Elif Tutuk, Global Head of Product at AtScale

Data-Driven Podcast

Listen to Elif, AtScale’s new VP product, share her passion for BI, and her philosophy for making analytics more consumable with a semantic layer. She talks about her experiences and how the semantic layer will add value to new personas like data scientists.

See All Podcasts Button Arrow
Quote icon

If you want to have increased analytics adoption, you have to build an easy-to-use analytics consumption experience. No matter what type of analytics consumption experience you create, if the user doesn’t understand what data means, then they won’t trust it and they won’t use it. And, I really think that a universal semantic layer is a good solution for that problem.

I actually started my AI / ML journey, maybe five years ago, with augmented intelligence and my view of using ML again, working at the consumption layer. I kept thinking how we can use this awesome technology and methods to kind of increase the analytics adoption. So my goal was to make the analytics creation much easier, and much more predictive, easily with the use of AI and amount. but what is happening right now, there’s a huge community of users, a new set of users, right. Data scientists, and their task is to generate insights for the organizations. And when you think about it, you know, it’s the same goal with analytics. So hence that set of personas should also use the same, you know, layer of, govern data, to be able to build their models.


Dave Mariani: Hi, everyone. Welcome to the AtScale data-driven podcast and today’s special guest is a leaf to talk and leaf, we just announced today is our, the AtScale new global head of products. So at leaf, welcome to the AtScale podcast.

Elif Tutuk: Thank you very much. They are. It’s great to be here

Dave Mariani: And welcome to AtScale. We’re so excited to have you, so in today’s podcast, everyone, we’re going to be talking about Elif and her path here, to add scale. she’s had a, she’s got an amazing background, and, and so I’m really excited, to hear about it and, and to share it with you guys. So, Elif, let’s just start with, first of all, just, you know, where how’d you get to where you are today and talk a little bit about, past present and, and we’ll talk a lot about the future.

Elif Tutuk: Yeah, great. So, yeah, I have almost like two decades of experience in data and analytics, like right out of college. I joined tree every Taylor part of their BI team. And, I was like living and breathing all the challenges with data and then enabling all the users to make data driven decisions. So that was a good experience. just, you know, just really being part of the BI team and creating those data structures for better decisions. and then after a while I said that, well, there should be better ways of doing that. And I decided to join a technology company, providing analytics and I drank like, it was 12 years ago and it was a fun journey. When I joined Qlik, it was a much smaller company, like 300 people. Now it’s around, I think, close to 3,000 and you only had one analytics product. And when I was leaving, we had a full cloud platform with data integration catalog, what AML, automations and analytics. So, yeah, it was a great journey. I learned a lot as a product leader, but more important that they like, again, you know, that 12 years really gave me a good understanding. What are the challenges there for real users to be able to, to, to make data-driven decisions and to be able to take actions based on that.

Dave Mariani: Yeah. It’s, it’s interesting because, you got started in industry sort of being a practitioner, were you using Qlik at the time and, and did the Qlik say, gosh, you’re such a great customer. You want, we want to, we want to have you onboard as a, as a product leader.

Elif Tutuk: Yeah, it was, it was something like that. Yeah. I was, I was using Qlik View at that time. and then during the company, I think you’ll have similar backgrounds. Right. So, you have been at Yahoo and then you actually had the challenges and that’s why, you know, you start integrating this awesome technology that’s scaled.

Dave Mariani: Yeah. You know, it look, it’s when it comes down to, you know, when entrepreneurs always sorta sort of asking, you know, what do you, what do you do Where did the idea come from It’s like, it’s it doesn’t come on your, you know, in the shower and, you know, pops into your head. And it comes from living pain and trying to find a solution and not finding one. and so, you know, that, that, that was sort of, my path to starting at scale. but enough about me. So what’s, what’s super interesting about, about your journey, Elif is that, you, you know, you started at, at the time when the company was pretty small and, you know, and for people who don’t know, Qlik is and Qlik view and Qlik sense, was really an innovative, visualization platform. that was really sort of the first to do this sort of associated, queries and dashboards where you Qlik on one, you know, one element and it automatically creates a filter for another. It was a really intuitive experience. And we actually, we’re a customer at Yahoo, which is where I had some experience with Qlik. So talk to me about like, what happened in terms of that transition from 300 to 3000, what kinds of things changed and how has your role changed during that sort of hypergrowth

Elif Tutuk: Yeah, so I think one of the first challenges to take on was, you know, are like that second generation of, of the analytics consumption. You know, when you think about the whole BI analytics consumption, you know, it started with like all that still needed very needed because, you know, every human brain would like to slice and dice and analyze the data accordingly. but we start seeing more and more the need for self service because, you know, we don’t want to, like, if you want to have increased analytics adoption, you really would like to have an easy to use, analytics consumption, experience. So that’s why I kind of take the challenge on like being the product manager for analytics area and then really understanding the user needs to have a self-service, you know, and all its experience. so that is where we have introduced Qlik Sense and really kind of enabling that self service.

Elif Tutuk: but then like, again, they might come back again. It’s just, the challenge is really, no matter what type of analytics consumption experience you create, if the user doesn’t understand what data means, then they won’t trust it and they won’t use it. So that’s why, like, in the last two years, I started really believing that there has to be a mechanism, easily read organizations, can reflect the business context and the business logic to data. And the way for that is like creating it, digital data to wean off the business. and I think like the market is also realizing that, like that’s why we are seeing more and more, catalog products out there, even the analytics catalog. Now it’s kind of the new trend. And the whole idea is how you can enable the users to gain trust, but the catalog technologies.

Elif Tutuk: Yeah. And they, they, they, they surface the metadata to the user, for, you know, you can understand where the data is coming from. They have a business with clinician, but it is siloed, right This is the catalog, you know, whereas you, when you’re consuming analytics, you’re in your BI tool. So that’s why I really start thinking that, you know, there has to be that middle layer where you can reflect, business definitions on complex data sets, and then just let the users go wild per se, because it’s also governance, to be able to really, you know, gain insights, it’s scale. so like that is really, again, I’m just coming back by, you know, I decided to join AtScale, I’ve seen that challenge. And I think, I really think that, universal semantic layer is a good solution for that problem.

Dave Mariani: Yeah. So, so you, you sort of got, took us there, but, I was going to ask that, so, you know, you know, you were, you know, you had a really important role at clinic, there for 12 years making the, sort of the move to, to get back into a startup like AtScale, and, make that transition had to be a big decision. So can you talk a little bit Elif about just, you know, about why you joined that scale and then how you sort of went through that tissue decision process

Elif Tutuk: Yeah, yeah. it was a big decision, you know, because, you know, you kind of put, I am a person who put my soul into everything that I do, and just creating that community and everything was, was great. But again, just like I’m looking at the technology trends, Dave, like, what is happening right now So, every organization, small or large, like they are, everyone wants to move their data to cloud. And there is a reason for that because, you know, in recent years, the cloud storage and processing power is at scale. so there’s that, that, that shift that is happening. And on top of that, there’s a lot of innovations that happened in the last four, five years. I would say that data management tools like DBT are a great example of that. Even now there’s any persona in Alteryx engineers, like who love to do SQL to get data in a constant consumable, you know, format. So when you combine that, you know, what has happened in the cloud, moment, and then also the data management innovations, on those tools, you know, that’s really created the modern data stack that everyone is talking right now.

Elif Tutuk: So like, that’s one thing that, but again, I keep thinking about, you know, when you look at those data management more than data, data stacks, like, yes, there’s a lot of innovations, but those innovations are focusing on making raw data in all ready, like either is it your house automation or change data capture or DBT, but still nobody’s focusing on the problem of making that analytics ready data business ready is how I refer it because you can create, well, you know, it good start schema, right. that can, you know, it’s a real time, it gets updated with CDC, but if the user still doesn’t understand, you know, what’s the, like what that data, like metric mean. and also what’s the best way, the right way to slice and dice that did the government are now, then they wouldn’t be really making the right decision. And also, you know, the analytics adoption will not be hard. So like, what was, are the technology trends that I really think that, you know, even making the need for semantic layer more obvious.

Dave Mariani: Yeah. I love that. I love, I love how you’ve made a distinction between business ready versus analytics ready. and, been a lot of focus on analytics, ready, not necessarily business ready where that semantic layer really comes in and makes it, you know, I like to talk about, you know, it makes data really consumable by everyone, not just somebody who can write SQL or understands schemas and understands how to connect to, you know, platforms like snowflake or Databricks. so, so when you think about that and you think about where we can go with a semantic layer, at least, what, what kinds of things do you think, where we can continue that sort of trend of innovation to get that data consumable by everyone

Elif Tutuk: Yeah. So what really makes my heartbeat lately, like, is really how we can shift that center of gravity, of making an Altec straight data business ready from the last mile of analytics consumption to the last mile of modern data stack. Like, what I mean by that is today. I know like every users, they, they, they, they do that conversion or they try to do that conversion it, the BI consumption layer. and what happens at the organizations, of course, there are many BI tools and it’s no news, right? So each of those BI tools have their own version of the truth or debate of defining the data. So really my goal is how we can shift that calibration and make it super easy for the data producer and business user to collaborate on the semantic layer in a very agile, like an awesome experience, to create that, business ready data to be consumed by any BI tool. So that’s, Dave, like, you know, right now is, you know, every business unit has their own choice of BI tool. Like when I was at Qlik, I used to say that, you know, every time a user exports data from Qlik to Excel in angel banks, it’s actually,

Dave Mariani: Yeah, I love that a lot. I was just going to, I was just going to ask you about that. That’s like something I love, because it could, because a Excel is deemed by it. And also obviously a visualization vendor Excel is deemed as the enemy, right. It’s like, oh, no. It’s like, it’s why would you ever want to use Excel and get out of our beautiful garden but, but that’s ultimately what people want to do.

Elif Tutuk: Yeah, because again, that goes back to human psychology, perhaps like everyone would like to see the data and seeing the data in a tabular format is what makes sounds so unique. so I don’t need to be, I can’t compete with Excel. It’s just the reality. and just by the way, I also want to give credit, like, it’s our previous CTO, Anthony Dayton, who starts saying that. So, but just, I love the fact that an angel dies and it really

Dave Mariani: Definitely, it definitely makes an emotional connection there. and, and, and, and just for everybody who’s listening, we don’t think Excel is evil. we think that, if you just dump data in Excel, I could see that, and use Excel as a database and it’s ungoverned and an unstructured. Yeah. You know, you can get into trouble. but, with our interface to Excel with that live connection, it is governed and it is live and that you don’t have spreadsheet and spread marts. because each cell in Excel can point back to a cell in the AtScale, you know, virtual model, which all, all that data sits on the cloud data platform, like a snowflake or a Databricks. So you can have your cake and eat it too, and you don’t have to fight the Excel, pull that, that a lot of users have.

Elif Tutuk: Yeah. That blow my mind. They’ve like couple of weeks ago, we were looking at the product together. You were doing a book through for me. And when I saw how in Excel user, a tough user power BI users can use the same set of defined metrics and dimensions all running either snowflake or Databricks or Google BigQuery. It’s just, I said, yeah, that’s it like, because I know, again, coming back to, you know, every business unit, they have their own needs, to use, you know, because finance users, they would like to have that sales specific access. And that’s why they love Excel. No, there is a group of users who love visualizations, who love, you know, Tableau, there’s a set of users who allow, you know, Qlik and others power BI. So that’s the other thing, like, you know, like imagine having, you know, what it will be our tool of choice you’re using and then, you know, any cloud data warehouse, and just being able to access the same set of governed layer of, metrics and dimension definitions is huge, but yeah.

Dave Mariani: Yeah. So, so why do you think it’s taken us so long to get to the point where, we actually start to think of a semantic layer as universal versus, you know, tied to the visualization tool, like, you know, like for 12 years, right. At Qlik, you know, the semantic layer was Qlik, as far as, as, as far as, Qlik and Qlik customers were concerned. So w why do you think it’s it’s, what, what do you think was the sort of impetus to actually get us to start thinking this way

Elif Tutuk: so I think just the world, they’ve like, I know, like, you know, every dashboard users found they want to create a dashboard, they create a data model, but that definitions of, you know, the logic business logic gets stuck with that data dashboard. And when you look at, you know, recent trends and conversations around data fabric and data mesh, people are realizing that yeah, business units, they would like to own the data because they have the domain knowledge. and now, you know, they start having the resources, like in all six engineers to get that data ready in an analytics format and then consume that. so I think that’s why the result of conversations around data fabric and data mesh. but now thinking about the case where every business unit has their own data, domain and data, and there’s a need to be able to combine those data models with the, you know, common dimensions, common attributes, maybe you’re an inventory department looking, you know, analyzing imagery, then there’s the sales department analyzing sales. And just combining those two data domains with a common time dimension is a huge value. So that’s why I think, you know, this is the right time for a semantic layer to shine because for the first time, I think the technology came to the point where those business units are enabled and it’s okay for them to drive their own data domain. And now the need to be able to have an organizational 360 degree view on that data. And that’s why you need the semantic layer more than ever now.

Dave Mariani: Yeah. Yeah. That’s it, that’s it, that’s a great answer because you look it’s, everybody is talking about data mash. It’s a more distributed way of, of, of having the business themselves build and control their data products, but without an architecture to share and define what those data products are without having a common language, you know, it’s just a, it’ll just, just be back to chaos. Again, everybody will be doing their own thing. So it’s great to have a semantic layer and, backed by a shareable data models to be able to be that common thread that allows those two different business units to create their data assets, and also share them with other business units in a cocky, in a, in a common, plugable way.

Elif Tutuk: Yeah. It’s kind of like having the Lego pieces, like it is all government, all searchable. You find a model and you see, you know, what’s the use of it. And then you say you’re finding another model and just the system is smart, knows what’s the best way of connecting them and you connect and you start, you know, your insight driven journey with that data. so I think those are the things like, you know, there’s a saying, there’s a saying in Turkish, and I think it’s the same in English as well, like before the perfect storm, you know, all the ingredients needs to come together. And I think, you know, for its scale and for semantic layer, that has happened right now, like with all the technology and the personas and the needs that we are saying, you know, there’s more need for a universal semantic layer.

Dave Mariani: Yeah. You know, I leave, let’s continue change gears a little bit and just talk about AI and ML. so, so, where do you see things going when it comes to AI and ML and that, that market and how does it affect, or how does it, how does it, integrate or, fit in with the semantic layer What’s your, what’s your, what’s your view on that

Elif Tutuk: so I think, you know, I actually started my AI ML journey, maybe five years ago with the augmented intelligence and my view of using AIML again, working at the consumption layer. Like I kept thinking how we can use this awesome technologies and methods to kind of increase the analytics adoption. So my goal was to make the analytics creation much more easier, and much more predictive, easily with the use of AI and amount. but what is happening right now, there’s a, there’s a huge community of users, new set of users, right. Data scientists, and their task is to generate insights for the organizations. And when you think about it, you know, it’s the same goal with analytics. so hence that set of personas should be also use the same, you know, layer of, govern data, to be able to build their models.

Elif Tutuk: you know, one of the, you know, hot topic in the AI community is, you know, getting data ready for modeling. and that is not a new problem for us. Like in BI and analytics world, we have been saying that, you know, 80% of your time goes to creating that data, making it business ready, and then the cost comes after that. So that’s not different for them. And it just amazes me now that, you know, there’s the BI users, and six consumers and the data scientists, and they shouldn’t be able to use the same semantic layer so that they are generating is based on that digital twin, half the business with the data. so I think again, you know, it’s just the consumer for the semantic layer is increasing even more with that new community of users.

Dave Mariani: Yeah. I love it. So, yeah. So a new persona to basically consume the semantic layer and get value out of it. How, how about a lead how about, applying AI and ML to the actual process of defining that data model, or even augmenting the data model itself, augmenting the semantic layer, where do you see things going, in that direction

Elif Tutuk: Yeah. So, one of the unique things, you know, just again, spending, you know, four or five years just building that type of technology. I had the aha moment, like when you create a model to predict a metric, let’s say sales, right One of the notes of the amount of model, it also tells you the key drivers, or I think the data scientists refer to them as like the features, like, saying that, okay, you know, how did the model come up with that prediction Because these are the key drivers, the key attributes that are affecting that, which is a great information by the way, for, in Alteryx consumer, to be able to know, because one of the other hot topic with AI is, you know, building the trust of the, you know, that black box, it shouldn’t be black box. So that’s why I really think that there’s also a great opportunity is a value for those data scientists community after they create their model, like they do their feature engineering and it’s very time consuming work.

Elif Tutuk: And it is, you know, they do that over and over and over. So what if, you know, those features generated by the featuring becomes part of the semantic layer and they weren’t, and they become reusable and they are not only for the data scientists, but they’re also for the business user because ultimately they are the ones who’s going to consume that prediction. And just knowing that these are the key drivers or how the model made the decision about that fact is there are some future values. So that’s where I see the overlap between the data scientists and the modeling and how it taught is part of the whole semantic layer and what the value is for different types of personas.

Dave Mariani: Yeah. You know, the semantic layer, it could really break down those silos between those two teams because traditionally, you know, the business analysts and the data scientists are sort of, you know, off on their own, doing their own thing. And sometimes they may meet. but the semantic layer can really unify them and allow them to, allow the data scientists to create predictions and features, and then allow the rest of the community to actually leverage them to make decisions. So it’s a really nice, that’s a really nice combination.

Elif Tutuk: Yeah, exactly. So the other area that I see a great innovation opportunity today for semantic layer is to have the concept of lessons or applying context. And I think I’m getting up to speed with its scale and the power of its scale. A couple of days ago, you showed me like how it scale can apply a more governance, like PII type of context so that you don’t see, all of the fields, if you’re another love to see, but that’s really kind of triggered this idea about, okay, what if I made like a, you know, the system knows me that, you know, maybe I’m a marketing analyst, but I just started my job. So it’s my first month. What if there’s a Lance that can be applied on that semantic layer so that, you know, I can progress in my analytics journey. Like there’s so much opportunity when you think about shifting the focus to that middle, like semantic layer and enable in multiple personas to collaborate over there to make an ethics ready, data, business Reddit, but then also like start thinking about more personalized consumption of analytics and how semantically are, can really help for that. So those are the things that Mike and like that make my heart beat.

Dave Mariani: Yeah. I love it. Super exciting, you know, because, you know, because we see every query, you know, coming through the semantic layer, we can, we can, we can, we know what people’s intents are, what they’re interested in. you know, right now we, that information to, to automatically performance enhance that experience. Right So we create aggregates automatically based on that sort of user behavior, but you, right. We can go so much deeper into making it a more personalized experience and making that, making that user smarter, you know, through, you know, through machine learning and, and, and, and through that, that, that intelligence, so an L it could be in the semantic layer, so that regardless of whether they’re using Excel or power BI or a Qlik, or, you know, or, or, or, Tableau, you know, they’re going to be equally smart.

Elif Tutuk: Yeah. Again, just going back, I know other BI vendors are thinking about that, like Netflix, like experience it’s personalized, but again, you know, it’s great. Like, yeah, let’s think about users making things personalized, but that shouldn’t sit on the, you know, a BI layer, that’s fine again. Those are innovation areas that we will be working together.

Dave Mariani: And I love it. I love it. I love it. So, at least, you know, you, you obviously are really successful. you know, you immigrated from Turkey, you established yourself, and, and have been a product leader, both, at Qlik and now at AtScale, started as a practitioner. So what are some of the things you could share with the audience, to, that your, your secrets of your success, to becoming, the, the, the, the analytics product leader that you are today

Elif Tutuk: Oh, I think a day. Thank you. really just, you know, always stay curious and, you know, continue to learn and make sure that you’re doing whatever you love to do. You know, I’ve been always driven by data analytics, you know, since I started, you know, as a user, and I keep, you know, going after that, you know, the ambition that I have, and then I started loving building technologies and innovating and interacting with users. So just know what you love, know what your strengths are and how you can combine them and, you know, continue to stay curious and, you know, major captains after that, you don’t have to think about anything else.

Dave Mariani: Yeah. I love that. That’s great advice for everyone listening, and Leafs. I am so excited and I really sincerely mean this to have a partner and have you here, helping us on achieving that vision of the, of data for everyone. and so, and, and so it’s for everybody listening, watch out, because you’re going to see some amazing things coming out of, of scale, with Leafs leadership. And so I just want to thank you for joining the journey. and I’m so excited to see, you know, what we could do together in the future.

Elif Tutuk: Yeah, the same here, Dave, like, there’s a reason why I fell in love with this technology and kudos to you that you have leave and Brita the problem. And then I love when the technology comes from an actual problem and it’s going to be a great journey. I’m so, so, so happy to have the chance to work with you and the other rest of the product and engineering team at, at scale.

Dave Mariani: Well, thank you, Elif. And thank everybody. Thank you for listening and staying data-driven.

Be Data-Driven At Scale