Alter Everything

AlteryxMatt

To help us catch up on data and AI trends, we are joined by Krishnan Venkata, Chief Client Officer at LatentView. Krishnan discusses adopting AI with the fail-fast method, how to approach solving important business problems, and all things trending in the data analytics industry. He also tells us what he's most excited about for the future of data science and the analytics industry.

Panelists

Krishnan Venkata, Chief Client Officer @ LatentView Analytics - LinkedIn
Megan Bowers, Sr. Content Manager @ Alteryx - @MeganBowers, LinkedIn

Topics

Ep 164 (YT thumb).png

Transcript

Episode Transcription

Ep 164 Data Analytics and AI Trends for 2024 and Beyond

[00:00:00] Megan Dibble: Welcome to Alter Everything, a podcast about data science and analytics culture. I'm Megan Bowers, and today I am talking with Krishnan Venkata Chief Client Officer at Latent View. In this episode, we chat about trends in the data industry, adopting AI with the fail fast approach, solving important business problems and more.

Let's get started.

Hi Krishnan. It's great to have you on our show today. Love if you could give a quick introduction to yourself for our listeners.

[00:00:33] Krishnan Venkata: Thanks, Megan for having me. My name is Krishnan Venkata. I'm the Chief Client Officer at Ton View. We are one of the premier analytics consulting firms based out of India and the only analytics firm based out of India that is listed in the stock market.

[00:00:47] Megan Dibble: Very cool. Well, yeah, I'm excited to hear some insights from you being at Layton View. I think you'll have a unique perspective and so we're here today to chat about some. Analytics trends. So I'm curious, what are some key analytics trends that we should be watching for in the remainder of this year?

[00:01:06] Krishnan Venkata: I think one thing that they are seeing quite a lot in the market is Gen ai and a lot of organizations have been speaking about making investments in this improved concepts.

What we are going to see in the second part of the year is a lot of large scale adoption of Gene I in use cases. So. They're going to move from just proof of concepts to actual large scale implementations, and we are gonna see a lot more of that where enterprises are going to derive value out of gene ai, not just bank on the promise.

We are also going to see a shift in the analysts becoming more storytellers than data crunchers, where tools that are already available to be able to do the information gathering, the data crunching. All of this available analysts can actually focus their time in terms of building the storyline for the business into driving decisions and impacts.

And I'm seeing a lot of conversations already pivoting towards that. A lot of work that we are doing is also pivoting towards being more insightful and decision oriented than analysis oriented.

[00:02:09] Megan Dibble: Yeah, that's super interesting. And to touch on the first thing you mentioned about organizations looking to get more value, are you seeing any.

Roadblocks or challenges that companies are facing when they're trying to move from this proof of concept phase into more of a production phase of generative ai and like how are they overcoming those?

[00:02:30] Krishnan Venkata: As with all new technologies, there is resistance in change within the current ecosystem. Our adoption.

There's always a fear saying, is this something that's going to truly benefit? Is the change worth it? The way we are doing the business is currently meeting its current business goals and things like that, right? A lot of the promise of new technology is for the future and not just for the short, the medium, and in the short of medium term, you definitely will have pain of changing the way you work and doing things.

I feel that's the biggest hurdle in terms of taking things from the, I would call the farm to the production where you are seeding a few good ideas and good concepts, but taking them into production and industrial scale takes some doing. So cultural mindset is a big blocker to that. The second part of it, I feel is inertia in itself in businesses saying they've done business in a certain way.

They're very comfortable in doing business in a certain way. Really changing the way to do business becomes pretty hard, right? That's there with a lot of technologies, but things like gene AI are touching every walk of life, and so that becomes a big impediment into getting this. But it is slowly getting there.

Organizations and people are also realizing that this can be an enabler and can actually be a disruptive enabler in things. And so there is a lot more focus into seeing how this can be imbibed into the organizational processes.

[00:03:51] Megan Dibble: That makes a lot of sense. And the other piece that you touched on, the data analysts, moving from data crunchers to storytellers.

Could you elaborate a little bit more on what that looks like or what you know? We have a lot of analysts that listen to our show. What kinds of skills or trainings or things should they be looking out for or ways that they can uplevel in their storytelling to stay ahead of the curve?

[00:04:17] Krishnan Venkata: So I think if you even look about a decade back analysis was a lot of getting, making sure that we are able to get the data from different systems, bring that to a central repository called an analytical data mart, and doing the analysis, building the visualization, building the analytical models from scratch or by hand as they would call it.

And then presenting those analysis out there. And lot of these things used to take weeks and sometimes months to do, and business leaders used to feel it could take a lot of time for us to analyze and bring back to the business. Now, a lot of these platforms be itself serve BI platforms that are available, realtime analytics platforms that are available.

Gen ai, as well as an enabler, has made a lot of this data crunching and analysis building much faster. So what could be done in weeks is now being done in days. What that means is you have information on the what has happened much faster and what the business has doubt pivoted to asking is, okay, now I know what has happened.

Can you tell me why did this happen? Or what are the things that I should do so that I can make sure of what will happen? And now why this has happened becomes a story in terms of you need to identify, drill down, and tell. These were the reasons that this product was not selling in this market. These were the contextual changes that were happening in the market, and these are the things that you need to do to make sure that you meet your forecast or goals.

Now, this part of why it has happened and what will happen becomes the storytelling part of it. So analysts need to move from the idea of just presenting data or analyzing and presenting pivot tables, et cetera, to coming and saying to the business, you need to be focusing on this product or these features.

You need to be doing this or investing in this so that you can get these outcomes. And that is where I feel the analysts need to pivot. And this is all about a mindset change from an analyst to say, how is my analysis going to be used for empowering business decisions? And starting from that, rather than starting from what data do I have and what analysis and insights can I give?

[00:06:21] Megan Dibble: I love that last piece. I think that's really applicable to everyone out there, and it's definitely a mindset shift. We've had a few recent episodes that have touched on data storytelling, and I just feel like it keeps coming back. It's definitely, I guess, a trend of this year and beyond. That those skills are just gonna become increasingly important.

And as well, like you said, focusing on the why and then focusing on the what's next. I imagine there would be some more predictive analysis modeling that analysts could be getting into in the next years with that as well. So I'd also like to hear from you on just how companies should be navigating this AI landscape.

How should they be? Approaching AI adoption in a way that will move them forward and not just perpetuate this hype cycle, you know, just because everybody else is doing it, how do we really adopt AI and use it at companies? What have you been seeing?

[00:07:22] Krishnan Venkata: I think it's important to start with a clear strategy and some defined goals, right?

McKensey had a recent study which said that 70% of AI initiatives start without a clearly defined goal. That's okay. Wow. Because initially you're just trying to try certain things and you're really not sure what will actually stick. So that's okay. But we are now at a state at which we cannot afford that amount of inefficiency in terms of budgets, and we are also in a budget constrained environment.

So I think the idea is to not go just after saying, okay, I need to do gen ai, or I need to do AI ml, or I need to do analytics. The question is, you need to identify what are the places where this business problem could be solved. I. Using a transformative thing like gen ai. So that pivot needs to be, that's so starting with a clear strategy and defined goals is important and with clear timelines so that you can adopt a fail fast approach.

So you say that this is what I would like to accomplish. These are the metrics that I'm going to look for success, otherwise I'm just going to learn from it and keep the learnings and then cut my losses, right? The second part of it is you need to make investments in data quality and infrastructure.

You're only good as the data that you have, and even if this is across the enterprise, you need to make sure that your quality of data, the data governance mechanism needs to be still there and your infrastructure needs to support some of this. You cannot cut investments in infrastructure and expect gene AI to really meet those needs.

So I think investing in data quality and infrastructure, as cliche as it may sound, is extremely important. The third thing I feel is we need to encourage employees to. Start looking into learning more about data and gene ai, right? They need to invest into upskilling themselves to understand what capabilities do AI break, what are the limitations that are there?

So we should be encouraging organizations to learn and employees to learn from the implementations, do research around a tech training of required, et cetera. There should be enough focus within the organization to encourage that, and we call this in late view. We call this. Encourage, enable, and empower.

You need to do all of this, which is you need to encourage employees, you need to enable them to be able to do it and empower them to take decisions around it. And I would say the last part of this would be to focus on a faced approach, which is, you can think big, but you start small, right? You say, this is what I, my end goal is, but start with a small bite-sized chunk.

Do a quick POC, understand what you are learning from that, iterate with the business, and then basically scale fast. I think a lot of businesses think about AI as a big bang approach and that would be very risky and dangerous to think about it in that way because you will make a significant amount of investment without getting the returns.

What we have really found is looking at a very IT trade processing, two sprints, six weeks, come with A POC, go with the feedback, get a lot of the feedback, reiterate, come back and do. This. Works very well in some of these kind of initiatives, which are. What we call the untreaded path, right? You need to iterate with the business or the stakeholders, and that actually helps a lot in making sure that you are looking at the right use cases.

You're showing the business the benefits of the POC and then helping them scale.

[00:10:37] Megan Dibble: Do you have any use cases of that, that iterative approach and that fail fast with clients that you're able to share?

[00:10:44] Krishnan Venkata: Absolutely. We have this platform for innovation that we have built internally within Latent View, which is a JDI enabled platform that goes to the market, takes a lot of the data sources that an external as well as internal and provides trends, the drivers for the trends, what should companies be doing and the entire works around us.

Now this system could take about a year to 18 months to fully implement it full scale. And a lot of times that is too big a bite to actually take. So what we try to do initially is we try to do what we call quick scans or quick discovery exercises where we do a four to six week exercise to take, we'll take a category or a subcategory in your business and say, what if a few modules of this were implemented, what kind of trends you will be seeing?

What the strengths would've helped the organization. What the benefits will be derived. So I create a business case based on that quick discovery, and because it's a four to six week exercise, it's really quick, the business sees the value. Once they're able to see the value, they keep coming back to us and saying, now how do I scale this across the business?

And that gives us a roadmap of succession, and we don't scale from one to hundred, we go from one category to a family of categories, and then to the overall business, and then overall across countries, et cetera. This iterative approach has actually helped us in a lot of cases, even with dashboards. When we built dashboard with clients, we actually start with couple of views.

We take it back to the business user, like love the initial views, they love the insights, and then they come, Hey, but I would also like to see this, or I would like to see that. And we take that as input and build the next iteration. So a typical dashboarding exercise, while it may take six to nine months to actually build the dashboard, will have so many sprints.

The business starts seeing value very quickly, and that is extremely successful with our client engagements.

[00:12:36] Megan Dibble: That's great. That's really interesting. I think having that iteration, the sprint cycles, the quick feedback, I can see how that would be really useful. I also would love to touch back on something you mentioned.

You were talking about the encourage framework that you have with your clients. Would you say it's like data literacy training in a way of enabling more of a data culture? At those companies or what are your thoughts on data literacy and how companies can strive to improve that?

[00:13:06] Krishnan Venkata: So the Encourage, enable, and Empower framework is what we use internally for our analysts inside Latent View as part of our mission statement saying that we feel that innovation comes from the flow when people are working with data, being able to think about this.

I would say that the same pivot is what we apply to the clients as well, which is. Internally as they look for new technologies, you need to encourage people to adopt new technologies. You should be allowing them to fail and say that even the failure is a success because we encourage you to actually try something new.

Right now, the enable part of it is being able to back them. Right now you can encourage them, but enabling is where you're saying you need to attend data literacy programs. You need to attend conferences. You need to be taking trainings that are there certifications that will be required or. Budgets to actually implement certain pilots that are not there as well.

And the third part is empire, where you're saying that we give you the freedom to take and run pilots, which you feel are important, right? We are not going to say, oh, have you thought through all of this? Allow them to freely experiment and do this. If you think about encourage, enable, and empower. Now you may have 10 projects that are done, out of which nine may fail, but you'll learn a lot more.

When the tent becomes a success and it comes from ground up, they are very invested into making that successful and that drives adoption across the organization. I feel that culture is extremely powerful if you're able to do that, and that's what we tell our clients as well, that when they look to pivot and do some transformative things with Gene ai, let's work with the teams, let's work with those employees and encourage them to be part of this journey.

And so they share the success when some of this become successful as well.

[00:14:51] Megan Dibble: It's a lot more empowering when it's really adopted by those teams and made a success internally. I can imagine that is a lot more sticky, adoption wise of Gen ai or even just analytics. I'm curious how you work through with clients like sorting out the hype of AI and bringing AI in just because we feel like we should or we should be doing machine learning from what really drives business value.

[00:15:19] Krishnan Venkata: I think when we work with clients, obviously clients come to us with a place where they may start saying that we want to look at implementing gene AI internally, or they may have specific problems that they're coming to us with. If they're coming with a ladder, then it's a fairly simple thing. They have a business problem.

We are looking to what the business problem is, what are the gaps that are there, and then we come with whether analytics data or a combination of data analytics can actually solve the problem. If the client is coming with bringing a technology and saying, we would love to do a lot more because we are getting pressure from the top management saying we need to adopt more others, is when we start doing a more consultative and iterative approach and we have a consulting organization internally which starts saying with, why don't you list some of the problems that you have in terms of what are your key objectives as an organization?

What are you focusing on right now? What is important? Is productivity, very important? Is growth, fairly important? Is. Rationalizing some of your spend. What, what exactly are those parameters? And then we start diving into the different areas which are there. And then we identify problems and roadmaps to this.

So if they don't start with the business problem, we take their approach to say, we want to implement Gen ai, but we start going to identifying what are those business problems where we could look at Genea. At the nutshell, I would say. If you identify a business problem that is quite important to solve for the business, which really can make an impact, then identifying whether the gen AI or analytics or even simple analysis would help is the next factor.

And we work with the clients around that. So we try to pivot them from going with a technology, shiny AI adoption first to going let's solve business problems. And all of this will make sense there.

[00:17:03] Megan Dibble: Do you find that the business problems often don't need gen AI or have a lot of good problems come up where gen AI is helpful?

[00:17:12] Krishnan Venkata: I would say both ways, right? So there are problems that fundamentally don't need to have Gen AI on iteration one. If foundationally, there are things that need to be set, right? Why would you need to do JEI when you can do some foundational things? I remember a project that we did with a market online marketplace where.

We found out that they were advertising on a social platform like Pinterest, and there was a lot of visitors that were coming from Pinterest into this online marketplaces, and they were not converting. We did the lot of analysis and found out that, okay, you know what? We need to personalize the user journey where they're going from a different marketplace coming into into this online marketplace.

How can we improve the conversion? When we dug deeper, we identified. That the reason that people were not converting is if I looked for Pashmina Shaw in Pinterest and I got a certain link and it said, Hey, this is the online marketplace that hosts it. And I clicked on it. That came to the online marketplace and that link was broken.

So all the feed that was coming from Pinterest was coming to a page not found URL on the online marketplace. So guess what? As a user, he doesn't see this. They just log off. They're not going to research for, oh, I want to look for pash. So the solution for this is not analysis solution is fix the link.

And that did work. So I think at some point, sometime some analysis when we try to do these analysis, et cetera, the good thing is we went with an analytics framework to identify which channels are performing, which channels are not performing, where are the visitor dropouts, how can we improve the personalization and things like that.

But we chanced upon something which was very basic. One of the projects we did with the leading browser ecosystem, and we can't quote them because it's confidential, but there are not too many browser ecosystem, but this is the leading one, is we found that, and they had a very stiff target in terms of how many browser downloads that they wanted to have on launch.

We were analyzing a lot of the unsubscribe, so the uninstalls that were happening on the browser ecosystem, and what we did when we found out that. A lot of markets that were not English speaking, were having a high amount of uninstalls. Now this is a good analysis. And we said, let's find out what's happening.

And then we had a lot of hypothesis around it. When people uninstall the browser there, there is a comment saying, why did you uninstall this? Right? And so we said we take all those comments and analyze it. We did topic modeling around that. And guess what? The reason the browser was getting uninstalled is because it did not give local language support.

And that was funny. Because if I'm in Germany and I have a browser that does not support German, I obviously don't want to be looking at that. I want to have something in my native language and they just start to turn that feature on and that automatically solve the problem. Right? So the beauty of analysis is you may start with an analytics exercise, but you may chance upon certain things that can have core things that you can actually solve.

And you solve them before you go to more complex analysis or complex solutions is what I would say. So businesses can be solved with simple solutions as well.

[00:20:16] Megan Dibble: Yeah, those are some great examples and good examples too, of digging into the why. Obviously you all looked at what was happening and then that next step that we talked about of.

Digging into why it's happening sometimes maybe that does lead you down the path of creating some kind of model or gen AI thing, but then sometimes it leads you down the path of. It's happening 'cause of a code issue or because of a simple business problem. So we've been chatting about trends in analytics, trends in ai, things that your company is doing.

I'd love to hear from you on what makes you excited in the data space for the next year and beyond. What technologies or trends make you excited for the future in data?

[00:20:58] Krishnan Venkata: I think we've seen a lot of excitement around gene ai, but we are very excited about it personally in terms of the impact that it can have across the organization.

We look at this as a three-pronged strategy internally. One is problems that can be solved significantly using gene AI applications, and this was search and discovery. We have a platform called Laser Internally, which does that. We have gene AI based insights and personalization for communication, which we are using in what we call AI pen pal.

So we see solutions where there are problems in the past that can totally be solved with gene AI based solutions. So we are very excited about that. The second part of it that we are seeing is the natural language generation part of gene AI is useful in some of our platforms. So be it match view, which is used for store comparison, be it connected view in supply chain.

All of these platforms that we already have built now have a natural language generation, which we're using for giving insights. So we are now telling the businesses not only what has happened, but our reasons for why this is happening, and this is already there in some of our solutions and platforms.

The third part of JI is the productivity improvement that we are seeing on analyst floor, right? Can we make the analyst faster, smarter, better in doing his job? And we are seeing. Cases in which we are seeing 20 to 40% productivity improvement across different pilot teams that we are working on. Now, again, we are doing POCs on that and trying to find out why those things are there, but we feel that there could be a significant amount of what we call the empowered analyst that we could have because of GNI.

The second thing that we are very excited about is authorizations earlier used to either do product analytics, marketing, analytics, and a bit on other places. We are now seeing that the organizations across the functions are actually looking at embracing analytics, and there is more cross-functional collaboration to look at data and then make insights.

So the sales is not doing it in silos. They're also looking at marketing data. They're looking also at production level data, customer issue data, et cetera. And all of this has been possible because of the amount of infrastructure investments made in the cloud in terms of data warehouses and things like that.

I think that has been a big unlock for us. We are seeing a lot of use cases where data collaboration across different teams are actually helping us. The third part we are very excited about is organization doing cross collaboration across their data sets. So organization A and organization B look at saying that this data that I have on my customer could be a great opportunity for organization B to partner and cross sell or upsell into.

Those are fairly exciting opportunities that we are seeing now coming. I think it's a fairly exciting time to be in the analytics industry right now, and the graduates and even analytics practitioners are finding it each day to be quite innovative and exciting from that perspective.

[00:23:57] Megan Dibble: Definitely. And that's really encouraging to hear that you're seeing organizations where silos are breaking down, where analytics is being adopted across departments, you know, data sharing because it's the cloud.

I think that's really exciting. Just imagining all the things we can accomplish, and if we're more productive too, with the help of generative ai, then yeah.

[00:24:18] Krishnan Venkata: So

[00:24:19] Megan Dibble: who knows what it's gonna look like in the future, but I think that's definitely some hopeful pieces. So yeah, thanks so much for coming on the show, sharing about trends from your perspective, from what your team is seeing.

Really appreciate the insights you shared with us today.

[00:24:34] Krishnan Venkata: Thank you for inviting me.

[00:24:36] Megan Dibble: Thanks for listening. To learn more about top expansion in this episode, head over to our show notes on community.alteryx.com/podcast. See you next time.

Ep 164 Data Analytics and AI Trends for 2024 and Beyond [00:00:00] Megan Dibble: Welcome to Alter Everything, a podcast about data science and analytics culture. I'm Megan Bowers, and today I am talking with Krishnan Venkata Chief Client Officer at Latent View. In this episode, we chat about trends in the data industry, adopting AI with the fail fast approach, solving important business problems and more. Let's get started. Hi Krishnan. It's great to have you on our show today. Love if you could give a quick introduction to yourself for our listeners. [00:00:33] Krishnan Venkata: Thanks, Megan for having me. My name is Krishnan Venkata. I'm the Chief Client Officer at Ton View. We are one of the premier analytics consulting firms based out of India and the only analytics firm based out of India that is listed in the stock market. [00:00:47] Megan Dibble: Very cool. Well, yeah, I'm excited to hear some insights from you being at Layton View. I think you'll have a unique perspective and so we're here today to chat about some. Analytics trends. So I'm curious, what are some key analytics trends that we should be watching for in the remainder of this year? [00:01:06] Krishnan Venkata: I think one thing that they are seeing quite a lot in the market is Gen ai and a lot of organizations have been speaking about making investments in this improved concepts. What we are going to see in the second part of the year is a lot of large scale adoption of Gene I in use cases. So. They're going to move from just proof of concepts to actual large scale implementations, and we are gonna see a lot more of that where enterprises are going to derive value out of gene ai, not just bank on the promise. We are also going to see a shift in the analysts becoming more storytellers than data crunchers, where tools that are already available to be able to do the information gathering, the data crunching. All of this available analysts can actually focus their time in terms of building the storyline for the business into driving decisions and impacts. And I'm seeing a lot of conversations already pivoting towards that. A lot of work that we are doing is also pivoting towards being more insightful and decision oriented than analysis oriented. [00:02:09] Megan Dibble: Yeah, that's super interesting. And to touch on the first thing you mentioned about organizations looking to get more value, are you seeing any. Roadblocks or challenges that companies are facing when they're trying to move from this proof of concept phase into more of a production phase of generative ai and like how are they overcoming those? [00:02:30] Krishnan Venkata: As with all new technologies, there is resistance in change within the current ecosystem. Our adoption. There's always a fear saying, is this something that's going to truly benefit? Is the change worth it? The way we are doing the business is currently meeting its current business goals and things like that, right? A lot of the promise of new technology is for the future and not just for the short, the medium, and in the short of medium term, you definitely will have pain of changing the way you work and doing things. I feel that's the biggest hurdle in terms of taking things from the, I would call the farm to the production where you are seeding a few good ideas and good concepts, but taking them into production and industrial scale takes some doing. So cultural mindset is a big blocker to that. The second part of it, I feel is inertia in itself in businesses saying they've done business in a certain way. They're very comfortable in doing business in a certain way. Really changing the way to do business becomes pretty hard, right? That's there with a lot of technologies, but things like gene AI are touching every walk of life, and so that becomes a big impediment into getting this. But it is slowly getting there. Organizations and people are also realizing that this can be an enabler and can actually be a disruptive enabler in things. And so there is a lot more focus into seeing how this can be imbibed into the organizational processes. [00:03:51] Megan Dibble: That makes a lot of sense. And the other piece that you touched on, the data analysts, moving from data crunchers to storytellers. Could you elaborate a little bit more on what that looks like or what you know? We have a lot of analysts that listen to our show. What kinds of skills or trainings or things should they be looking out for or ways that they can uplevel in their storytelling to stay ahead of the curve? [00:04:17] Krishnan Venkata: So I think if you even look about a decade back analysis was a lot of getting, making sure that we are able to get the data from different systems, bring that to a central repository called an analytical data mart, and doing the analysis, building the visualization, building the analytical models from scratch or by hand as they would call it. And then presenting those analysis out there. And lot of these things used to take weeks and sometimes months to do, and business leaders used to feel it could take a lot of time for us to analyze and bring back to the business. Now, a lot of these platforms be itself serve BI platforms that are available, realtime analytics platforms that are available. Gen ai, as well as an enabler, has made a lot of this data crunching and analysis building much faster. So what could be done in weeks is now being done in days. What that means is you have information on the what has happened much faster and what the business has doubt pivoted to asking is, okay, now I know what has happened. Can you tell me why did this happen? Or what are the things that I should do so that I can make sure of what will happen? And now why this has happened becomes a story in terms of you need to identify, drill down, and tell. These were the reasons that this product was not selling in this market. These were the contextual changes that were happening in the market, and these are the things that you need to do to make sure that you meet your forecast or goals. Now, this part of why it has happened and what will happen becomes the storytelling part of it. So analysts need to move from the idea of just presenting data or analyzing and presenting pivot tables, et cetera, to coming and saying to the business, you need to be focusing on this product or these features. You need to be doing this or investing in this so that you can get these outcomes. And that is where I feel the analysts need to pivot. And this is all about a mindset change from an analyst to say, how is my analysis going to be used for empowering business decisions? And starting from that, rather than starting from what data do I have and what analysis and insights can I give? [00:06:21] Megan Dibble: I love that last piece. I think that's really applicable to everyone out there, and it's definitely a mindset shift. We've had a few recent episodes that have touched on data storytelling, and I just feel like it keeps coming back. It's definitely, I guess, a trend of this year and beyond. That those skills are just gonna become increasingly important. And as well, like you said, focusing on the why and then focusing on the what's next. I imagine there would be some more predictive analysis modeling that analysts could be getting into in the next years with that as well. So I'd also like to hear from you on just how companies should be navigating this AI landscape. How should they be? Approaching AI adoption in a way that will move them forward and not just perpetuate this hype cycle, you know, just because everybody else is doing it, how do we really adopt AI and use it at companies? What have you been seeing? [00:07:22] Krishnan Venkata: I think it's important to start with a clear strategy and some defined goals, right? McKensey had a recent study which said that 70% of AI initiatives start without a clearly defined goal. That's okay. Wow. Because initially you're just trying to try certain things and you're really not sure what will actually stick. So that's okay. But we are now at a state at which we cannot afford that amount of inefficiency in terms of budgets, and we are also in a budget constrained environment. So I think the idea is to not go just after saying, okay, I need to do gen ai, or I need to do AI ml, or I need to do analytics. The question is, you need to identify what are the places where this business problem could be solved. I. Using a transformative thing like gen ai. So that pivot needs to be, that's so starting with a clear strategy and defined goals is important and with clear timelines so that you can adopt a fail fast approach. So you say that this is what I would like to accomplish. These are the metrics that I'm going to look for success, otherwise I'm just going to learn from it and keep the learnings and then cut my losses, right? The second part of it is you need to make investments in data quality and infrastructure. You're only good as the data that you have, and even if this is across the enterprise, you need to make sure that your quality of data, the data governance mechanism needs to be still there and your infrastructure needs to support some of this. You cannot cut investments in infrastructure and expect gene AI to really meet those needs. So I think investing in data quality and infrastructure, as cliche as it may sound, is extremely important. The third thing I feel is we need to encourage employees to. Start looking into learning more about data and gene ai, right? They need to invest into upskilling themselves to understand what capabilities do AI break, what are the limitations that are there? So we should be encouraging organizations to learn and employees to learn from the implementations, do research around a tech training of required, et cetera. There should be enough focus within the organization to encourage that, and we call this in late view. We call this. Encourage, enable, and empower. You need to do all of this, which is you need to encourage employees, you need to enable them to be able to do it and empower them to take decisions around it. And I would say the last part of this would be to focus on a faced approach, which is, you can think big, but you start small, right? You say, this is what I, my end goal is, but start with a small bite-sized chunk. Do a quick POC, understand what you are learning from that, iterate with the business, and then basically scale fast. I think a lot of businesses think about AI as a big bang approach and that would be very risky and dangerous to think about it in that way because you will make a significant amount of investment without getting the returns. What we have really found is looking at a very IT trade processing, two sprints, six weeks, come with A POC, go with the feedback, get a lot of the feedback, reiterate, come back and do. This. Works very well in some of these kind of initiatives, which are. What we call the untreaded path, right? You need to iterate with the business or the stakeholders, and that actually helps a lot in making sure that you are looking at the right use cases. You're showing the business the benefits of the POC and then helping them scale. [00:10:37] Megan Dibble: Do you have any use cases of that, that iterative approach and that fail fast with clients that you're able to share? [00:10:44] Krishnan Venkata: Absolutely. We have this platform for innovation that we have built internally within Latent View, which is a JDI enabled platform that goes to the market, takes a lot of the data sources that an external as well as internal and provides trends, the drivers for the trends, what should companies be doing and the entire works around us. Now this system could take about a year to 18 months to fully implement it full scale. And a lot of times that is too big a bite to actually take. So what we try to do initially is we try to do what we call quick scans or quick discovery exercises where we do a four to six week exercise to take, we'll take a category or a subcategory in your business and say, what if a few modules of this were implemented, what kind of trends you will be seeing? What the strengths would've helped the organization. What the benefits will be derived. So I create a business case based on that quick discovery, and because it's a four to six week exercise, it's really quick, the business sees the value. Once they're able to see the value, they keep coming back to us and saying, now how do I scale this across the business? And that gives us a roadmap of succession, and we don't scale from one to hundred, we go from one category to a family of categories, and then to the overall business, and then overall across countries, et cetera. This iterative approach has actually helped us in a lot of cases, even with dashboards. When we built dashboard with clients, we actually start with couple of views. We take it back to the business user, like love the initial views, they love the insights, and then they come, Hey, but I would also like to see this, or I would like to see that. And we take that as input and build the next iteration. So a typical dashboarding exercise, while it may take six to nine months to actually build the dashboard, will have so many sprints. The business starts seeing value very quickly, and that is extremely successful with our client engagements. [00:12:36] Megan Dibble: That's great. That's really interesting. I think having that iteration, the sprint cycles, the quick feedback, I can see how that would be really useful. I also would love to touch back on something you mentioned. You were talking about the encourage framework that you have with your clients. Would you say it's like data literacy training in a way of enabling more of a data culture? At those companies or what are your thoughts on data literacy and how companies can strive to improve that? [00:13:06] Krishnan Venkata: So the Encourage, enable, and Empower framework is what we use internally for our analysts inside Latent View as part of our mission statement saying that we feel that innovation comes from the flow when people are working with data, being able to think about this. I would say that the same pivot is what we apply to the clients as well, which is. Internally as they look for new technologies, you need to encourage people to adopt new technologies. You should be allowing them to fail and say that even the failure is a success because we encourage you to actually try something new. Right now, the enable part of it is being able to back them. Right now you can encourage them, but enabling is where you're saying you need to attend data literacy programs. You need to attend conferences. You need to be taking trainings that are there certifications that will be required or. Budgets to actually implement certain pilots that are not there as well. And the third part is empire, where you're saying that we give you the freedom to take and run pilots, which you feel are important, right? We are not going to say, oh, have you thought through all of this? Allow them to freely experiment and do this. If you think about encourage, enable, and empower. Now you may have 10 projects that are done, out of which nine may fail, but you'll learn a lot more. When the tent becomes a success and it comes from ground up, they are very invested into making that successful and that drives adoption across the organization. I feel that culture is extremely powerful if you're able to do that, and that's what we tell our clients as well, that when they look to pivot and do some transformative things with Gene ai, let's work with the teams, let's work with those employees and encourage them to be part of this journey. And so they share the success when some of this become successful as well. [00:14:51] Megan Dibble: It's a lot more empowering when it's really adopted by those teams and made a success internally. I can imagine that is a lot more sticky, adoption wise of Gen ai or even just analytics. I'm curious how you work through with clients like sorting out the hype of AI and bringing AI in just because we feel like we should or we should be doing machine learning from what really drives business value. [00:15:19] Krishnan Venkata: I think when we work with clients, obviously clients come to us with a place where they may start saying that we want to look at implementing gene AI internally, or they may have specific problems that they're coming to us with. If they're coming with a ladder, then it's a fairly simple thing. They have a business problem. We are looking to what the business problem is, what are the gaps that are there, and then we come with whether analytics data or a combination of data analytics can actually solve the problem. If the client is coming with bringing a technology and saying, we would love to do a lot more because we are getting pressure from the top management saying we need to adopt more others, is when we start doing a more consultative and iterative approach and we have a consulting organization internally which starts saying with, why don't you list some of the problems that you have in terms of what are your key objectives as an organization? What are you focusing on right now? What is important? Is productivity, very important? Is growth, fairly important? Is. Rationalizing some of your spend. What, what exactly are those parameters? And then we start diving into the different areas which are there. And then we identify problems and roadmaps to this. So if they don't start with the business problem, we take their approach to say, we want to implement Gen ai, but we start going to identifying what are those business problems where we could look at Genea. At the nutshell, I would say. If you identify a business problem that is quite important to solve for the business, which really can make an impact, then identifying whether the gen AI or analytics or even simple analysis would help is the next factor. And we work with the clients around that. So we try to pivot them from going with a technology, shiny AI adoption first to going let's solve business problems. And all of this will make sense there. [00:17:03] Megan Dibble: Do you find that the business problems often don't need gen AI or have a lot of good problems come up where gen AI is helpful? [00:17:12] Krishnan Venkata: I would say both ways, right? So there are problems that fundamentally don't need to have Gen AI on iteration one. If foundationally, there are things that need to be set, right? Why would you need to do JEI when you can do some foundational things? I remember a project that we did with a market online marketplace where. We found out that they were advertising on a social platform like Pinterest, and there was a lot of visitors that were coming from Pinterest into this online marketplaces, and they were not converting. We did the lot of analysis and found out that, okay, you know what? We need to personalize the user journey where they're going from a different marketplace coming into into this online marketplace. How can we improve the conversion? When we dug deeper, we identified. That the reason that people were not converting is if I looked for Pashmina Shaw in Pinterest and I got a certain link and it said, Hey, this is the online marketplace that hosts it. And I clicked on it. That came to the online marketplace and that link was broken. So all the feed that was coming from Pinterest was coming to a page not found URL on the online marketplace. So guess what? As a user, he doesn't see this. They just log off. They're not going to research for, oh, I want to look for pash. So the solution for this is not analysis solution is fix the link. And that did work. So I think at some point, sometime some analysis when we try to do these analysis, et cetera, the good thing is we went with an analytics framework to identify which channels are performing, which channels are not performing, where are the visitor dropouts, how can we improve the personalization and things like that. But we chanced upon something which was very basic. One of the projects we did with the leading browser ecosystem, and we can't quote them because it's confidential, but there are not too many browser ecosystem, but this is the leading one, is we found that, and they had a very stiff target in terms of how many browser downloads that they wanted to have on launch. We were analyzing a lot of the unsubscribe, so the uninstalls that were happening on the browser ecosystem, and what we did when we found out that. A lot of markets that were not English speaking, were having a high amount of uninstalls. Now this is a good analysis. And we said, let's find out what's happening. And then we had a lot of hypothesis around it. When people uninstall the browser there, there is a comment saying, why did you uninstall this? Right? And so we said we take all those comments and analyze it. We did topic modeling around that. And guess what? The reason the browser was getting uninstalled is because it did not give local language support. And that was funny. Because if I'm in Germany and I have a browser that does not support German, I obviously don't want to be looking at that. I want to have something in my native language and they just start to turn that feature on and that automatically solve the problem. Right? So the beauty of analysis is you may start with an analytics exercise, but you may chance upon certain things that can have core things that you can actually solve. And you solve them before you go to more complex analysis or complex solutions is what I would say. So businesses can be solved with simple solutions as well. [00:20:16] Megan Dibble: Yeah, those are some great examples and good examples too, of digging into the why. Obviously you all looked at what was happening and then that next step that we talked about of. Digging into why it's happening sometimes maybe that does lead you down the path of creating some kind of model or gen AI thing, but then sometimes it leads you down the path of. It's happening 'cause of a code issue or because of a simple business problem. So we've been chatting about trends in analytics, trends in ai, things that your company is doing. I'd love to hear from you on what makes you excited in the data space for the next year and beyond. What technologies or trends make you excited for the future in data? [00:20:58] Krishnan Venkata: I think we've seen a lot of excitement around gene ai, but we are very excited about it personally in terms of the impact that it can have across the organization. We look at this as a three-pronged strategy internally. One is problems that can be solved significantly using gene AI applications, and this was search and discovery. We have a platform called Laser Internally, which does that. We have gene AI based insights and personalization for communication, which we are using in what we call AI pen pal. So we see solutions where there are problems in the past that can totally be solved with gene AI based solutions. So we are very excited about that. The second part of it that we are seeing is the natural language generation part of gene AI is useful in some of our platforms. So be it match view, which is used for store comparison, be it connected view in supply chain. All of these platforms that we already have built now have a natural language generation, which we're using for giving insights. So we are now telling the businesses not only what has happened, but our reasons for why this is happening, and this is already there in some of our solutions and platforms. The third part of JI is the productivity improvement that we are seeing on analyst floor, right? Can we make the analyst faster, smarter, better in doing his job? And we are seeing. Cases in which we are seeing 20 to 40% productivity improvement across different pilot teams that we are working on. Now, again, we are doing POCs on that and trying to find out why those things are there, but we feel that there could be a significant amount of what we call the empowered analyst that we could have because of GNI. The second thing that we are very excited about is authorizations earlier used to either do product analytics, marketing, analytics, and a bit on other places. We are now seeing that the organizations across the functions are actually looking at embracing analytics, and there is more cross-functional collaboration to look at data and then make insights. So the sales is not doing it in silos. They're also looking at marketing data. They're looking also at production level data, customer issue data, et cetera. And all of this has been possible because of the amount of infrastructure investments made in the cloud in terms of data warehouses and things like that. I think that has been a big unlock for us. We are seeing a lot of use cases where data collaboration across different teams are actually helping us. The third part we are very excited about is organization doing cross collaboration across their data sets. So organization A and organization B look at saying that this data that I have on my customer could be a great opportunity for organization B to partner and cross sell or upsell into. Those are fairly exciting opportunities that we are seeing now coming. I think it's a fairly exciting time to be in the analytics industry right now, and the graduates and even analytics practitioners are finding it each day to be quite innovative and exciting from that perspective. [00:23:57] Megan Dibble: Definitely. And that's really encouraging to hear that you're seeing organizations where silos are breaking down, where analytics is being adopted across departments, you know, data sharing because it's the cloud. I think that's really exciting. Just imagining all the things we can accomplish, and if we're more productive too, with the help of generative ai, then yeah. [00:24:18] Krishnan Venkata: So [00:24:19] Megan Dibble: who knows what it's gonna look like in the future, but I think that's definitely some hopeful pieces. So yeah, thanks so much for coming on the show, sharing about trends from your perspective, from what your team is seeing. Really appreciate the insights you shared with us today. [00:24:34] Krishnan Venkata: Thank you for inviting me. [00:24:36] Megan Dibble: Thanks for listening. To learn more about top expansion in this episode, head over to our show notes on community.alteryx.com/podcast. See you next time.

This episode was produced by Megan Bowers (@MeganBowers), Mike Cusic (@mikecusic), and Matt Rotundo (@AlteryxMatt). Special thanks to @andyuttley for the theme music track, and @mikecusic for our album artwork.

Alter Everything

Episode Guide

164: Data Analytics and AI Trends for 2024 and Beyond

Panelists

Topics

Transcript

This episode was produced by Megan Bowers (@MeganBowers), Mike Cusic (@mikecusic), and Matt Rotundo (@AlteryxMatt). Special thanks to @andyuttley for the theme music track, and @mikecusic for our album artwork.