The Impact of Cleaner Data, with Manish Sood, Founder, and CTO of Reltio

(Joel Beasley at 00:00:03) Hello, my friends. Today we're talking to Manish, founder and CTO of Reltio. And we discuss how Reltio helps companies accelerate the value of their data, how cleaner data can have an immediate impact on the bottom line of your business, and why every company needs a solid cultural foundation in order to scale successfully. All of this right here, right now on the Modern CTO podcast.

(Manish at 00:00:31) Here we go.

(Joel Beasley at 00:00:32) This is the Modern CTO podcast.

(Joel Beasley at 00:00:41) So how did you first get into technology?

(Manish at 00:00:44) Great question. So out of engineering school, I majored in mechanical engineering, which I had some exposure to software because of a lot of convergence that was already taking place. And this is, I'm talking about 1995 when I graduated from my engineering school. So even mechanical engineering had some amount of software baked into it. That was my first starting point introduction to software, to the world of software.

(Manish at 00:01:16) But then I stepped into the workforce, and in the workforce I started working on areas related to mechanical engineering, but they all had heavy dependence on software and technology. And that sort of got me warmed up to how, in the business scenario, software was being put to practical use. So that was sort of my initiation to the software technology world.

(Joel Beasley at 00:01:42) That's really cool. A little while ago, one of my favorite episodes we did was with a company called Augury. I nerded out on it because I'm an audio guy. They record the audio of the machines and then run AI on the audio recordings to hear how the machines are doing.

(Manish at 00:01:59) Yeah, I still remember back in college, one of the first mechanical engineering projects where applying software sort of became a necessity, at least how I learned about it, was this area called finite element analysis, which is essentially structural analysis of bridges or machinery to find out where the failure and stress points may exist in that structure. So you're going through a lot of mathematical modeling in order to figure out where those failures may occur. And because of the complex computations, using software is a much easier path to solving that problem.

(Joel Beasley at 00:02:41) Yeah, absolutely. What was that applied to, some real world structures that you were working on? That's really cool.

(Manish at 00:02:49) Yeah. So any time you think about any kind of structures, whether it is buildings or bridges, you have these trusses that you use to put them up. And figuring out where the stress failures can happen was sort of the practical application of that.

(Joel Beasley at 00:03:05) That's awesome. So how did you find your way to founding Reltio and what you're working on today?

(Manish at 00:03:12) Yeah. I started Reltio in 2011. And in 2011, it was a very interesting point in time. The cloud infrastructure was becoming available, so Amazon Web Services type of capabilities were coming to life. The second thing that was happening in parallel was companies like LinkedIn, Facebook, even Amazon or Netflix, if you remember, they were all building their infrastructure capabilities on this thing called big data stack. And they were talking about how they were not using the traditional databases or the traditional technologies that came from companies like Oracle or others. And they were working heavily on open source projects that were trying to redefine some of the core technology constructs that were in use. And things like Hadoop or MapReduce was talk of the town back then.

(Manish at 00:04:10) Or NoSQL databases were being heavily leveraged by these companies. So when you look at that combination of those two things, it was a unique opportunity or a point in time opportunity where the reinvention of some of these technologies from databases all the way up to how applications would be defined in a completely horizontally scalable manner. Right? Not just that you need bigger machines, but you can take smaller machines and have a farm of those machines that you can keep adding more nodes to and keep scaling the infrastructure for it in order to solve bigger problems.

(Manish at 00:04:50) And that really got me interested in this area of work, which I had been involved with, master data management, where the problem was getting more complex because there were more data inputs that were being provided. But the limitation of existing technologies was that it was starting to run into a scalability performance type of a wall.

(Joel Beasley at 00:05:14) Like there wasn't enough compute available?

(Manish at 00:05:17) Well, there was enough compute if you could scale vertically, if you could go to a bigger box. A bigger box, more memory, more hardware type of box. But the concept of horizontal scalability wasn't really being applied to enterprise grade systems. If you wanted more performance, go buy a bigger box to run it on. That is the answer. So that created a perfect opportunity to think about if data is going to be a growing problem where we don't know what the upper bound of the limit is going to be, then how do we solve it in a better manner? And what would be a better way of solving an age-old problem that's only growing in magnitude with the new technology stack that is becoming available? So that was sort of the genesis of the idea behind Reltio. And thinking about how we could completely reinvent the solution by using these modern technologies and bringing the solution to life so that it could serve the purpose for customers for the next decade or two as we looked ahead.

(Joel Beasley at 00:06:30) So when you were starting it, you were setting out to solve the problem of infinite scalability of data. Is that accurate?

(Manish at 00:06:38) More data, more sources of information. How do you then keep accommodating that growing set of information into a single coherent, unified view of information? For example, if I wanted to find something about a company like IBM, there are so many sources of information available across the internet, across some of the third-party datasets that are available for commercial use, or even the information that a lot of the organizations who are trying to sell to IBM will try to gather in their own CRM or ERP systems, all the internal first-party systems that you use. Everybody's capturing some bits and bytes of information, but none of these sources have the total comprehensive unified view of that information about the customer that they're trying to service. And that was sort of the core problem that we wanted to solve, that we wanted to create Reltio as a platform or a destination of choice where customers can bring in any number of third-party datasets, any number of first-party systems, combine the information to create a single unified view out of that information, which will then power all their analytics, data science, real-time operations, or business processes that they're trying to use this information across.

(Joel Beasley at 00:08:08) Awesome. So you're helping translate just raw data to business value.

(Manish at 00:08:14) Data to business value.

(Joel Beasley at 00:08:17) Mhmm.

(Manish at 00:08:17) And making sure that all the friction out of assembling that information together is taken up so that this is a model of constant curation where more inputs can continue to flow in, and you have trusted, cleansed, enriched information that you can use to drive business value out of it.

(Joel Beasley at 00:08:39) So I imagine you'd probably have to implement some cool AI behind the scenes for understanding all of it, and then you're figuring out the trends of what's happening in the data to provide business intelligence. Can you take me through what the flowchart, I guess, looks like from data to intelligence?

(Manish at 00:08:57) Yeah. From data to intelligence. And again, this is an ecosystem that we play in because when you think about any enterprise technology organization, for example, it could be a traditional life sciences business or could be a retail business or even financial services, they all have lots of different applications that they're using. But each one of those applications, I'll use customer data as an example, have some parts of customer data. So the first step in this flowchart is identifying all the systems where you have customer information.

(Manish at 00:09:36) And the second part of that flow is identifying all the third-party systems that you're trying to gain some kind of enriched understanding of customers. So it could be sources like Dun & Bradstreet or, depending on the industry, you may have sources for life sciences where they get information about physicians or healthcare organizations because they have to, for compliance reasons, get the up-to-date information about them. But creating a list of all those sources and then taking the core critical inputs into Reltio—name, address, any other type of demographic information that is mentioned in those systems, any kind of derived attribution. Because in case of retail, if you're a customer of some retail company where you go browse the website and you see specific information that is of interest to you, it may be based on some kind of association to colors or types of clothes that you like. That information is taken in.

(Manish at 00:10:45) And then we start to run algorithms on that information because these are all inputs where we may find some overlaps because every system will have your name, address, phone, email type of information, and sometimes they may have conflicting information for those different attributes. Or sometimes they may have complementary information because one system only has name and email, another system has name and address. Right? So if you join those two systems together, then you have name, address, email, all of the information assembled together. And we take that detail from all these sources.

(Manish at 00:11:27) We run the algorithms to identify if there are any kinds of variations in the records. For example, somebody fat-fingered your name and spelled it incorrectly, or somebody was lazy and they say, we don't really care about the first name, so a last name is enough to mention in that input record. And a lot of times we find these variances. We are able to bridge those variances to arrive at the conclusion that this is really about the same person across all of these systems. And that is done algorithmically. And that is where some of the core IP of Reltio is applied.

(Manish at 00:12:20) And then out of that, we create that disambiguated record, which has details of which is the best address to use, which is the best phone number to use, what is the best email ID to contact someone at, or here are some of the derived information based on all the interactions that have taken place with that person. And I'm just using this as the example of customer information, but it could be the same case for organization type of data, product type of data, supplier type of data. Given the supply chain issues that every company is facing today, they have to look at, do they really have a full view of that supplier that they are working with? Because they will have to check against, are they under sanctions or not? Are they on the list of any kind of fraudulent activity that would allow you to gain visibility to the fact that you should not be doing business with them anymore?

(Manish at 00:13:20) Those types of details have to be inferred. So we do all of that work, and we provide the curated output out of that. That really then allows every one of these organizations to use that information for business insights, for data science type of work that they do, so that data engineering can be eliminated out of that equation or reduced while they can focus on the data science part of the work that they have to do to infer better insights out of it. And clean data leads to better outcomes is sort of the basic hypothesis in this whole process. And a lot of times customers will see, as an example, on the marketing front, they will see that they had an uptake, 70% uptake in their conversion rates from lead to sale because they used clean, better trusted data in order to make the decisions.

(Manish at 00:14:27) And that's just one example. Think about revenue loss prevention. If you're able to identify upfront in the cycle where you have fraud or other types of compliance violations, then you're able to prevent that revenue loss as well. So those are the types of outcomes it starts to drive once you have clean, trusted information available as a starting point.

(Joel Beasley at 00:14:54) That's really cool. So I'm starting to wrap my head around this. You guys are solving lots of different problems around data from siloed sources. You've got to combine your internal data sources as well as third-party sources. So you mentioned you help other companies, but I imagine dealing with a lot of internal data that's pretty sensitive, especially customer data, and today there's a lot more regulations and compliance coming out. Is that a challenge at all for you working internationally? Do you have a team dedicated to working with the regulations? What does that look like?

(Manish at 00:15:29) The privacy regulations are evolving across the globe. If you think about some of those regulations, even in the US, there are multiple regulations that are evolving all the way from what California is doing versus how other states are starting to follow suit. And at the same time, GDPR type of compliance requirements in Europe or other countries are starting to come up with some of the same type of regulatory details. For example, China introduced the PIPL type of information restrictions for any kind of person-related data and how it should be managed. We do depend on industry resources to keep us up to date because even by putting a team together, it's not enough.

(Manish at 00:16:18) So we work very closely with some of the law firms and other parties that are involved in shaping some of these compliance type of dialogue and requirements. But on top of that, what we're trying to solve for is creating Reltio as the platform where customers can come in and start to track how they should comply with these requirements. So in most of these cases, for example, HP Inc. is a customer. It's a global customer where they have across the entire planet about close to a billion customer entities that they manage inside of it. And across those billion customers that they're working with, they are trying to address any kind of data service request that may come in.

(Manish at 00:17:12) For example, if you raise your hand and go to them and say, hey, I want to be forgotten and please do not market anything anymore to me. That request comes to Reltio as the system of record to see across how many internal systems is your information spread across their landscape. Right? And they have 300-plus different applications that may be consuming some part of that customer information because at some point you bought a printer, another time you bought a cartridge, another time you bought a monitor from them, all different lines of service.

(Manish at 00:17:53) And you could have bought it in store or you could have bought it on the website. Any of those pieces of information and which channel you engaged through sits inside Reltio. And Reltio then becomes the map across the entire HP landscape to outline that here are the, out of the 300 systems, here are the 30 systems that were involved in touching or handling someone's information. And this is the work that will need to be done to scrub that information out of it because only once that is done, then the request for the GDPR type of request will be considered complete. So we become the system to track and manage that entire process or coordinate that entire process.

(Manish at 00:18:45) So not only are we trying to keep track of the requirements, but we are also trying to make sure that we are creating the system capabilities that will become the way in which these customers can solve these types of growing problems.

(Joel Beasley at 00:18:59) So for a company like HP, of course, there's a really broad landscape of areas where data is stored, and Reltio needs to integrate with each of those. How long is the time to value from when you partner with HP to where you are fully integrated with all of their different lines of business with full visibility into their data, able to provide insights to them?

(Manish at 00:19:25) You mentioned something that is critical to this part, fully integrated. And I would say that there is never a state of fully integrated. It is a continuous process because there are always new systems, new business processes that need to be integrated. So the way we have been trying to drive value for customers faster towards value is by helping them understand that they don't have to boil the ocean on day one. They have to start with whatever system they can start to plug into Reltio.

(Manish at 00:20:01) And as soon as that system is plugged in, they can start to see the flow of information, the unified view of that data, even if it is one system. Because a lot of systems inherently don't have the capability to deduplicate any errors, right? Because most of the systems, like even a CRM system like Salesforce, wasn't really built with the assumption that somebody would enter a duplicate record for the same customer. And that's why even plugging in one system starts to get customers to value.

(Manish at 00:20:38) And this can be done in a matter of days. If the customer wants more systems to be added, once they've completed the first system getting plugged in, then they can go to system number two, three, four, the thirtieth system, the 300th system that they want to keep plugging into this mix and keep expanding that footprint. And that's on the data input side. On the data value side, the value is almost instantaneous because as soon as you land data into Reltio, you start to see the refined output out of it. This is a very simple example.

(Manish at 00:21:17) As soon as the data comes into Reltio, if it is Adam's record and somebody, let's say you created through a self-service mechanism your own information, you entered some of the details about your residential address where you want the package to be shipped, Reltio will receive that information, and Reltio will make sure that it is cleansed as per the USPS standards for deliverability. And that will happen automatically, right? Nobody has to do anything additional in order to get that input, which in turn saves companies like HP or any other retail company hundreds of thousands or millions of dollars.

(Manish at 00:22:01) If they didn't have the right address or if it was not checked for deliverability, then they would have to pay a lot more for that package to be shipped out versus if it were properly cleansed, hygiene was addressed, and the information was encoded correctly, then it is a big saving for these companies. And that's just one small example of how customers can start to get to value right away.

(Joel Beasley at 00:22:33) So what are you personally focused on in your day to day? What are you spending most of your time on?

(Manish at 00:22:41) My focus is on the product and working closely with customers. When we think about what we have been able to build at Reltio, we are providing this as a capability that customers are running and using Reltio for to run their core business functions. Every business process, if it has to look up customer information, product information, supplier information at the point of engagement or at the point of experience, it's an API call that is being made into Reltio to fetch that information as an always-on, always-available data asset that can be used across their entire enterprise. And when we go talk to these customers, they're asking us to improve performance even further. They say 200 milliseconds is not good enough.

(Manish at 00:23:36) Can you bring it down? 200 milliseconds as the response time. Or it is great that Reltio delivers 99.95% uptime all the time as a guarantee, but we would really like to make sure that Reltio never goes down. So can you operate at four nines or five nines type of availability? And there's always that constant improvement that can be further brought into these types of solutions. So a large part of my focus is how do we go back to the drawing board and create the next generation of our capabilities?

(Manish at 00:24:21) Because if we don't, somebody else will.

(Joel Beasley at 00:24:23) Right. Yeah.

(Manish at 00:24:25) So that constant mode of disruption where we are thinking about what would be the next generation capability and the way of solving this problem in a faster, easier manner that we could bring to life is the area of focus for me at Reltio.

(Joel Beasley at 00:24:44) That's awesome. I love your outlook there of that it's never finished. That's consistent with when you corrected me on fully integrated. No, you're never fully integrated.

(Joel Beasley at 00:24:53) It's a journey. You're continuously improving on it.

(Manish at 00:24:56) Yeah.

(Joel Beasley at 00:24:57) So you're very forward thinking with the company looking for more improvements. What's a project within the company or something that you're really excited about for the future within the next couple years?

(Manish at 00:25:08) Yeah. Adam, one of the things from day one when we started Reltio in 2011, our hypothesis was that this kind of a capability should reside in the cloud because every company out there has on the ground systems. They're going to make a transition to in the cloud type of systems, and they will need a bridge. They will need a bridge that can be the bridge between on the ground, in the cloud systems so that there is constant continuous exchange of information that is taking place. And Reltio in the middle can serve that need as we move forward.

(Manish at 00:25:50) So with that in mind, one of the key decisions that we made back in 2011 was that we are going to set up Reltio as a software as a service capability, but build the software as a service capability on public cloud infrastructure. Why? Because with the type of innovation that was going to take place in the lower layers of that infrastructure, networking, hardware provisioning, doing infrastructure as code type of enhancements that were going to come out where you don't have to know specific details about the machine, but you can run it as a cluster or a big farm or use innovations like Kubernetes, all of these things were going to be on a faster innovation cycle on the public cloud infrastructure. Because companies like Amazon, Microsoft, Google will all continue to invest in that layer cake of innovation.

(Manish at 00:26:51) And for us, that literally has become the layer cake of innovation where our capabilities sit on top of all those capabilities provided by these cloud providers. And we have to focus on our core IP versus the context provided by these lower layers of infrastructure innovation. We become a natural beneficiary of it. And that is why some of the things that we are working on today that I'm really excited about is driving that next generation of further investment in the cloud native capabilities. So one of the big things that we are doing is in 2011 when we started our journey, our database of choice was Cassandra, NoSQL database, which we had to manage and run ourselves in the public cloud infrastructure. Versus now, we are moving to cloud native databases that are offered by each hyperscale cloud provider.

(Manish at 00:27:57) For example, we just completed the first step in that journey, which was to get our AWS footprint or Amazon Web Services footprint up and running on DynamoDB, where we get multi-region capabilities as a natural benefit out of it. We get better resiliency out of it. We get higher throughput and performance without having to sit there and manage any of the capabilities ourselves. And if we have to scale up, for example, some of the customers that we work with, for example, Lululemon as a customer, when they had to run their Black Friday promotions or the holiday season type of promotions, they didn't have to do anything to handle the spike of traffic that went through the Reltio system. Reltio automatically scaled up and scaled down as we work through that promotion period.

(Manish at 00:28:56) And that's the type of dynamic scaling that we can handle faster than ever before by investing heavily in these cloud native type of capabilities. So I'm really excited about doubling down on that original thesis of building the software as a service solution on the public cloud infrastructure and how we can drive faster innovation with that as we move forward.

(Joel Beasley at 00:29:24) That's really cool that you had the foresight back in 2011 to build the company like that back then, and now that's still continuously the thing to be excited about.

(Manish at 00:29:36) Absolutely. I call it the layer cake of innovation. We don't have to reinvent the wheel at every layer, but we can certainly benefit from every change, every innovation that takes place on a daily basis in each one of those layers.

(Joel Beasley at 00:29:52) Absolutely. That's kind of like those day one principles from Amazon. Focus on day one, and it sounds like you guys are still doing that.

(Manish at 00:30:01) Yeah.

(Joel Beasley at 00:30:01) Before we wrap up, I want to ask you a couple leadership questions. Is that okay?

(Manish at 00:30:05) Yeah. Absolutely.

(Joel Beasley at 00:30:08) Awesome. So I saw you're over 400 employees now, which that's got to be a pretty incredible experience as the founder. What are some of the culture items that are really important to you at your company as it grows? And how do you keep that culture strong as your company grows more and more?

(Manish at 00:30:27) Adam, culture is the foundation of our business. When you think about the fact that we are in the software business and we are running everything on public cloud infrastructure, so we don't even own capital equipment. There is no hardware that we own. Everything is software and all of the software that is our business is created by people. And in order to create the right foundation so that we can continue to grow and scale to the massive market opportunity that is in front of us, that cultural foundation is the key investment that every company needs to think about.

(Manish at 00:31:11) And for us, that has been the case as well. And we have been very focused on making sure that we build the right cultural foundation. So making sure that we are investing in people, thinking about our employees, our shareholders, our customers, and our partners has been a key part of that framework that allows us to define what are the values that are most important to us and how we are going to move forward where we continue to use the cultural foundation as the way to onboard and bring on more members into the team. Because everybody will come from different backgrounds, everybody will have different experiences. But those different experiences, the only way to create the common outcome out of those different experiences is to have culture as the key normalizing force across the entire company.

(Joel Beasley at 00:32:10) So what are some of the values that are most important to you?

(Manish at 00:32:15) In this type of journey where you're going from an idea all the way to, today, we are north of $100 million in subscription ARR. Wow. And the opportunity that we see in front of us is to build a multibillion-dollar revenue size and scale company. We have to, first of all, make sure that we are taking every Reltio employee through this process of understanding the market opportunity, understanding how we execute to that market opportunity. So transparency is very high on that list because transparency builds that bond across all layers of the organization.

(Manish at 00:33:00) That allows us to move forward as a team. We are in different geographies. The COVID type of situation has further distributed the team because we have now fully embraced the distributed work culture even though things are starting to come back to normal. We believe that we will be a distributed company as we move forward. But in order to drive a coherent outcome and lead everybody towards that common goal, transparency has to be a key part of that value system.

(Manish at 00:33:34) And we put in a lot of effort as a team to make sure that our principles around transparency and the culture around transparency allows us to build on that. I would say that is one of the key principles or values that we adhere to as a part of our culture.

(Joel Beasley at 00:33:54) That's really cool. And that makes a lot of sense how it makes your employees feel more a part of the team and also adds value to the company. So what is something that you are learning right now as a leader? Something that's challenging you, you're actively overcoming.

(Manish at 00:34:11) One of the big things that I have learned over my eleven years of working on Reltio, we have to constantly ask the question, do we have the right team that will take us to the next milestone of our journey? A lot of it is tied to stages and phases of growth because at every stage or every phase of growth, you need different kinds of skill sets. When you are a $0 revenue company, you need people who can think creatively, who can understand the customer needs, who can really go hard at creating a solution out of nothing. Versus when you're a $100 million company, you have to think about scale, where you have to think about how to go rinse and repeat, how to build more structure and process. And you need a different mix or proportion of those skill sets at every stage.

(Manish at 00:35:17) You can't really say you don't need creativity anymore even though you're a $100 million company. But at the same time, you cannot be zero on process either. So how do you create the right balance, and what is the right ratio as you continue to evolve as a company? Because that ratio will change with every next milestone that you will hit. So that is something that is easy to understand as a concept, but it is hard to really pin down what is the precise ratio that you need for the next eighteen to twenty-four months.

(Manish at 00:35:54) So that is something that if you can anticipate and get better at anticipating where those needs are going to come from and what types of problems you will have to solve is a little bit of looking around the corner and being able to see the picture so that you can anticipate that curve in the road and start to think about how you're going to navigate that. That is something that I find both very interesting and constantly challenging because, hopefully, the next curve in the road is a different kind of a curve in the road and not the same one that you encountered before.

(Joel Beasley at 00:36:30) Yeah. I mean, first of all, thank you. That's super solid advice. Like you said, it's going to be hard to suss out what the precise ratio you need coming up is, but just having it top of mind that that's a problem you need to be thinking about as an executive is going to help you a lot if you are actively thinking about it. You'll find that ratio faster.

(Manish at 00:36:52) Yeah. You used to go from being reactive to being more proactive around those areas.

(Joel Beasley at 00:36:58) Yes. Yes. Well, Manish, is there anything that we didn't get to touch on today that we want to make sure we cover and get out to the world?

(Manish at 00:37:07) No, Adam. This has been great. Thank you so much for having me. I really appreciate it.

(Joel Beasley at 00:37:13) Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics that you'd like to hear discussed on the podcast, either add me on LinkedIn or send me an email [email protected]. Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.