Hi everyone, welcome to Figuring Out Fabric. We're I asked them questions and you get smart answers. And I've got uh Elena here to help help us figure out how how to navigate uh fabric in the nonprofit kind of space, which I think has its own set of issues. Um Elena is a digital analytics professional, she's worked in a bunch of different fields, so she worked in healthcare, tech, education. She's been uh working in analytics for five years and currently she is working with the National Council on Aging and she's leading the digital analytics strategy. And I think that's interesting because in my limited experience, uh digital analytics is it is a challenge to navigate and understand because you have all these red herrings. So for my example, uh, I helped my old communications professor at Penn State, where I would give his students access to my blogs, Google Analytics, and they were supposed to give me advice. And uh one of them would always have like really smart insights, two or three of them, so like say out of six, two or three of them would have no idea. And then the other half. They would kind of just look at whatever numbers and facts were on the Google Analytics page and just infer a conclusion. So like I'll give you an example. Um one person I remember distinctly said, hey, you seem to have a lot of viewers, readers from India. And so maybe you should have your blog auto-translated into the languages that make sense there. And the thing is, if you know about the actual context, then you know that one, India has a very high English speaking population. That's I I believe kind of the primary language that they use for a lot of government documents and stuff because there's a dozen different native languages. But also they have a very high IT population or tech population. And so it's not necessarily that they were super excited about my blog in particular, but just A billion people live there and a lot of them work in technology. And so it's it's really easy to kind of get distracted. So I think there's some interesting things to talk about. Do you wanna elaborate a little bit more about kind of your background and your experience? yeah. Hi everyone. Hi, Eugen. Thank you so much for having me on this podcast. Super excited about our conversation. Uh I mean for what you shared. Thank you so much again for talking about my background a little bit. yes, a big a big part of my experience definitely touches digital analytics. So it's mainly web analytics using as you said, Google logics for, but also AWS analytics. So I'm also working a lot with AWS analytics. The market Analytics, uh social media analytics, then we also use and since I'm working in a non-profit in the past year and a half, as I said, National Council and Agent, uh, it's also lots of fundraising data. So this is comes from different different platforms, email analytics, so lots of lots of uh digital sources, and we are trying to unite user journeys. So Google X4 definitely touches different channels, right? But it's still it's still a pretty limited user journey. Journey and also Google Logics 4 definitely does not capture all the data because of all the cookies, right, and the uh browser restrictions. Right, there are actually lots of missing data, lots of mess and key data, and that's the way we are trying to connect uh different data sources, bridge them together, uh to have that view into a unified user journey. Yeah, so just to kind of reflect back what I'm hearing, it sounds like one of the big challenges compared to say a manufacturer or a consulting company or something like that, is that uh you have a lot more sources of data and a lot more inferences that you have to make. Whereas, you know, uh say if I was doing reporting for a consulting firm or something like that. some of those numbers would be a lot firmer because I would know here's my costs, here's my um, you know, how much money we made on each project, all this kind of stuff. And a nonprofit is still to a degree a business and you still have that. But you've got a lot more of trying to figure out, okay, how do people get connected, involved and what's this full life cycle? So it it sounds like it's it's a lot more spread out and insubstantial and you have to do more work kind of separating the wheat from the chaff with all that stuff. Exactly, yeah. I mean if you're talking about like digital user journey, right? So then you find like different like drop off places, right? Helping technical helping technical users to move to a final destination, right? For us, for a non profit, it could be different actions. For example, helping older adults to apply for a key benefit program. So filling out all those like questionnaire forms, right? Like putting the zip code, trying to understand like what what are the pain points. right on their on their journey towards lead l reaching that final final action let's say reaching applying to the key benefit program reaching some key resource but also we are doing again like fundraising right like you know assisting donors in their donor journey uh trying to help uh our partners again to identify those like key key benefits on the website so those are some of the things regarding user journey yes Yeah, no, I mean that makes sense. And so you from what I understand, you decided to do a pilot with fabric instead of just jumping in. And I think a lot of times uh when people talk about fabric, it's just presented as if like, okay, this is the obvious kind of conclusion because we're already maybe we're already using Power B I or already using Office, so let's use their thing. and I think sometimes that can be really frustrating. Mm-hmm. Because fabric so much feels less like going to IKEA and getting some furniture just needs a little bit of assembling and much more like going to Lowe's or Home Depot where it's like, all right, we've got twenty different types of lumber and all these nails and go figure it out. So what did what what factors caused you to decide to do a pilot and what did that process kind of look like? Yeah, so initially when I was hired even my role, so digital analyst, right, about a year and a half ago, I feel like so before me they never had a even a dedicated like data professional for my division. I feel like before me they just had different like marketing people and everyone was trying to do like different like you know parts of analytics, right? But there was no not a single person and when I joined data estate wasn't not a good place, so there was lots Of inaccurate tracking, right? Like the data even was tracked like incorrect. There was no like proper data standards, very limited data governance. So in the older dashboards, they were constantly breaking because again they didn't have reliable data pipelines. So lots of lots of issues, and I feel like even in my data in my description, right, of my job description, I was mainly tasked to solve like those like issues on the surface, right? Like just like I needed to fix the tracking, right? Like through Google Tech. Manager does define like proper uh tagging strategies, right? I had to maintain dashboards. So and I feel like first like half a year when I was just trying to understand, you know, like what the company is about, right? Like what is they're doing, and yes, I mean I did I did uh implement those strategies, but I feel like so many of those issues will still reoccur because I didn't have a proper data uh architecture, right? Like under all those different data sources they were doing. And it's not just what I mentioned, there's so many others. We have we have so many. platforms so many platforms we got for free because we're non-profits so we we try yes yeah so as uh as a non-profit they do have uh lots of technical capabilities and in terms of resources and eagerness to innovation and like using new technologies they're pretty open which makes me at least very happy because you know they they're open to innovate and try new things and we do have a robust set of technologies but again there was never Proper data architecture. That's why no matter how many times I would fix those dashboards, you know, because the architecture is not set up properly, uh, the issues would reoccur. And last year, so last summer, the organization finally decided to standardize a single BI tool. Because before we would also use different BI tools, we would use uh Tableau, we would use Looker Studio, we would some people would use Power BI. So they finally made a decision, they did a study. They did a uh data um what was it? data maturity study in the organization, right? And they decided to standardize a single BI tool, which was Power BI. So when I started working with Power BI last summer, I also realized I saw it's connected to fabric. And once I started learning more about fabric, right, I realized that this is actually can help us to solve so many fundamental data issues and just uh build a proper data architecture uh foundation. And honestly. Initially, I was I mean I my vision was like okay this will solve you know like all the data problem in the organization right because again like I'm an analyst from my department more like a digital department but we have so many other departments we have a research department right that they have their own data we have department which kind of connects to uh like federal partners and they're doing like advocacy, so different different various departments with different data. So my vision was like this can be a unified place to solve all the data problems with in in the organization. But of course to rule them all kind of thing. Yeah, for sure. So but again uh because we don't have that uh first of all like technical tal many technical talent in the organization and the cost is a big thing. Of course, you know they uh wanted to make sure that it justified, right, before like rolling out, before committing to expensive model, right? And rolling out within the organization. so we wanted to commit on More like a limited cost and coming to more like a scope-bounded pilot. So and scope-bounded pilot which will have a sp which will have which will be uh going on for a year, which will have a specific use cases, so it will be so we'll be able to measure uh the outcome, right? And if the cost is justified. And now it's been about seven-eight months since we're in the pilot and it's going great. We definitely learn a lot, but I can also see that other departments. Seeing this pilot's work and we already achieved some results. Other departments now starting to create their own pilots. So I feel like soon we'll have like pilots, you know, within different departments, right? And then finally, hopefully, we will be able to uh commit for the whole organization with fabric. So the the decision to start to assess the current landscape and then consolidate, did that come more from bottom up, like folks like you being like, Hey, this is a bunch of problems, or was it just a top down decision where someone at the C suite decided, Hey, we need to like consolidate? Like where did that kind of come from in your instance? Yeah, it's a good question. So the study was done before I joined, but I know it was definitely a vision of the leadership. They knew that they have so much data and they knew that they d they have to uh improve, you know, the way uh like the data is managed, the governed and uh also how people access data. So and the study was done by IT leadership. So they were the one who did the survey, they did the focus groups and then they consolidated the fundings and they made Decisions based on those uh fundings. Decisions were the standardized BI tool. They also roll out uh like a more like Like a data stewardship program, so where each department will have a data analyst and they will be in charge of doing the data cataloging. Because the the issue was also that many people in the organization they had no idea about data from other departments. And the data could still be useful to their work, but they just had no idea what those metrics means, you know, the definitions. So those were some of the decisions we made based on that study. Yeah, it's interesting to hear the in your instance, you're you're you're trying to do this kind of I don't want to say necessarily hub and spoke kind of model, but you've got this situation where you're trying to create some centralization because you want to have a reliable set of data, but because you don't necessarily have a dedicated data team like a large for-profit organization might, you're having to kind of move the data and the management, the stewardship closer to the stakeholders. Um, and so that's that's definitely something interesting because it's not something I've heard a lot um about before. But it makes it makes sense, right? Like there's gonna be certain people that are most invested in having the numbers be right. Yeah. Um so what I I mean, um so what were some of the particular challenges with the the pilot or some of the difficult kind of integrations or things that you had to to deal with with getting that off the ground. Yeah, I mean first of all like a learning curve of course. So simultaneously while started working on fabric, I've decided to uh take that exam, DP six hundred analytics engineer. Uh because for me uh as I was kind of the initiator of this project, right? So I was also uh like I was leading it, but uh and I had like limited technical support. So in my team we have a software engineer uh who again he didn't have Direct experience with fabric, but he has you know experience with like data integration, like similar platforms. So he was pretty helpful. And we had IT leadership as well to like kind of help me with governance, help me with pricing, understanding the pricing, how it works, right? Like understanding the all those kind of components. So we encountered several issues uh initially with data integration with some of the sources, not being Incompatible and it's like things maybe didn't go as smoothly as you would imagine when we just started using fabric. Uh, at some point when we uh stopped using fabric trial because for 60 for the first 60 days we use fabric trial because we wanted to understand like what what pricey model we should commit to, and we tried to get to support to fabric support to kind of like get the consultation from them to understand better. what would be the ideal pricing for us because again we want it to be price uh effective, right? It's a non profit. Yeah. Exactly. you're going to your donors and you're telling them, Hey, for every dollar that you give us, eighty cents or seventy-five cents or ninety cents is gonna go to help people in need. And then there's a certain amount that we're gonna need for administration. And so one of the I mean, to my understanding, one of the core metrics of a success of a nonprofit is how many cents out of the dollar donated gets to the people that are in need. So yeah, it's it's a big deal. It's not you know, it's not Fortune five hundred we're gonna be like, Yeah, let's just run an F one twenty eight, it'll be fine. Yeah. Exactly, exactly. And for us, honestly, like our funding comes like ninety percent from the government and I would say like eighty per s eighty percent is corporate partners, smaller percent from individual actual giving, what you described. But for all these funding sources we are doing an extent reporting, right? As you said, like we have to report like what how those money are used, how they're contributing to all the, you know, initiatives we're doing for our audience. So pricing was definitely a big uh Deal for us, and it was very complicated because initially we use the trial, and trial gives you uh one of the biggest capacities, 64 computer inching capacity. So, and I was trying to do those like initial tasks, right, while still learning and monitoring on that app. What is it? Uh capacity metrics. Yeah. So I was just trying to understand, yeah. It's not that yeah, it's not the easiest to read, unfortunately. And it's not always easy to just look at the percentage and be like, okay, well, we're only using an eighth of this, so we only need an F eight, because you have to consider like peak utilization and all this stuff. So it's not it's not as easy as it maybe could be. And the other issue, I get why they do this, but the sizing is in powers of two. Mm-hmm. And so if you need just a little bit more than an F eight, you're paying twice as much for an F sixteen in most cases. There's some edge cases where like spark billing and stuff, but that probably wouldn't apply, you know, org of your kind of type. So it's it's challenging. And I just wanna, you know, reiterate, like it's really, really frustrating and overwhelming when you're like you're going through the cert, you're trying to learn what is all this stuff. And you're expected in sixty days to figure out how much money the organization should be spending for this thing. So like I if I was in your shoes I would have felt overwhelmed personally. So No, that's how I followed you. I mean, I was trying to make my best decision, but again, when you're just starting using the platform, right, in your first two months and you don't have much support, it's very hard to estimate and understand like how the the whole year will look like, right? Like what what uh features we will end up using, right, and what it will drive the capacity. So I feel like and also the model the the the regarding discounts, right? So from us, from our experience, we didn't At a specific discount, but they do have like this policy when you commit to a year, you get forty perc up to forty percent off, right? And then like if let's say that uh capacity is not enough, which you committed to a year, you can pay uh pay as you go, but then you won't get a discount. Yeah, exactly. Yeah. Yeah. So doing a reserved instance has a lot of benefits and it's kind of the normal way to do it. But you're taking a guess and they've they've made it so you can go a little you can go above that, but again, like you said, you're not getting the the discount. So it's not as dire as it sounds sometimes, but it's still it's a real operational challenge. And it's not it's not like Power BI in that it's great for being able to do a pilot. Like Power B I was awesome for doing a pilot because you're buying a license per person. And so you can start with five people, try it out, develop the internal support with the organization, just demonstrate the value. So with fabric, it's a lot harder to do that. Um and there's a lot more pieces, right? Like um, you know, I with with Power BI it's like making a cake. I just need the ingredients and I make it, it puts on icing and the show, okay, I made a cake. Yeah. fabric to demonstrate value, you need to like buy a whole whole oven and the pans and all this stuff. And so there's more layers you have to implement for a vertical slice to show, okay, we can do this and it's actually a returning value. So it's it's a harder challenge for sure. Exactly. I love your example of the oven. right. Yeah. Yeah, totally. And I feel like what we end up like commission to so for now we are still pay as you go. I think we still have yeah, so initially what we end up commission to it was also not enough. So because I was thinking it's not but then we had to increase and pay what you go like. So we actually initially committed to eight, but then we after a few months it was not enough. So we had to increase to sixteen, as you also mentioned, right? Like there is no like eight oh Soon probably we'll be commission to a like a yearly commitment. IT is still like hesitant if they if they're ready, although it will be cheaper at this point. Um but that's our situation for now. Yeah, and I think those numbers are helpful to people because it's very hard to find like specific case study numbers and all that kind of stuff. Um, so you had mentioned uh that you were going through the cert. How how was that? Because I've heard good things about the fabric certifications, but in my experience with older Microsoft certs, like I'm dating myself, SQL Server twenty twelve or SharePoint twenty sixteen, um there they tend to function a lot of times as a marketing device. And so what I mean is, um, as a fabric example, I've heard there's a lot of like KQL and real time streaming kind of pieces, which in your case, you're probably never gonna need, right? Like I have to imagine you're you know, you're if you're trying to do this user journey, you don't need like a Klaxon siren going off because someone clicked donate fifteen minutes ago. Like, you know, you're looking at daily kind of Yeah. and there's a lot of pieces to fabric. So, you know, how how helpful did you find the cert? How was the experience of of going through that for you? I mean honestly I liked it. I think for someone who had a prior Power B I experience, like that certification was kind of a good breach, you know, like from Power B I it touches it touches power BI a lot. It touches data transformations, you know, semantic models, right? So it was a good breach to understand like how fabric, you know, like how how power BI is like under that umbrella of fabric, right? How it's all connected. Just to have a like a bigger, broader picture and uh Honestly, some of like I got familiar with other features of fabric and they helped me also in my other projects in fabrics, not also about like unified user journey. So we've recently started doing this project with charm prediction models for donor retention. And I feel like that certification, like I was able to learn a lot about the notebooks, right? And uh like other like functionalities of fabric which kind of helped me to do that project with uh charm prediction models. Yeah, no, I'm I'm glad to hear that. And I I definitely agree that a lot of times with fabric, it's just dealing with like the unknown unknowns. Like I just if you can just understand like what is out there and be aware of something, then a lot of times you can do kind of the just-in-time learning that you need. And I do recommend for people who feel lost or just getting started, even if you don't want to pay the $150 or whatever it is to test out for a cert, the objective list and the practice. exams that they have are still a nice way to just like start with your own study guide. And then you can pick and choose or augment however you see kind of fit. But, you know, if you have no idea where to start, those those exam objectives and the practice questions and all that kind of stuff are a good place to to get going with that. So I mean that makes sense. I'm glad to hear it was good worked out well for you. Yeah, thank you. And I think also like companies, I mean, care a lot about like, you know, getting people certified just to have that proof, you know, that like you're certified to use this platform. Right? So I feel like when I got my certification, my leadership was, I don't know, they saw me more trustworthy of like leading this fabric initiative. All the gas I still have a pretty limited experience and it's just one single certification, but it still added up some credibility. So I feel like for people, you know, when they will be uh like they would want to do more things in their own company with fabric or looking for another job. I mean certifications leadership still looks at them. So it's just a good credential to have. Yeah, it definitely can help. I think it's it's so interesting because I don't know how things are today, but earlier in my career it was kind of a mixed bag in IT because in IT, like experience is so, so highly valued. And so in that space, in some ways it was better to get like spend a year working on a help desk and get your foot in the door than necessarily have like all these certifications. But I think there's other situations like what you talked about where leadership wants to know that if they put you in charge of this program or project, that you can kind of run with it. Especially for someone like you where you're coming from a particular business domain with digital analytics and that kind of stuff and you're extending outwards into some area that I mean fabric's not fabric's new, you know, it it's only been around for three years and the first year, year and a half were pretty pretty bumpy. So I mean, I think that makes a lot of sense. Um, as we're kind of getting near the end, are there any other kind of tips or particular insights you want to share, whether it's about like how some of this experience related to non profit or just anything that was particularly surprising with with all this for you? No, I mean like Using fabric versus let's say using just Power BI, like I found so many useful features. Like even even like let's say shortcuts, right? Like for us, uh before we would sometimes use let's say API connectors to like some newly data sources which you would add to Amazon S three, right? So let's say shortcut was a very I mean, a very like quick way to connect to data sources in in a virtual way, right? And get access to like a real time data. So uh definitely was interesting to see like all those interesting features and uh for me personally uh I see Fabric is a great like learning environment to understand more what I can do in my data analyst role and actually how it fits like where my role is actually going in the future. So because right now like with the I the goal also of this pilot is to technically prepare organization data for like AI analytics, right? Like AI readiness and build like a proper data foundation right with data governance and uh the the making sure we have a chart forced a data foundation so technically working with fabric is just a great learning environment to understand like all those data components like you know like behind the AI right so understand like data architecture how it all functions data governance because that's exactly where our role of data analyst is moving uh AI can already do many so many data analyst tasks right they can do data carrying they can do data transformations uh building building dashboards so uh the future i feel like the future of our role would be more like in a data architecture and working with those data agents right like working with the i tools like understanding the data understanding the full data journey right from data source to actually providing business value so for me it's been a great experience to just expand my capabilities while also assisting my company in bringing bringing lots of value. That's I'm I'm honestly really happy to hear that. and yeah, I I think you know you and I met at Power B I DC Days and you were talking about, you know, getting your model ready for AI. And one of the things I appreciate is that so much of the talk was just about governance and just the shape of your model and all that kind of stuff. And I think that's gonna be a thing that people underestimate just over and over again, where there so much of actually using AI with stuff is not the fancy latest models, but it's all these things that we've been told to do and probably haven't. Um there's a couple articles I'll put in the show notes. So Anthropic put out an article about um their experience with setting up data models uh for AI and then um some of my coworkers at Tabular Editor have some as well about kind of AI readiness and and all those sorts of things. But a lot of it comes back to like Is your stuff named well? Is it documented? And then in certain cases, you're adding in, say, if you're using Fabric or Copilot, those AI instructions or verified answers and all that kind of stuff. But that's kind of like the cherry on top in a lot of these cases. And so yeah, I'm glad that Microsoft was able to give kind of a smooth funnel outwards from Power BI to all these different pieces of the data platform with fabric. So that's that's awesome to hear. I a thing. so let's go ahead and wrap up with our our usual questions. Um so the first two questions is what is your most favorite and least favorite fabric feature? for the most favorite. What's maybe you know like you mentioned shortcuts, like what what is something that's just been really useful or you're just it's just a pleasure to work in or something like that? Mean honestly like the one leg concept, just like to have like all your data in one place, I think it's amazing. That solves my primary issue with silent data. With like, you know, seeing all those data sources like separately, right? Like now I can see them like all in one place, I can do data transformations, I can finally create like unified metrics, right? Because before people would see like the same metrics and different data sources and they would be like, Why they look different, right? Like why they don't match? Because there is like so much going on in data collection. Right. And uh it's it's lots of bad bad background uh like information, but now like in one single source I can do like all the data transformations, like and I can like define one single metric, right? One single source of truth. So I think that just for me the the main the main use of fabric, the single source of truth and the one lay concept. Yeah I'm happy it it's nice to hear that from someone who's had to implement it.'Cause that was always like the sales pitch at the beginning, but you know, you always wonder how often people manage to get there. So that's very cool. And then was you know, was there anything that you felt was kind of a paper cut with fabric or a feature that just really didn't quite stick the landing for you? so the pricing was definitely very challenging for us since we talked about it before. Also um the when we were buying the pricing we also had an issue with Committing to a region. So all our data sources were in East region, right? Uh we actually when we wanted to commit to buy fabric in the East Region, we actually couldn't. The East region was not available. And again, we we were going back and forth with support of fabric, but they couldn't assist us. So we ended up um buying actually capacity in the central region because that was available. And uh as a result we lost, we couldn't transfer all our work in east region. So what we already built, some of the integrations we already made, we couldn't transfer them to the central region. So uh we ended up rebuilding everything from scratch. So that was just frustrating experience that in those early uh early months we were working this fabric. Yeah, no, I can I can definitely imagine. Very cool. So the last question is um, you know, where can people find you uh online? Is there anything you wanna pitch or promote or or talk about? Uh I mean people can find me on LinkedIn, uh Olena Kulish. So I go my short version of name name is Lena, but LinkedIn is goes Olena Kulish. So uh I mean I don't promote anything right now, but uh Yes, but I'm always I'm always very passionate about, you know, like conversation with people, like having coffee chat. Yesterday actually I met with this amazing uh woman. she's actually a data professional in American University here in Washington DC. And we also met at the Power BI Days conference. So she was in my session, and yesterday we actually had an amazing in-person coffee chat. We also talked a lot about the fabric, we talked about Power BI and how they're doing that adoption of uh fabric in the university in a higher ad right uh organization. So it was just yeah, I love those conversations. I love you know like sharing perspectives, learning from people all the time. So, yeah, so please feel free to send me a connection on LinkedIn and feel free. child always super happy to have those conversations. Thank you so much for having me, Eugene.