Scaling Value with Azure Databricks

In this session, you’ll learn how to scale the use of data science and artificial intelligence (AI) for accelerated business results. You will gain insights into high impact use cases and learn why a design led approach helps you achieve a higher success rate to accelerate value enterprise wide.

Watch more Spark + AI sessions here
Try Databricks for free

Video Transcript

Scaling Value with Azure Databricks.

Scalina Value with Azure

My name is Luke Pritchard, I lead our global data practice here at Avanade. – [Susie] And I’m Susie Lira-Gonzalez, and I am part of the data and AI CoE with Avanade, and focus a lot on solutioning architecture. So before we get started, I want to make sure that at the end of this, you don’t forget to rate and review the sessions. It’s really important for the overall summit. So please don’t forget to do that. So let’s jump into it. So today we will be covering a full agenda, ranging from who is Avanade, and how do we help clients go through some of the market context, our approach questions for clients, go through a couple of client examples, and then finish off on where to get started.


So who is Avanade and how do we help clients? So Avanade was founded in 2000 by Microsoft and Accenture. And essentially we bring the best of Microsoft experts to our client solutions. So we also recently launched an official partnership with Databricks in April 2020, to activate our global resources. We have over 80 locations across 24 countries, and this partnership is helping us scale and bring data science and AI to our clients.

So that is a little bit of what we do as a firm, so Avanade, all up, helps our clients accelerate the value of data end to end. So this includes anything from a high level strategy, to incorporating and enabling AI and Data Science at scale, which Spark is a core pillar as it relates to end to end data platform. And so we have clients all over the world, and we’re helping them realize new revenue streams, new business models, enhance customer experiences, reduce stress, on wide range of clients in a wide variety of industries, ranging from healthcare to retail to manufacturing. So anything data related from strategy to scale, we help our clients in that realm.

So with that, I’ll hand it over to our global lead here, who’s a pioneers for Avanade, Luke. – [Luke] Thanks, Susie. Before we get started in some of the cool things that we’re doing with clients, some of provide some market context, some are tied back to our approach here.

Data is the Fuel of the New Economy

It’s a really exciting time working in the market. Data is really the fuel of the new economy. It’s driving everything. Even in today’s current environment, data is very core to everything that’s going on, in terms of, you know, how we look at the world, and how we make decisions, and how we influence the different systems, and applications, and processes. Some of the key areas that we’ve seen as from our customers, is market differentiation, that data is core to that. It also comes into their new business models, new products and services, and then driving better experiences, and processes throughout the org. And using data also, you know, technically does reduce risk and improve compliance. Not as exciting as creating a new business, but data is really core to that in it’s growing. And this is recognized by all the different market research firms, and organizations worldwide.

What our clients use Azure Databricks to achieve

You know, and how does Databricks fit into this, and what are our clients trying to achieve? We see Databricks as core to these turnings core, to what our clients are doing with data. Very often it is being used as a pillar in Azure specifically, for scaling Data Science and AI. Through doing that, new business models are emerging, competitive differentiations emerging, and there’s a freeing of data from legacy systems, that is also being applied to this to even make it richer, as organizations go on this journey.

So our approach here at Avanade, is pretty simple. We have something we call Use Case Based Solutioning. This is an area where we ensure, you know, the technical solutions that we design, and recommend to our clients, aligns to the business strategy, that also fits within their key processes, ties back to the business outcomes that we’re looking to do on a project, and then also considering the end user, and developer experiences, in the ecosystem. When using Data Bricks, this is a key construct. Often… We see too often that organizations or architects, they will focus too much on one particular software, one particular package, or one particular service from a vendor, without looking at it from an end to end perspective. And this really helps us gravitate to make that right selection. It’s very business value focused. So it’s not just about IT, and what they’re trying to achieve, but also business users, and in the top line business goals, how does that align? It’s also a proven approach. We’ve done this lots and lots and lots of times. We’ve forecasted out costs, we’ve come up with technical selection roadmaps to make it flexible. But really at the end of the day, the reason we do this, is we want to drive a technical selection and recommendation with Azure that fits the needs of the business, and the developers in the IT organizations. And as you work on your own projects, this is very much a process that you could take to heart, yourselves as well, and it will be, well, in building aggregates in your organization.

Avanade’s Use Case Based Solutioning Approach

So what is the approach or solution approach in detail? I think number one is really the strategic alignment. You know, its just tied to a cloud transformation, a business transformation, digital transformation, or is there a key initiative going on organizationally? And maybe there’s a business disruption around that. But, you know, what is this project time tying into? What’s the value around that? What’s the success criteria from all the different groups? You know, it’s… a lot of you come from organizations that are heavily matrix, their complex. But how do you get the viewpoints to say, once you’ve gone and use this technology, you’ve done a project and you’re running something. How do you go to those people using the system to say, hey, this was successful. And there’s different points of view. And I think it’s important to capture those points of view early before you get started, and also take it into your design as well, from that perspective. And this ties directly into the business process and the user experience. So understanding the business processes, the organization, where users fit. But then as, you know, you look at the experiences, whether it’s a developer, or maybe a data scientist, a data engineer, or just business user, or a customer that’s using a website. How do you ensure that technical selection is eating the desired experience. And this is a key construct that we work within, to come up with the right technical selection. And from there we go through a functional architecture, and then a deep dive technical data flow that takes some ties to all the efforts below. And we typically advocate let’s do it, and BP approve the value. And then industrialize the business value, and then have a roadmap to continue growing around that. And we have a number of tools that we’ve done, as we’ve done this a number of times with our clients. But this is an approach that you should consider going as you work through the Azure Data Services, and integrating Databricks as a broader part of your data platform.

IT and Business Success Criteria

So what do we mean by IT and business success criteria? I’ve heard a number of different things for success criteria. We’ve actually broken it out into two different pieces. And the main reason is, there’s technical success criteria, that we often hear it is, and we have one… one client said, “We have to support 3000 columns.” That success, if it does that, then the project is good. However, the business on their side of things, they actually looked at it and said, “I need to be able to do “many more clinical trials in a week than I am today. “I need to do 400% more clinical trials.” So they were divergent, in terms of their success. And this is why we captured. But when we were thinking business success criteria, what we have on the screen. These are very similar to what, you know, the businesses were talking to, on what they wanna be able to do, whether it’s reducing risk, or it’s reducing costs, or it’s improving satisfaction, improving efficiencies, or maybe it’s generating new revenue, from that perspective. These are the top line business outcomes that you should be looking for, as you go through your journey, and working through building out your data platform and the use cases that the platform will support.

And also, you know, thinking, you know, we have a lot of experience, so we receive a lot of questions from our clients or on a regular basis, from that perspective.

Common Questions

And you know, there’re four broad categories that we do receive a lot of… They’re common. You know, Cloud Services, Technical Refresh, Data Science and Data Engineering. And then the other thing too is, we get a lot of questions about Spark. Just in general, or who do for Databricks or open source? And I think, from this perspective, this is where really data bricks has a great story around these questions. These questions often come up in a data transformation, or a business transformation journey, or a very large program. It may not seem clear cut, but a lot of these programs, are really focused on data, focused on data science, AI, and these are new concepts. And I think, as we work through with our clients, we recognize these questions, but how can we make these journeys easier as time goes on. And what we found is many of our solutions, Databricks is a good fit to not only, help scale Data Science and Data Engineering and the org, but it’s also to simplify the open source piece around Hadoop, given that it’s in Azure, there’s a great integration from that with element teams, security, automation from that perspective. And then also, you know, the (indistinct), the user experience that they rewrote, so and the ease of using Spark, really helps out and being able to help from a consumption based format perspective.

With that, I’ll go through a couple of client examples here. I think these are always the most interesting things to talk about, is what are our clients doing, and what are we helping them with? From the perspective, and I think everybody’s doing something a little bit different, but every single one, there’s a lesson. There’s lessons to be learned, both good and bad. And I’ll share some of those with you.

Global Pharma Company

The first client I’m gonna talk about is a Global Pharmaceutical Company that really specializes in creating vaccines and medicines, through clinical trials. They have a lot of data. There’s a lot of standard healthcare data. They’re generating a lot of field data, or technology metrics. And they’re very data driven company, from that perspective. They also live in an era where, you know, the full profit, they rely on patents, or they’re relying on innovations research to come up with that (indistinct), you know, to keep their business running so they can continue, you know, innovating from this. And patents only have a certain amount of shelf life. So there’s a big urgency to make sure that these things are going. And with COVID coming out, in the race for vaccine, this really fits, you know… that illustration really fits well as a company they tried to do every day, and they’ve done for years and years and years. And really, we worked with them around, they had a pretty large legacy data estate. And the way they were working was essentially getting the extracts out of that, data ecosystem on prem, with a couple different sources. It took days and days and days to get the data, and then they would model it in some add-on software. Then they would repeat that process again, especially if they had some new cable data pivot that they wanted to go with. From that perspective, it was cumbersome. They saw it as there’s all this new technology coming out, how can we leverage that? And then also there were finding and limitations in the current software setup around, you know, what they could do. And they were really looking for how can we get into the cloud? How can we benefit? And how can we actually get to this top line goal? You know, from the IT perspective, they were really looking for, how do we move to the cloud? How do we get to the cloud? From that perspective. And there was some misalignment around that. But as we went through the use case… the use case based solution approach, became pretty clear there was a clear use case where, the business, sort of clinical research group, they had a pretty fair use case from that perspective.

And from moving to data to the cloud, this was a good win for IT because they had a use case that can move the cloud and business would say, yes, it was, this was a good thing. As part of that, though, you know, there’s a technology gap. So how… you know, this organization wasn’t familiar with Azure Databricks, that they somewhat knew what Hadoop was, that was about, they knew a little bit about Spark. But we introduced Azure Databricks, and it solved a couple different issues for us. Number one, it was notebook based, which most of the researchers, scientists, they were used to. So they were happy with that from a user experience perspective. But more importantly, it integrated securely with their other Azure Data Services in the cloud, at a scale, and they could scale up and down, they could support the loads. I gave them that modern compute where they could do these clinical trial research even faster. This is something that we’ve been on a journey with them, and we got them up and running from that perspective, and they are doing more and more clinical research trials, and hopefully they’re finding out something to help with COVID here, from that perspective.

Financial Services Company

Another one, I’m switching industries on everyone here, is Financial Services Company. That’s… they have a number of customers that use them for banking loans, credit cards, and a full range of financial services. A lot of their customers, though, are just… they’re just using one service, so they may be a banking customer, but they’re not using the credit cards that are offered, or they may be using some other savings account, but they’re not using maybe the checking accounts. But there’s a number of different combinations of financial service products, this firm has. One of the key things that that was coming out of it, that was more of a data problem, but it was also looking through some of the logs was, that an issue credit card application process being abandoned. So you know, number of clients will go in, without a credit card, but they just leave. Part of it was, you know, it was pretty tactical, in that there was some challenges with the website. But what was even more so the challenge was, what was being recommended as the critical product to the person, to the applicant. And so, in this turned into being a data issue, for the most part, an algorithm issue. And so to improve insights into the customers, the data scientists started leveraging using Databricks, to start spending through all the different data coming up with their different modeling. But that, you know, that was one piece of it. The second piece of it is, how can we actually scale this into our website, so we can continue improving the model based on what happens. And so as a result, after modernizing the website, then also modernizing the algorithms for the next best offer, they’ve been able to, you know, increase the revenue greatly, just by fixing this problem in their website, and also making better recommendations to their customers to apply. And they have been seeing a massive increase from that perspective. But the key thing with data bricks, is it seamlessly worked with Azure ML, it seamlessly worked with the security in Azure, and the backup security around that perspective. And it really, really helped on that. And it not only for the biggest science work, but then operationalizing and scaling Data Science. Okay.

UK Insurance Company

Another one that I’m gonna go through is an insurance company. They’re based in UK. A little different challenge that they have.

Insurance companies that provide lots and lots of quotes. They have lots of different policies.

They have different things going through, lots and lots of data. I think when we first started working with… What’s the company they said, we’re a data company, but we really don’t have a lot of data. I think that that was misrepresented, as we talked through. The fact was they were generating lots of new data every single day, at a rate that they didn’t even have insights into, from that perspective.

You know, and they had some robust Data Science. They had a lot of models going through. So they’re pretty advanced on the Data Science side of things. The challenge that they really had was reporting on the results, producing out, you know, what’s the price optimizations, in making that more visible throughout at a scale. And really where Databricks came into, is it fit within the Data Science community of the client very well, but it also fit into the data ecosystem, that have to help scale out those algorithms, and make it more aware to process this data in coming up with the different pieces. And it was really core to data engineering, scaling up Data Science. Without data bricks, this would have been a very challenging and difficult problem to have. I think the other thing with Databricks is, given it’s in Azure, it worked within all their security perspectives, and policies from that. And this turned into being a pretty impactful service that the client put together for their business users to help, you know, improve the efficiencies of their business, and be more competitive in the market. – [Susie] Great, so we heard a lot of great client stories, and what we’re seeing in the market from Luke. And now I wanna talk through how we get started. So what we offer is an Azure Data Studio for clients who are looking to transform their business, and digital transformation. And we have three key pillars for the studio, virtual, portable and physical. As many of you are aware, the reason we’re having this conference virtually is because of the COVID-19 pandemic. So our physical and portable models are probably up to the state our health crisis. But before all this, we did have a studio in Toronto, Canada, and there’s more coming globally. And then we have the portable one that can be used at any of our Digital Studios. But what I wanna highlight here today, and our virtual world is the virtual Azure Data Studio, which helps… where we bring in our clients, and they come in to a virtual session, and we talk through that end to end data transformation for them on Azure. And regardless of whether they are very preliminary, and they don’t know any of the features within Azure too. I have a PLC I wanna scale across the business, we can help them at any part of that journey. And then also in that day in the virtual studio, we can have them talk to Avanade SME’s client specific industries, SME’s as well, to really help them align their business, and technical ambitions as Luke alluded to earlier, that highlights are approach to use case based solutioning. So we really wanna make sure that the stay in the studio gives them immediate value, and a way to to get started.


So as part of our Data Studio, our virtual Data Studio, we have four key workshops, that help our clients in their journey. So it goes from data on Azure ideation. So a client comes in, and they can come in and learn about the benefits of Azure or their data, or really understand ideas for where they can go. And then a natural next step, is to go on to create… Co-creating your data, vision and strategies, by creating a roadmap of what that looks like specific for that client. And then the next one workshop that we offer is, designing your use-case. So really coming in and understanding, okay, we have clients say, we wanna do 100 things, and they’re all super important. So we help them prioritize the most… the highest value use-case, and prioritize that, create detailed design documents, so we can quickly prove value for the business. And then finally, for more mature clients that are already underway, and they have a PLC, and they’re just really looking to scale across the business, we help them leverage that by creating an enterprise scale operating model. So you can see here, these are the four workshops that we offer as part of our virtual Azure Data Studio.

To ultimately improve efficiency, and productivity for our clients so… And this is again tailored back towards this key new partnership that Databricks and Avanade recently launched. So excited to see some folks in the studio. Virtual portable or the real one in Toronto.

Watch more Spark + AI sessions here
Try Databricks for free
« back
About Susana Lira-Gonzalez

Avanade, Inc.

Susie is a Solution Consultant and a Data Strategy SME within Avanade’s Global Data & AI Center of Excellence. She is based out of Denver, CO. Her focus areas include helping clients understand their data strategy, business process optimization, and human centered design to ultimately meet their desired business objectives. Prior to her time with Avanade, Susie worked for Battelle as a Mechanical Engineer in the Chemical, Biological, Radiological, Nuclear, and Explosives (CBRNE) group. Susie earned a Master of Science in Technology Entrepreneurship from the University of Notre Dame and a Bachelor of Science in Mechanical Engineering with a concentration in Leadership from Gonzaga University.

About Luke Pritchard

Avanade, Inc.

Luke Pritchard is the global leader for Avanade’s Intelligent Data Platform business. He is responsible for GTM strategy and execution of Avanade’s Data business that helps enterprises, through Design-Led approaches, lay the foundation for their cloud and digital transformation journeys to infuse AI and data science at scale. Luke joined Avanade in 2010 and has been deeply involved in AI and Data Platform space for 15+ years.