift and this is why fabric makes a lot of sense because it helps with that evolving and shifting you’re able to come back to your system and look at it all the information there and say oh I do need to adjust or change or modify what data we’re pushing into these systems so that it makes more sense and when we tie it back together

32:54 sense and when we tie it back together at the end we can actually design the full circle yeah and if I had a 100 button or a like button on my desk I would be repeatedly hitting it right now because you’re your spot on I I’ll rephrase what I was saying to I think to make a little more sense if we’re already integrating the data in powerbi desktop and from these systems it’s already too late for us to have the argument because it’s already some structured but because we can get more of the raw version of it in Fabric and we have a higher level view or a

33:24 and we have a higher level view or a more Upstream that allows us to have have that have that conversation before it’s too late I agree with that I I semantics words matter I’m that guy sure I I I don’t like the word integrating I like the word probably consolidating or okay I I I get what you mean but like when we talked about the other point there’s a difference between those systems integrating with one another so that they pass the same data together okay when we connect to those s

33:57 together okay when we connect to those s that is not our job yeah what we want to do is make sense of and make meaning out of the data and provide you insights that you can’t see individually in each one of those systems it’s the combined sets of data that make the mo that provide the value and that’s the only point I want to make here is like whatever we call it whether or not we’re integrating like merging transforming those all the data sets consolidating them together on the keys on the things make them work obviously

34:28 on the things make them work obviously there’s going to be like the buckets there’s our 2% which is our perfect data right like 50% is like in a state of somewhat needs repair and the other 48% is irreparably damaged go fix go fix it system right but it there that there’s a lot of that and that’s like that that’s what I’m just saying we integration okay fine yes we we’re integrating all of the data from the

34:58 integrating all of the data from the systems into a platform that we’re we’re trying to enable the business to access that data holistically not individually in each particular area so I’m pulling up my sleeves right now because I’m I’m ready I’m trying to get past it but I just can’t right now Seth we have this amazing integration tool from a business in it that allows us allows us to actually have a platform that we can do the holistic view of

35:29 that we can do the holistic view of integration why wouldn’t that be our job if that’s not available if we’re always blamed that the data doesn’t connect and we can’t get this analytical solution shouldn’t be what do you mean the data doesn’t connect so to Mike’s point where it’s like why can’t we connect YouTube and my CRM system I I need to see my YouTube views per cost to to do see you need to see the YouTube views per cost okay sure whatever the metric is if that’s coming from different systems and we now actually have that

36:00 systems and we now actually have that capability this was my capability or my responsibility at one point on the bi side to try to make sense of these different systems so they can how do you how do you fix that data we had a we had to create tracking we created the tracking that’s not fixing it that’s identifying where it’s broken I think what Seth was alluding to is more of the as data comes in you’re actually looking at data and you’re pushing it an integration would mean a two-way street between the fabric

36:31 two-way street between the fabric environment and then what is in that other system now it it doesn’t say you can’t do that there’s maybe some more grayer lines here around okay let’s look at Power automate let’s look at other in the notebooks in Python you could potentially push data out to other systems talk to an API and not only just read it in but also push it back out too so okay yeah I do think we can do some level of integration with fabric it’s possible do I see a lot of organizations doing that right now probably not

37:04 doing that right now probably not yet yeah I’ll refrain from the word integrate because I see what you’re saying the two-way street yes what I’m saying is let’s call it connecting where we’re actually having an ID or we’re creating that customer account or whatever that ID is between different systems it’s not living in that tool so it’s not integration but at least in that tool I have to create some between multiple systems of a single customer ID whether it’s one has email the other has some transaction

37:34 the other has some transaction ID or some account ID the other has an customer address none of them have the same single customer ID in all the systems so I need to find a way because they’re all matter they all want that analytical Solutions you’re talking about customer Master you’re talking you about customer Master you’re talking making that customer Master know making that customer Master something in this example in this example which is very common and rather than doing it in the tool where we’re pushing which would the integration pushing a single Master customer ID we have that ability to do

38:05 customer ID we have that ability to do that in fabric now or some semblance of it if it’s not available by our business I don’t want to say we have that responsibility to connect and merge all of these tools together but that allows us the opportunity and I don’t know why we wouldn’t take more of a proactive point of this if we don’t have that already I don’t really I wouldn’t argue too much around that thing as well well I maybe it’s two different

38:37 well well I maybe it’s two different ways to solve it right I I guess if if I’m going for a an endtoend solution and and sometimes it’s not possible sometime like what what I’m saying is if it is imperative for the business to understand that Tommy to your point let’s draw back if if I’m trying to acquire companies try to acquire somebody and

39:08 companies try to acquire somebody and that follow and so that’s my marketing if I’m in an opportunity or an account that’s my CRM and then anything Downstream after that it integration like if it’s important for the business to understand how those things tie together then those two system that store The Source information should at least communicate on a base level with one another that should be the ideal if they don’t which predominantly they

39:35 they don’t which predominantly they won’t correct moving towards a hey here’s why these things need to like we’re g to create a campaign that has this hook in it and then when we create an opportunity we’re going to reference that or we’re going to talk to somebody and we’re going to get that hook and we’ll understand that now we have a connection between this o this new opportunity this account etc etc that’s what I think businesses should engage with in integrating like core parts of the

40:06 integrating like core parts of the system so they aren’t completely siloed yeah the reason I say that is all we do from a business intelligence perspective is consolidate provide Insight we aren’t the system of record so the more we create data architectures or Data Systems that the only place that logic could ever exist or be is in a platform that’s just designed to produce information about your Source

40:37 information about your Source systems I I I don’t I’m not saying like we we can’t do it we shouldn’t do it because there’s a lot of precedents around having to do it but there should also be a way for these Source systems to clean themselves up so that our system like is also then part of it pulling the right stuff and not having to fabricate its own entity for the business what I’m saying yeah so that that’s my only point this so what I I like that point and I think what I’d like to maybe just slightly

41:08 what I’d like to maybe just slightly adjust the conversation here towards is in Mark’s wrapup he talks about things that are in fabric that make it easier for you to do exactly what you’re saying Seth like the system like to your point Tommy right there is going to be problems and disconnects between systems someone will need to go in and figure out what those relationship pieces look like and or we will likely have to produce reporting that says this stuff’s wrong here’s here’s where there’s disconnects so we produce the reporting

41:38 disconnects so we produce the reporting around here’s where the source system has missing inaccurate not right information and then we produce the reports around the observation of that and then go back to that team and say okay here’s here’s rules that we’ve run here’s some information that you’ve provided to us and we able to say these things are wrong please go fix it and so the whole idea here is that the source system is system is always getting cleaner or cleaned periodically and this is anyways there’s

42:09 periodically and this is anyways there’s more things to say on that one for now but in the last portion here Mark starts talking about like there are shortcuts there’s these things called mirroring and fabric where you can have a single data table that lives in a place and you can mirror or copy or do a readon copy of that table your only cop copying the metadata but the table still lives in its original location and I have found personally in the last couple weeks working with direct Lake working with data sets that connect to objects in Lakehouse boy do I really like these

42:43 Lakehouse boy do I really like these newer features that are data let’s call it data virtualization inside fabric wow they’re very that’s the objects within fabric that’s the objects within fabric but but I’m really saying like and if you inside fabric it works really well and it it also other things that the Microsoft’s trying to produce if I think about like Amazon what’s stuff going on there there’s storage accounts that are happening there that you could put data in and you can go read that data with a shortcut meaning I

43:13 read that data with a shortcut meaning I can go look at it read it but I don’t have to consume or copy it over to another Cloud which saves me a ton of time Cosmos DB and SQL Server right these tools can appear as if they are a standard table inside your fabric environment but the data stays where it lives you’re you’re not actually doing anything against it you’re not moving the data out of its source system so one of the challenges I think that the business always faces is how do we get our reporting to look

43:44 is how do we get our reporting to look as close to what’s in the actual system that the data is coming from right I’m looking at well if I have this view in this system why doesn’t it look the same here ex exactly right right and and there’s just always idea of like okay well we’re taking nightly loads we’re copying it over so you may make a change in the source system but doesn’t appear for days later or something along those lines it’s so it’s it’s a it’s again a data culture an educational thing that has to happen but I think Microsoft is really attacking these things and I one

44:14 really attacking these things and I one thing I think is really good here I think it’s genius that Microsoft is looking at like making snowflake an integrated connection as like a shortcut directly into powerbi because then snow can be retained by it and the organization to do whatever they got to do with the snowflake side side of things or Amazon or whatever whatever other teams you’re working with yeah then you just directly connect to it and again I’m not worrying about copying the data I’m just accessing their system and reading the data out of it so I think

44:44 reading the data out of it so I think that’s where I’m really enjoying the features of fabric and I’ve been very impressed with the speed and the ease which these other Integrations like shortcuts are working for for models and one really clear example that I had was we were looking at testing a model and had the opportunity to create a model on top of some tables with tabular editor we’re like oh we’re not sure if that something doesn’t seem right the model doesn’t seem to be correct so we were able to spin up new

45:16 correct so we were able to spin up new workspaces very quickly go in create the workspace and with tabular Editor to we’re able to redeploy that model to a brand new workspace with different configur ations different settings different options but all the original data wasn’t it which was direct L to the original location So within seconds or minutes I had redeployed a brand new model made some changes was able to tweak it adjust it without actually having to have an import and this was a 28 gigabyte model this was

45:47 this was a 28 gigabyte model this was huge so I could let the data lie where it lives and almost direct query over the lake information itself it was very impressive to see what see what it was doing here and we were able to spin up two more models just by playing with things so this whole data virtualization thing I think is underrated I think direct lake is a feature that is underrated right now and more organizations should be focusing on like those parts of fabric shortcuts you like those parts of fabric shortcuts direct Lake these things I

46:18 know direct Lake these things I think will add a lot of value and potentially speed up your whole ecosystem if I don’t have to spend in in the old in the old powerbi world before before fabric your main two measures of calculations came from rendering the front end and the other one was your back end just refreshing the models so every time you get a load on a data model you saw compute through the roof and that’s just how I observed it whatever it is that that’s that’s an

46:48 whatever it is that that’s that’s an expensive operation to do if I can now use other tools or connect to other tools without having to do that virtualize that data all those large spikes in loading information now can disappear the speed in which I update those data those data tables are now dependent on the source system and not what I can do in fabric well it’s going to create a different problem because if if you’re mirroring into a source system and you’re hammering it hard where’s the load of that it’s the source system yes

47:18 load of that it’s the source system yes right so there there is going to be Reliance just in a different place correct I guess I I guess the the key word that I missed when first reading this article was the future because so some of the things you’re there there’s two things I’d I’d like to because outline one is all of the things that we talked about are one of the largest data engineering problems that individuals

47:48 engineering problems that individuals have to to deal with whether whether or not you’re doing those engineering as the bi developer or you have dedicated people doing data engineering tasks where you’re plugging into a source system extracting data then the hard part is replicating what the business logic was from that Source system in your new data set right where am I excited about the potential of mirroring if that is going to solve some of those problems

48:18 of those problems absolutely is it there no right and that’s like if I if I look at things how they’re released so what Mark is implying here in the article is he assumes and that’s a big assumption that Microsoft is going to start plugging into third-party systems okay how many third-party systems did we already just talk about in this conversation yep right that’s a good point plug into third the third party systems look at a table okay that would mean that they’re

48:48 table okay that would mean that they’re like take dynamic CRM that would mean that that thirdparty system would instantiate that view and and their UI view of that table in a mirrored application that I could go plug into plug into now if you’re familiar there are many columns and things that you can create in those systems Salesforce included that will only show in that application when you go connect to that even with like the Salesforce connector right

49:19 like the Salesforce connector right now you you will not see a whole bunch of the dynamically driven measures and calculations and things that Salesforce has now added in as

49:29 that Salesforce has now added in as capabilities into those tables well now we’re we’re assuming that miring is going to pick that up because otherwise I can’t reference that table I have to do the same exact thing I did and miring is only going to get me the object itself at which point I need to do all the Transformations again miring in my mind would be the most powerful thing if it kept the everything the business uses in their views and the business business logic right right in the source location

50:00 logic right right in the source location then I’m 100% on board with this idea but that’s a lot of ifs like in there and that and that’s my only point right because right now mirroring if we’re talking and people aren’t familiar mirroring only connects to SQL databases or Azure SQL right and Cosmos DB because there’s a structure under there yes so those are Microsoft owned those are Enterprise grade those are relationships are are what they are there’s no views or added filtering or things

50:30 no views or added filtering or things like that and I think when we’re talking about this topic mirroring is is the most important part of this mythical future of making this customer 360 thing work really really really really well you bring up shortcuts you bring up all those but that’s after we get the data into the fabric ecosystem that I 100% agree with like keep the data where it’s at in the Forum whatever allow us to connect to it light years ahead right the platform itself love the idea takes

51:01 the platform itself love the idea takes away all of the complex complexity that Enterprise systems have to do today where you have your orchestration tool you have your storage you have your processing you have all this stuff it’s all together but this customer 360 is the initial ETL part of it and and I think you hit the nail on the head because to your point this has to be holistic or it’s nothing if the the does not include those columns or those tables which are usually what the business is relying on or dependent on

51:32 business is relying on or dependent on or critical right then it doesn’t matter if it’s not it doesn’t matter if it’s not part of the mirring because it’s not going to be vital those are those are those columns those Dynamic columns that oh no everything lives through those fields or metrics that we’ve custom created and that is critical again if it doesn’t have that it’s nothing and I’m going through the from the reference point from the mirroring and the integration side of I know we know I

52:03 integration side of I know we know I could close my eyes throw a dart board at a list of companies and go oh you have problem throw the actual dart board or the dart at the dart throw the dart actually what we’re gonna change it up I’m gonna throw the dart at the dart yeah Dart you just put the dart on the wall and you throw the board at that’s that’s a t-shirt is that why never wanted okay that’s why you have a lot of holes in your walls I have a lot of holes yeah not a lot of wins a lot ofs discus yeah but a list of organizations and I could tell you that they have

52:34 and I could tell you that they have problems with sales and marketing data integrating but and it’s to I think sess Point 100% where great in theory the miror is great but it has to be holistic yeah I I think this is going to continue this is part of the reason why we have jobs and still will have jobs for a long period of time in the future cuz there there’s just so many things mechanically to work out with where data is coming from how to get it there where do you store it how how’s it going to blend together these are not going away and for

53:04 these are not going away and for all the really cool things that AI is doing these days I don’t see AI doing this stuff yet I think it’ll be a good supplemental help to a lot of what we’re doing already but it feels like a lot of this I I we’ll have to see where the products go in the future but I don’t think AI is going to solve these problems anytime soon so for that standpoint we still have a good solid job moving forward for at least a little while longer if you do want to see AI in action and working though true the tips

53:35 action and working though true the tips plus themes generator we will throw out a little that’s a is this going to be a new challenge on how many episodes can I throw AI into every conversation so at some point with that we’ll we’ll do a wrap here but we do really encourage you to go check out the new harbi tips plus so themes. power. tips go check it out we now have a new feature on the tool that alls you to go ahead and generate your scrims and then from your scrim you can generate positions of your

54:05 scrim you can generate positions of your visuals on those scrims automatically it’s pretty dang slick we really like it it’s getting the feature is pretty cool it’ll save you a ton of time just using the tool once I think would save you enough time to justify the cost for a full year at that point so anyways with that being said we really appreciate your listenership we thank you very much for for talking with us and talking through this topic today let’s do a quick wrap up here on Final thoughts around this topic I’ll just jump over here to Tommy Tommy what is your final thought around this topic I I

54:35 your final thought around this topic I I think the biggest thing is we’re getting closer and closer and if you haven’t looked at Fabric’s capabilities with from an engineering higher level point of view of we’re no longer living in the semantic model this is another point in that that favor Seth any final thoughts as you wrap up what keen insights your pulling away from this one yeah specifically around this topic I would say just keep your eye on mirroring and and how much how many different sources are what types of connections can be made because from

55:06 connections can be made because from simplifying the data engineering Journey especially in a representative example like the customer 360 the the more investment there is massive Time Savings I I think potentially and keeping Source data that is transformed that is business owned in that location without us having to replicate that or or recreate that in our environment which would be an amazing time saer after that the system’s already almost in place

55:36 the system’s already almost in place right like not almost Fabric’s already there right shortcuts in the ecosystem things are working really well it’s just that part keep an eye on that and don’t like this is one of those lines of like make make sure reality butts up with expectation right mirroring mirroring is key in in how many different data sources they plug into I think from my take away from this one really is talking about the data culture is very important here and I

56:07 culture is very important here and I think we have to realize we cannot force every single business unit to always pick the best tool for the solution for the from the data perspective so it will be be messy as best as you can work on your organization to talk about the data culture what do you want to do with the data and have at least a voice on the decision board around okay we’re going to go pick XYZ new cool tool that has some maybe some really amazing AI in it that’s going to help you do whatever

56:37 that’s going to help you do whatever your business unit job is great we love that I want to encourage that but I always want to have the conversation around okay so what what now after you use the tool how does the data come back and is there any considerations for what organizational data needs to join with data generated and 30 third party tools because I think whenever I’m picking tools either my company or helping other companies decide one of one of the criteria pieces is how well can I

57:08 criteria pieces is how well can I integrate with it how well can I get the data out that’s going to be very important so that’s my my my nugget there around data culture and things with that thank you so much for your time today we appreciate everyone listening we only ask if you found this topic insightful if you found it a little bit more thought-provoking around thinking through what a customer 360 is and maybe some of these challenges resonate with you or if you’re not there yet and you’re trying to pick a third party tool that you’re going to use with fabric potentially listen for these kinds of things we’re telling you this this will happen you’re going to hear

57:38 this will happen you’re going to hear people asking more details and integration steps you’ll be like oh yeah I I know what’s going on here we talked a podcast about that so anyways please share it with somebody else someone else who may find this valuable we really appreciate your sharing of the episodes Tommy where else can you find the podcast you you can find us on Apple Spotify or wherever get your podcast make sure to subscribe and leave a rating it helps us out a ton do you have a question an idea or a topic that you want us to talk about in a future episode head over to powerbi. com

58:24 data platform data which know we already we we already have internally these same problems we we’ve used third party tools that have no data in them that we can get to through powerbi or fabric yet saw my power query just to get the freaking YouTube data I we’ll figure it out generate list and if Tommy didn’t stop just picking Rando tools to use things and we would have had such a better time integrating all the data together so been real out of the box tools YouTube and and Spotify excellent thank you all very much and we’ll see you next time

58:56 much and we’ll see you next time [Music] [Music] n

Thank You

Thanks for listening to Explicit Measures. If you’ve got a topic idea for a future episode, send it in—and if you want more Power BI and Fabric content, subscribe to the PowerBI.tips podcast feed.