Adel Nehme: Hi everyone. This is a Dell data science educator and evangelist a data camp. If you've been listening to data frame for a while now, you promptly know that data democratization is one of my favorite topics to discuss on the show every now and then I get to take a bird's-eye view with guests and discuss the broad ways organizations can become data-driven.
And this is one of these episodes today's guest is Muna. Manal is a data and analytics, strategist, and transformational leader with over 22 years of experience, building data analytics platforms and driving enterprises to be insights driven. She has a wealth of knowledge and specialties, including the illiteracy programs, data monetization, enterprise data analytics strategies.
The others she's led data teams, various retail organizations, such as Macy's tailored brands and more. And she's one of the few data leaders I've spoken to that can really articulate a simple coherent strategy for data democratization. Throughout the episode, we talk about the components of data, democratization, data, culture, and people, the importance of standardizing business metrics that you've data democratization, which we spoke about at length, how to enlist data champions is analytics leader.
And. If you've enjoyed this episode, make sure to rate, subscribe, and comment, but only if. Also, don't forget this week, we'll be hosting data, camp radar, our digital summit on June 23rd, during the summit of a variety of experts, including many people from different backgrounds, we'll be discussing everything related to the future of careers and data.
Whether you're recruiting for data roles or looking to build a career in data, there is definitely something for you. Seats are limited and registration is free. So secure your spot today on events.datacom.com/radar. The link is in the. Now onto today's episode. Now it's great to have you on the show.
Meenal Iyer: Likewise. Thank you so much for the opportunity to speak in this.
Adel Nehme: So I'm excited to deep dive with you on your work. Leading data science and theater brands, the importance of data democratization and how data leaders can really create a vibrant data culture. But before, can you give us a bit of a background about yourself?
Meenal Iyer: Absolutely. So, thanks again for letting me speak about data democratization. That is one of my favorite topics to speak about. So as terms of the background, I've been in the space for about 20, over 20 years and have experienced like the good, bad and ugly of the space worked across multiple industries, which gave me also the experience to solve for unique data problems I enjoy or enabling data enterprises to essentially become data driven.
And, you know, that has kind of become my motto to provide the right information to the right people at the right. I'm currently working at tailored brands. I had data science and experimentation here in addition to data now.
Adel Nehme: That's really great. So you mentioned here, like data democratization is definitely the topic for this conversation. I want to set the stage for today's chat by first unpacking what it means to democratize data. You know, we've seen a lot of organizations investing tons of resources and becoming data-driven.
So I'd love to first start off the conversation to understand how you define data democratization.
Meenal Iyer: If you ask me, I think I found the best definition, as something from what Bernard Marr basically said and what he said. And I quote a data democratization means that everybody has access to data, and there are no gatekeepers that create a bottleneck at the gateway to the data. And the goal is to have anybody use data at any time to make decisions with no barriers, Lexis or understanding.
And if I have to kind of rephrase this a little bit, I would just say data democratization is not just about data access, but actually comfortable data ex. So you have a documented, ready to use platform with tools and a culture that actually provides data driven decision-making and data literacy. So in addition to just telling everyone that all we have centralized all of our data access, you actually need to meet that access very, very comfortable and convenient.
So end users can use it to become more self.
Adel Nehme: I love that concept of comfortable data access, because I think a lot of people stop at just the data access component of it, which really hurts the ability for people to make data-driven decisions at scale. So, as you mentioned, there are definitely multiple components to democratizing data effectively.
This can range from scaling data access and data infrastructure, centralizing data, having like strong governed data as well. Culture and upscaling and a lot of different things that need to come together to be able to create a data-driven organization. So I'd love it. If you can walk us through the different ways organizations can accelerate data democratization.
Meenal Iyer: Absolutely. I hope you have a pen and paper ready, Adele, Dell, because. Of items. So I would first say that if the organization is just starting with data democratization, the first thing they need to have is an enterprise data strategy with a prioritized list of priorities as the work needs to be done in data itself.
The reason why this is so key and so critical is the fact that this is a material journey and without the executive or enterprise support, it is going to be almost impossible to make this mission a success if we do not have the strategy in place. So this is very, very key and very important. So before anyone even embarks, you have to know what is it that you're setting out to do and have basically your longterm vision set out. Once you have your data strategy set out, you essentially start building out or thinking about a more forward looking architecture. So by we say forward looking architecture is that you have to at least look out three to five.
In terms of what your platform is going to be able to provide. So you have your near term priorities, you have your strategic objectives for the organization. Where do you see your organizations in terms of data over the next three to five years? And that's what your architecture is going to turn out to be so that you are not continuously you not doing migrations are not continuously trying to revisit this architecture.
And then you're always in the process of building out and never at the point where the organization is really self sufficient with their data needs. The third is. So again, we talk about where you try to reduce data silos within the organization. So that's one key part of data democratization is to have access to the data within the organization itself.
Now there may be some limitations to certain data, which you may, will not want to bring into a central warehouse or a central data lake. And that is okay. But for most part 85 to 90% of the organization's data, essentially. pipe into the system. And so you need a centralized data team who basically helps with the piping and who helps essentially with managing the access at the rosters. The fourth is essentially data. Now, this is very, very key and very important to everything that you do. You basically build data lineages and data quality as part of your process as you're typing data in. So ensuring data quality will also give credibility in terms of what is being reported out of your platform.
And also when you get to the point where this goes to execute is of a new start going into. Doing more data science or building data models and machine learning. You know, that your data is not skewed and not biased. You can always trust the quality of the data. So you trust the output of whatever is coming out of the data itself.
The fifth one and my most favorite is the semantic. So the semantic layer is very, very key just for the reason that that's the layer of the view that the end user has access to. And this is the view that they use to essentially consume the data that exists within your ecosystem. So when we talk about comfortable data, The semantic layer is the key to that comfortable data access, because it tells your business users that, okay, you can basically query me without having to write very complicated sequel or doing very complicated joints.
You have predefined metrics, you have predefined KPIs defined within that semantic layer. So there is no reason for the organization to have confusion in terms of what is it that you're looking at or what is it that. days. So once you have this platform in place, you'll meet regularly with your stakeholders and you tell them, okay, and at every point in the journey, actually you have to meet with your stakeholders and continue to update them on the status of what's going on.
And at this point, what you should do is you have to be ready for some maybe reprioritization because these are organizational goals that you're trying to meet as part of the build-out of this platform itself. So you may expect some shift in priorities. Here are some additional priorities which you have to account for, but the meeting with the stakeholders also ensures their continued support for this journey that you are on.
Then you focus on basically the tools for the end-users. So these would be a standardized set of tools that essentially meet again, the needs of almost 85 to 95% of your end-user base. There are going to be some end users who are going to have very specific use cases. And you have to think about how you want to address those, but for the bulk of the organization, essentially you have to think about what are the standard tools that we have to provide to them so that they are able to get the most value out of this.
And then you have data stewards essentially identified for all of the data that is coming into this platform for each source of data that you're bringing into this platform. You have data stewards identified two subject matter experts who essentially are able to help define or dictate as to what the rules of the data going from the system essentially coming in. Now, these people also become your data stewards for all your governance processes. And, you know, privacy processes and everything as well. Then we talk about data privacy specifically to support CCPA and GDPR. For example, customer data is very sensitive. You have HIPAA rules in terms of healthcare data.
All of this is very, very key. So the data stewards essentially work very closely with them to identify what's sensitive versus what's not sensitive and who can get access to. And then you work on data literacy, your black form is set up and you basically want to improve and increase the adoption of it across the organization.
So data literacy programs where you provide continuous training, that verite ensures that adoption and it ensures the continued usage of the platform so that it continues to stay. you have to continuously evaluate and translate the sticks on the platform and understand if there are certain areas of your platform that are not being used.
So you have agric functionality, but then no one is actually using it. And try to understand that, how did this become a priority and not get, um, used that as an exercise, as a leader that you have to continually keep doing. And if you do these, and this is like a small list of components, essentially that lead to data democratization, then you know, you are. I wouldn't say guarantee, but you're bound to move towards a successful implementation.
What is the most challenging and important leverage scale?
Adel Nehme: I love the clarity by which you approach the discussion here and how you position each element. So concisely now, of course, out of all of these levers, there are so many things that we can unpack from governance, infrastructure, upscaling out of these different levers. What do you think is the most challenging and important leverage scale?
And why is that? Why is that the case?
Meenal Iyer: So I think the most important one from my standpoint and what has been the most challenging is the data culture. And that comes to use it at all. And literacy itself. It's human nature, essentially. You know, you're very, you're very comfortable with what you work with and you're comfortable with what you have access to.
It provides you so much security that you are the one who kind of knows this stuff. And all of a sudden you come up with something new, which is going to basically automate. And then you're like, oh my God, no, 70% of my time that I was spending on doing my stuff has now been reduced to 5%. What do I do?
And that's, I think the biggest challenge. That organizations face. And it's more so with legacy organizations, organizations where, you know, which have been in place much before, you know, technology came into play. So affecting data is from my perspective, the most challenging and the most important level of the scale that adoption is very, very key.
If you truly want to enable your enterprise to be data driven. So we have to make sure that the organization countries. To the guiding principles of the platform regarding governance, privacy, the consistency and key metrics that are being reported. And I B probably we can talk about enterprise key enterprise metrics being consistent across the organization, the later point.
But you know, all of those are very, very key in them having to accept it just so that you can truly create a culture of. The decisions that are made by data again are not skewed and are biased and coming from a place where the data cannot actually be trusted. So I think changing the culture from my perspective has to be the most useful.
Adel Nehme: Definitely. And it's all about change management and giving the ability of people to kind of dig for data for themselves and creating that mindset shift as the leader of. Uh, lack of data, culture effected the adoption of some of the solutions that you've developed, even outside of creating a platform where people can do data for themselves, but as a solution for the data science solutions that you adopt.
And what are the ways that you've been able to battle through such adoption?
Meenal Iyer: So the biggest study, or that comes from a lack of data, but Judah is essentially. The proliferation of data silos within the organization, you have data silos, you have reporting silos, you have organizations various. 15 20 30 reporting tools within the organization. You have, don't even ask how many data architectures sitting over there.
Data is not in a singular place. Reporting is all over the place and numbers are never matching. And then organizations struggle to say, okay, which is the number that I should look at. I'm setting a goal to basically say my sales has to increase by 10% next year, but which sales should I look at when you look at it, if there is a lack of that data capture or the lack of.
You're not looking at a singular govern dataset, then all of a sudden these kinds of problems become more dampened and they exist within organizations. And it's not uncommon to have. You have increased total cost of ownership because now you have so many of these self managed data and datasets and then no ability to control.
So if you have sensitive data that moves across the organization, you have no ability to manage or control that. And the things like CCP that comes in, it becomes so important that very sensitive data essentially has to be managed so closely so that if a customer comes in and says, I need access to this data.
You have to remove that data out of the system. Then it's not easy to do if it's all over the place and you have no way of identifying where it exists. So I think that lack of data culture does cause a lot of problems, but some of them are tangible. Some of them are not, and some of them are visible and some of them are not, but it does.
It causes a lot of issues. How do we make data essentially a priority in every conversation with every initiative that takes place. We start talking about what are the success metrics for that initiative and how are we going to measure it? Because you cannot manage what you can't measure. have to be able to measure that. So how do we bring data into that conversation? Now it may translate to saying that we truly don't have data needs, but it is important for the data person to data leader to essentially have a seat at the table. So there, what that will allow is that, okay, now you have a whole list of stakeholders who are sitting with you and telling, okay, this is the initiative that is coming up.
And then it allows for you to have. The conversation and see, okay, this is what it is going to touch. This is what it's going to impact. Then for them to have a top down conversation where they cascade that information down to them, to their leaders, and then it goes down all the way to the . Bottom of the, of the enterprise itself.
Now, the other thing you have to look at is also bottom up. The top-down approach. Sometimes it's not very effective just for the reason that it's represented in a very different way. So to ensure that that message is reached, you have to ensure that there is also communication from bottom up. you show value essentially from what is coming out of your platform or what is going to get impacted and see that how you can essentially provide that value for their business unit and our team also.
And so that needs to be something that you have to be continuously communicating to them. So that's one of the ways in which you can ensure that that lack of data culture does not exist within the organization. It of course happens in a piecemeal way. In some organizations it's easy. They already know what they want in terms of data centralization or it exists.
It's just that they want to truly democratize data. So depending on where you are within the organization, that are, there's a communication that you have to do either from top down or bottom up, and that will help address the issues that basically come with that lack of data by jury, the communication is very key.
Accelerating Data Culture
Adel Nehme: That's really great and harping on the semantic layer that we discussed about and how it also empowers the data culture. I've seen you speak about this quite a bit, and we've mentioned this in our conversation so far. Is the importance of standardizing business metrics to be able to galvanize a data culture.
Do you mind further expanding into that and how it helps accelerate the data culture within organizations?
Meenal Iyer: Absolutely. So let me give an example in one of the organizations that I was working. There was a key metric that was in use within the organization. And we had 13 different definitions of that metric 13, not one or two, we had 13 different definitions, of that organization. So you can imagine just what happened, that when that metric actually got reported across, and that was a metric that was being used to compute to basically as part of our strategy to see whether we were making progress in our initiative.
So it was very, very important that that metric definition become consistent so that we know exactly what number we were starting from and what we were actually driving towards. I talk about this so much because this again is something that is rampant across organizations, finance, for example, has their own definition.
Marketing has their own definition. Sales has their own definition and that sold to other business units, having their own definitions of each metric. So bringing all those stakeholders together and trying to basically understand as to okay, if you have all of these definitions, but as an organization, we all need to go. Attached to one or one definition that we can come so that when we are actually communicating it is be at all, speaking as one, rather than speaking at six or seven different business units and not speaking six or seven different languages. So inconsistency in metrics essentially means where the metric is defined differently across different business units and is being used in different forms.
But it is hard to see. And that leads to a lot of chaos and it's very, very important that organizations essentially come to an alignment in terms of how they truly want to define it.
Adel Nehme: That's really great. And definitely, a nightmare scenario when you have 13 different definitions of the same metric. So if you're an organization or a data leader, Such a situation where there is a lot of messy definitions of metrics. How do you approach the journey of reaching consistency? What does it entail and what does that journey look.
Meenal Iyer: So, what we do here is so first you start identifying what are the key metrics across your organization? And we prioritize them. Think of it, almost like a data journey for you. I wouldn't say it's a multi-year journey, but it is a journey to essentially get all of this aligned. So you need to first understand what are the key metrics that are in use across the organization, and specifically focused on metrics that are used across business unit.
So if a single business unit has a metric that they use, that's not of too much concern just for the reason that it's used just within that business unit. But if there are metrics that are used across business units, that's where the concern stats until you identify basically the list of those metrics.
You sit down with the stakeholders and begin a conversation. Now this is like the toughest part. I don't think the build out of the metric or anything is difficult, but it is this conversation where you have to get alignment in terms of metrics. And the challenge with these conversations typically have been, is who's going to take ownership of this metric going forward.
So you land on a definition, but then who becomes that owner who becomes responsible for tech metric? So this conversation. It's very, very important and very, very key to basically have. you bring all of them together and you sit and then you tell them, okay, we have all of these. So let's come now up to an alignment.
Now, in this case, the data leader essentially is the facilitator or the coordinator of these conversations, but you essentially are waiting for the business to make that decision in terms of what is the definition that should be used forward. So you document and you take it across to the stakeholders and say that this is what we have now.
Can you tell us that this is the one that should be. Once they decide on that singular metric, we say, okay, now who becomes the owner of tech metric going forward? So defining that data steward or the owner of that metric, that person will be responsible now for all communication for that metric going forward.
So business changes the way we look at the baby pivot our business, or the way we look at our customers, our product changes for example, and the metric definition is likely to be. So this person then becomes responsible to say, you know what, just because of these changes that are happening going forward, this is what the metric definition is going to look like.
And that communication needs to be done. They need to get alignment from all the other stakeholders and say, okay, this is what. This is what it is going to be forward. Going forward. This person is also going to be responsible for that definition in a business glossary. So you have a business glossary within the organization.
So in some cases it's a fancy business glossary. Like it comes as part of your data governance tool, but it can be something as simple as a shared Excel document or a shared SharePoint. Or the shared conference document or something where, you know, that is already defined, their definitions are put over there for the organization, basically C so in short, basically this person owns the metric and to end, and then we just go back and it'd be this exercise until we covered all the KPIs and metrics that are in use within the organization.
So, as I said again, it's a journey. I wouldn't say necessarily it's, multi-year based on the size of the organization and based on the number of metrics that you have. you identify and get to that point, you just go to the you've covered all the metrics. Once you have all the definitions in place, then the data leader essentially goes back to his or her team and then goes and has that metric.
And in its definition, put as a metric within the semantic layer. So again, this is why the gear is very, very important. So if I have something called financial. Then financial sales becomes a metric within my semantic layer. So whoever accesses it going forward will get the exact same.
Adel Nehme: And would you consider this like a massive low-hanging fruit that can really accelerate data culture given that it's not a multi-year journey, but something that just requires alignment, people sitting in a room, and this is one of the easiest way a data leader can make an impact in an organization.
Meenal Iyer: Oh, my God. Yes. Yes. I think this is so key and so important and people fail to see that. You're absolutely right. I like that. I like the low hanging fruit. Yes. This is something that can be so, so easily achieved and can be done with just some very quick communication between teams and someone just taking ownership of something so simple, something which they're already working with and they just take ownership of it.
So, absolutely. I think this is something that is very key to bringing at least landing and bringing that data. You're much, much closer to what you're looking.
Role of the Data Leader in Data Democratization
Adel Nehme: That's really great. And I think this marks a really great segue to discuss kind of the role of the data leader and the marketizing data. I think the past few years, we've definitely seen the role of the data leader, whether that's a VP of analytics or data science, a CD. Uh, chief data analytics, officer, it evolved into much more of a culture Seward and a change management stored rather than someone who just sets the data agenda of the data team. So how do you, how have you seen this evolution over the past few years? And what do you think of the data leader's role in democratizing data today?
Meenal Iyer: So I think that role has evolved in leaps and bounds. So if you look at traditional data leadership, essentially the data leader was an order taker. So his or her responsibility was just fight the data into the system and make it available to the end user. There was nothing beyond that. You attack the reporting tool on top of it, and that's about it.
They were an afterthought typically when initiatives launched or projects launched. But over the past few years, when people started seeing, oh my God, data is the new oil data is your latest asset. And everything is all data, data, data, and people are like going all completely clean. This has all has not completely like 60.
And now what has happened is that the data leader has a seat at the table and their responsibility now is just not building those data pipes, but now they're actively responsible for changing the data culture within the organization. Making sure that organization is data literate. That data is being used very effectively across the organization.
In what it's basically, they're like they become an evangelizer of data and they have to become that change agent, which they originally like traditionally were not. So now in bit bit, basically the, just having like a roll of all you had to do was like your data engine. Now you'll have become like a data strategics, your detect transformist you have to think about how you're going to monetize data.
You have to think about, you have to pretty much think about everything from a data perspective. You have to see that an insured, that the value that the business thinks that they're going to get from the data you have to prove and provide that value to the organization. So the role has completely shifted where the owners of all of this is now falling directly on the data leaders.
Whoever is the data leader and who's sitting in the space has a great responsibility. And I'm not saying that they were not responsible before. It's just that a heavy responsibility has now been placed on them where they have to be so visionary. So forward-looking in addition to just being someone who can.
Adel Nehme: And what do you find are key guiding principles to succeed as a data leader in such a stressful, as well as high pressure environment, where you have to really decide on and drive the data strategy of the entirety of the organization.
Meenal Iyer: If you think about guiding principles? I think, um, one is communication is a big, big, big key, right? So when the leader gets into this. He, or she may or may not have access to what they need to do their task effectively. They have to go out, reach out and ensure that they fully understand what is it? That is the goal.
If it's not something that they understand at this point, and what you have to do also is that. You have to have, as I said, a strategy in place that strategy would kind of help you dictate as to what your next steps are going to be, what you are actually going to do. So that strategy is very important, not only like a data strategy, but you also have overlooking technology strategy for your team and for your organization itself.
So communication getting that executed. Having a strategy in place, essentially, a lot of what we actually talked about in terms of the components of data democratization become also your principles for what you need to do to basically make your life much, much easier. And communication holds a big place because you not only have to communicate to your stakeholders continuously providing that value that they're looking for, but also to the folks who are your end users.
And continually telling them in training them and telling them about the value of the data as well, so that they adopt your platform. So I would think in terms of guiding principles, I would probably go back again to my components of data democratizing. And see that, okay, these are, this is almost like a checklist for me.
And let me ensure that I'm going and performing each of these steps. This should help reduce that stress, that data leaders now face. And then of course, you keep yourself very, uh, engaged with the industry. I've attended a lot of data and analytics conferences where thought leaders come, they share their best practices.
They share their thoughts and challenges in terms of what's going on, where they are challenged and hearing and listening to other people, going through the same journey essentially helps you. You know, you have buddies who are essentially are going through the same journey as you and you learn a lot, essentially coming out of those.
Communication and Collaboration
Adel Nehme: I can imagine. And you mentioned the here communication being super key. Of course, a data leader can not do it all by themselves. How do you choose your collaborator collaborators when embarking on these large transformational projects and what does successful collaboration look like in the context of these data transformations?
Meenal Iyer: So I see collaboration happening in multiple. So again, there is this whole piece of where you have a data strategy and you have priorities for your data itself. So in that case, it becomes a little easier because your collaborators are kind of already defined and you already know that they have strategic initiatives, which tied to organizational goals and they become your partners very easily.
But if you are also building out and you are also looking to see that, how can you, how can you bring more business into your. basically reach out to the business unit. So again, that's where communication becomes very important. You almost have to be, I wouldn't say like a know it all, but you have to be someone who is aware of things that are going on within the organization itself.
You reach out to business units and say, you know what I've heard actually, that you'll have this big initiative that you're embarking on and data may be a concern for you. Is there any way essentially that we can step in and help you out? In some cases you are able to get this project funded, but in some cases it's almost like you have to prove it as a proof of.
And once you prove it, then you have a collaborator for life. And that person also becomes your data champion and allows you to promote it. So that's one kind of collaboration that you can do. The second kind of collaboration that you do is going and talking to the top folks. You can go down to the bottom folks and you talk to them and sit down with them, understand what their challenges are, and you can do a similar. terms of how do we, how can I help you? There is opportunity here. I see. And how can I make this easy for you? Go to them, reach out again. It may be funded, may not be funded again, you do proof of value. And again, you've been a data champion for life, or you've been a collaborator for life. I have done this successfully across my organization, and I have seen that one is not only that these data champions become my it's like the train, the trainer kind of mode.
They become. I'm being split across the organization. So there have been meetings where they sit in and I have not sat in and they speak to the capabilities of my platform, much better than I ever could. And so that becomes an automatic showcase for my platform without me having to say anything. So, as I said, we find collaborators in many forms.
You just have to be aware of what's going on from a data standpoint around the organization. You go to each out and you just have to go with the attitude as possible that you're just going to probably get pushed out or you may not get funding, but if there is value, you know, that has to be unlocked because sometimes people don't see the value that has to be unlocked.
But if you see that value to be unlocked, then you almost, you forced take that thing and you say, okay, you know what, let me prove it to you. There's no cost to you. I do it for you. And you tell me whether this is not going to change your life. Almost like 90 to 95% of the time it has changed. I think that sees a lot of success.
And for me, if you ask me that's the best way in which data championing or collaborations with others have democratize my data better within the organization.
Adel Nehme: That's really great. And I love this virtuous cycle model that you're proposing, where you start off with low-hanging fruit. It creates more evangelism of the data and generates a lot more opportunities to do a lot more low-hanging fruits. And it's like a cross-pollination of data within the organization.
Do you mind sharing to a certain extent successful low-hanging fruit? The projects that you've worked on, that we're able to create data champions. And how do you prioritize which projects to go after first and your data journey to be able to create this momentum and keep it going.
Meenal Iyer: Yeah. I have a couple of use cases that I could actually share. So in this one organization of ours, we had this whole team of marketing analysts, essentially, who were there. And one of those analysts essentially was spending 70 bucks. Of his week on doing this reputative task. Okay. So he was given an exercise and spend 70% of his week doing all of this manually.
Then again, he repeated that same thing, 70% of the week because the parameters have changed and, or shifted. It's a challenge. So you have the severely intelligent individual who is here, and that person is doing a manual task spending 70% on this time on probably something that he was not actually hired for originally, but there is so much more value he can actually provide to the organization.
So we start first to his manager and then eventually to him, and we said, okay, we can help support you. And we can actually automate this. Function out for you. And he would look a little wordy and he's like, there's no way. This is so complex. And everything said, why don't you just give us a chance? And we can show basically what you can actually do.
So it took us a good two and a half to three weeks to basically get everything that he was doing manually automated. And then not only did we like automate what he had to do, but we also gave him the capability that tomorrow, if he has to change or shift the parameters or variables within that, that sec, he can do that with just the click of a button. He saved not only that 70% of the time, but just because he had all this additional time, he got to work on more fun stuff. And he was so happy at the end of all of this, that he became like our champion for life. He became a key mouthpiece in all meetings. He would just keep saying all, if there's anything that you'll want to get done, this is the team who's going to do it for you.
It, my team's work was done because we are like, okay, we don't have to go and market ourselves anymore. We don't have to prove value anymore. We don't have to showcase anymore. We have someone who is going to do this automatically for us. It's so simple, right? As you very nicely. Low-hanging fruit. These are like low hanging fruit that you just have to go figure it out again.
You know, they exist and go take your opportunity and go after it. In another organization I was working, there was a very critical problem of where they were doing this computation. On-hand inventory manually. And it was at a very, very, very granular level. So there was so much, so much data to crunch that.
And again, since they were doing it manually, they were able to manage it only twice a year. It's a very critical thing that needs to be done actually at a weekly level, but one given just the size and nature of the data and the fact that they had to do it manually just allowed them to do it twice a year.
They could not do beyond that. Even that twice a year was like a nightmare. So again, we intervened, we got into play and they were a very, very, very skeptical team, very skeptical. And again, we said, so you have charge the bill, just go and prove it to you. And then if it works, all you do is basically have the sustained run on our system going forward.
And we went, we built it in, into our new platform. So not only then did we save on resource efficiencies now, basically the process even ran Vicky. As it was originally, this basically desired. So that was what the actual intent of that work was. And again, freed up a lot of time and then they could actually focus on other things that were sitting on that plate, which they were not able to focus on because of this whole nightmarish thing they would do.
So that data champion again, became so happy that he worked with us to eliminate that data silo and the reporting tool his team was using. So he said, you know what? We don't need this. Your platform is the one that we wanted to start building everything on. So when it came to migrating everything from his tools and his platform on, he became the champion for his team, he went and spoke to his leadership and said that this is what we need to do, and we need to move out of this.
So. low hanging fruit, something we can only get to. And something easy to do as a, as a leader, as a data leader, as I said, you have to be like a communicator and an evangelist. And as long as you are able to get to the root of the problem and then figure out how to fix it for them, you are going to be able to successfully ensure that your organization is using data in the right way.
And they adopt your platform, which essentially was just built for that.
Adel Nehme: That's really great. And I love these stories, especially when you showcase the enthusiasm that it created. Now, as a data leader, you want to balance out between the low-hanging fruit and the strategic long-term projects. And even the more technically demanding data science projects that require machine learning, highly complex algorithms and tools.
The only subject matter experts like data scientists, data engineers, machine learning, engineers know how to master. How do you balance out that roadmap between short-term and long-term wins? What are some of the exciting kind of long-term projects that you've worked on as well?
Meenal Iyer: Awesome. So this is how I manage, and this is specifically for me, how I manage. So the vid source and the way I think about this 70% I work on strategic. And present is essentially ad hoc where my low hanging fruit falls and any other ad hoc needs and requests that just pop in. And then 20% is technical debt.
Okay. So we do accumulate technical debt and stuff starts getting older. And then how do we keep it always new and fresh? So that's how I do the breakdown. Now, in some cases, what happens is that just because of timelines, one becomes 90% and then the others just become a little bit. But overall as an average, I maintain it like the 70 10, 20, um, percentages.
I think it was about eight months old. So we were almost going to be close to the, probably we're all having a beta customers and not production customers. And we were going to be. One of the first ones going into like going into the AWS Redshift. And the interesting thing was that my team had basically traditional data engineers, including myself.
So we hadn't had access to cloud technologies and we hadn't done that. And I created a training plan with one and a half months. All of my team became like cloud. I wouldn't say experts, but we became, we knew everything there was to know about the AWS. We probably could talk about that in a separate session in terms of how we did that.
But I was so surprised at how my team just, they just all pick it up. And in six weeks we were already, we built out the whole architecture, our whole architecture. Okay. And this was not only the batch architecture. We built a full streaming architecture in that. And we launched on time. Of course there was a lot of money savings because we were migrating out of a very expensive platform as well.
And then of course aligning with the broader technology strategy for the organization. And then in addition, what we were able to do is that we were able to build a fraud detection model on this platform and which would actually run in near real time and then spit out the, essentially back to the application, it would spit out a score to see whether this transaction was going to be fraudulent.
And that saves so much of the resources on the risk management team, because otherwise risk was going through these things actually manually. So our ability to basically be able to score the transactions and we had about a 92% accuracy, it made it all worthwhile. So a lot of exciting use cases to be worked on, but this one I'm very particularly proud of.
Call to Action
Adel Nehme: That is so nice. And as we end on a such an inspirational note talking about kind of the value of creating a training plan, do you have any final call to action? You know, before we wrap up.
Meenal Iyer: I probably repeat what I said when you asked me to define data democratization. So again, data democratization is more than just bringing data altogether into a central location. So it's not only about side solving data silos or centralizing data, right. It has to be. It has to be truly an interface that the user can become sell.
So if it has to be an interface where the organization truly can say, yes, we are moving in a data driven fashion, it has to be an interface in which you provide comfort. You provide the tools, you provide the access, and then you provide all the processes, the backend processes of governance and privacy and all of that, and help basically the organization succeed in.
Uh, data-driven environment. So I think for me, in terms of call to action, I tell all data leaders is that focus on what that end view for the customer should look like rather than just focusing on piping data in and just bringing all that data into a central location.
Adel Nehme: Thank you so much to me now for coming on the podcast. I really appreciate it.
Meenal Iyer: Thank you so much at Delta. Thank you for letting me speak.