Tarush Aggarwal – Using Infrastructure to Drive Value

Connect with Tarush:

Connect with Dwight:

• 
• dwight.brown@salary.com

Connect with David:

• 
• 
• david.turetsky@salary.com

Resources:

• https://5x.co/
• HR Data Labs Season 3 Episode 17
• HR Data Labs Season 3 Episode 15

• “A data team with a hundred people acts very differently from a data team with 3, 4, or 5 people . . . So unfortunately, what we see is that a lot of [smaller] HR analytics team don't really get to leverage the more mature technology of their sister, larger data teams.”
• “In reality, [HR analytics solutions of the future] will probably be hybrid, where you get 60-70% of what you want out of the box and [you can customize] the remaining 30-40%.”

Announcer: 0:02

Here's an experiment for you. Take passionate experts in human resource technology. Invite cross industry experts from inside and outside HR. Mix in what's happening in people analytics today. Give them the technology to connect, hit record, pour their discussions into a beaker, mix thoroughly. And voila, you get the HR Data Labs podcast, where we explore the impact of data and analytics to your business. We may get passionate and even irreverent, that count on each episode challenging and enhancing your understanding of the way people data can be used to solve real world problems. Now, here's your host, David Turetsky.

David Turetsky: 0:46

Hello, and welcome to the HR Data Labs podcast. I'm your host, David Turetsky. Like always, we try and find you interesting people from inside and outside the world of human resources to bring you the latest on what's happening with HR data, technology and analytics. Today we have with us Tarush Aggarwal from at 5x.

Tarush Aggarwal: 1:05

Hey, David, so happy to be here. Thank you so much for having me.

David Turetsky: 1:08

Our pleasure. And like always, we have with us Dwight Brown from Salary.com. Hey Dwight.

Dwight Brown: 1:14

Hey David. Hey Tarush. Good to have you with us today.

Tarush Aggarwal: 1:17

Hey, thanks so much. Looking forward to it.

David Turetsky: 1:19

Tarush why don't you dive in by giving us a little bit of your background and what you're doing it at 5x as well.

Tarush Aggarwal: 1:26

Sure. So my background, you know, I was fortunate enough, I spent my career in data, I got to be one of the first data engineers at Salesforce.com back in 2011, when no one in the Valley really had a data team spent a few years over there. And most recently, I ran data for WeWork with 100% data team jumpstarted data engineering data platform, moved to China and ran our platform efforts. And you know, believe it or not at the beginning of COVID, I find myself stuck in Bali, Indonesia, kind of taking some time, away from the valley and away from New York helped me you know, put things in perspective. And we so WeWork and Salesforce were amazing experiences. And you know, we were able to do a lot with data. But not many companies can, you know, go build 100% data team and kind of get started, right? And, but like 95% of the world, they want to get value from data. And they don't necessarily have the capabilities of, you know, building multiple layers of the stack. Like today, we have data collection, ingestion, storage, modeling, reporting, and, you know, AB testing machine learning on top of that, and we also care about privacy and security and GDPR and PCI compliance audit, right? So, you know, what do you want to do? Do you want to go sign these contracts and build a stack? Or do you want to get value from data? Right. So you know, think of 5x as you know, think of us as that large platform team, which companies like WeWork in Salesforce have. But instead of doing it for us, we can do it for like 95% of the companies.

David Turetsky: 2:57

So it's outsourced?

Tarush Aggarwal: 2:59

Yes, like you get a, you know, an end to end stack, you know, we bundled together the different providers, the sort of best in class providers, so you know, our houses snowflake, which is where the industry is going, and you don't need to sign any of the enterprise contracts for all these companies, you kind of get it all on day one, you get a monthly bill. And the if you're sort of good to go. And just like, you know, when you use any sort of platform, you get updates. And, you know, as we build more and more stuff into it, you don't need to worry about maintaining a stack, you just get a maintenance window, and it goes back and it comes back up. And it's better than ever before.

David Turetsky: 3:38

So, one fun thing that no one knows about you Tarush?

Tarush Aggarwal: 3:44

Wow, one fun thing that no one knows about me? Yeah, yeah, it's, I've spent a lot of my career I spent most of my career in data. And, you know, part of being tagged as a data person is this idea that you're very logical and you know, make data driven decision, make data driven decision that everything is a spreadsheet and all of that jazz. But in reality, actually, most of my decision making comes from intuition, which is not what most people will believe. And data is just kind of a means to sort of validate that intuition often. So I find it really, you know, kind of on a broader topic that, you know, intuition versus data driven are kind of two separate sides of the stack.

David Turetsky: 4:35

So So, if What if what we're hearing is your intuitive, you're not a geek, that's what we're hearing.

Tarush Aggarwal: 4:41

That that that's, that might be a way to say it.

David Turetsky: 4:46

Okay. Well, well, Dwight and I are geeks, so we'll stick with the data. That's okay. It's different, sir. It's all good. All good.

Tarush Aggarwal: 4:54

I can wear many hats.

David Turetsky: 4:56

That's all right. You can wear a data geek hat. It's okay. I'm wearing a hat today.

Tarush Aggarwal: 5:00

I like it.

David Turetsky: 5:01

So our topic for today is using infrastructure to drive value. And while we're gonna get into it with our questions, one of the interesting things is, you're talking about at 5x, being able to have an outsourced infrastructure company does enable that. But I think what we're gonna do is we're going to talk about, how does HR, or how can HR and the business, get more out of their data, by leveraging infrastructure? So let's go to our first question, then, which is what prevents companies from developing a stack internally to be able to manage the HR data themselves?

Tarush Aggarwal: 5:51

Yeah. So, you know, nothing really prevents companies from building the stack internally, right. Like, you know, snowflake was the largest tech IPO last year, you know, everyone's now heard of it, you know, the data warehouse is no longer a niche use case, it's, you know, the core, every company in the world should now be sort of looking at kind of having one. So it's, there's nothing really preventing companies from doing this, it's more so you know, them choosing to focus more time on the analytics. And until now, this this, they are, they have been kind of having to do this, in terms of setting up this infrastructure and do the analytics. And they kind of want to be spending more and more time on the analytics. So they can do it themselves. As you know, what's happening now is that they're trying to read the basic data, refocus, and kind of spend more time on actually delivering the business value.

David Turetsky: 6:49

Is there any reason why they can't leverage some of the other business analytics stack, or the data stack that other parts of the organization have been building or leveraging?

Tarush Aggarwal: 7:00

I mean, in an ideal world, they would, you know, a lot of these companies who have HR analytics teams, they probably have sister organizations inside the same company, which are the sort of larger data teams sort of focused on the entire business. Now, historically, these teams aren't able to really interact a lot, because the HR analytics team very often has people sensitive information, which you don't kind of want to share broadly, across the organizations. So they're forced to, in some ways, have their own stack, which is a little bit more secure. And given that these HR analytics teams very often are a small fraction of the size of the sort of sister or brother organizations, or data team, with 100 people acts very differently from a data team with three, four or five people. So you know, even just in terms of what stack they use, and how that's maintained, and how that's built and operated. You know, they might have a few pieces in common, they might be using the same database technology, they might be using a similar ingestion layer, but the way they're really set up until now, it's pretty independent. So, you know, unfortunately, what we see is that a lot of HR analytics teams don't really get the leverage a little bit more mature technology of their sister larger data team works.

David Turetsky: 8:25

Yeah, what we've been advocating. And what a lot of companies have been trying to do is democratize the HR data. So it's not HR data anymore. It's more business data. And I totally agree with you that sometimes the data is too sensitive. But in many ways, you can firewall the more sensitive pieces, like pay or like reasons why people leave an organization. And you can then take the aggregated and anonymized data, especially for things like headcount, which are much less sensitive, and then be able to put that alongside of finance and supply chain data. So that the people who need the Insight while they're building the plans that they're trying to execute against or measure against those things. That's where HR data really needs to become business data and and cease to be HR data.

Tarush Aggarwal: 9:21

Yeah, no, I totally agree. And, you know, we now have the ability to encrypt certain columns, right, you know, sort of a warehouse like snowflake kind of, in the enterprise solution, it does run the box where you know, even working with PII, or your social security, and all of this stuff, where all the data is going into the warehouse, but at source, it knows what's the PII column, and it gets kind of encrypted. And, you know, you have to have special sort of special roles who actually have who can unencrypted but otherwise, for the most part, it kind of gets aggregated up. And, you know, I think we're probably going to start to see this trend kind of get more and more mainstream and you know, This is probably a more recent thing in the last one year, where the technology is now able to do sort of selective bi, filtering and, you know, encryption of like certain columns. So, you know, I think we are going to see more and more of that historically, it's kind of lived completely separately.

Announcer: 10:20

Like what you hear so far, make sure you never miss a show by clicking subscribe. This podcast is made possible by Salary.com. Now, back to the show.

David Turetsky: 10:32

Why don't we transition to question two, which is why doesn't mean a one size fits all solution work for everybody? And I think what we're trying to say in this question is, you know, do companies need to kind of build based on their own bespoke needs? Right?

Tarush Aggarwal: 10:50

Yeah. You know, given that HR analytics is not subject to, you know, a particular industry, it's not just, you know, one type of company, and even in terms of sizes, right, like, you know, obviously, anything under midmarket is probably not really a candidate for HR analytics. But you know, there's, again, a big difference in 1000 person, company, and 100,000 person company, and kind of how they kind of look at these HR analytics, right. And then, especially for some companies who look at this a little bit as the competitive advantage companies that, you know, need to hire, you know, the best people in the world and really care about retention, if you look at the tech company, you know, people raise of, you know, making sure you have the best engineers, compared to a more traditional business where, you know, you obviously want to hire good people and retain them, but it's not, you know, the center core of like, what your business kind of focuses on, you know, you have so many different variables over there. And, you know, every sort of solution, every company chooses to emphasize on certain things and, you know, not emphasize on certain things, when you do something sort of, well, and something's not Well, and, you know, do you want to a lot of companies around the HR analytics space are, you know, optimizing for ease of setup, or in terms of rich feature set, or integrating with kind of non bespoke vendors. So, you know, there's enough different levers out there that, you know, one size fits all, doesn't really kind of make sense, right, you're always going to have some sort of nice use cases where some companies are kind of very happy with us sort of standard out of the box solution, whether that sort of Visio, or some of the new sort of competitive, some companies, you know, are just going to always do it in house, like, it's just too much of a competitive advantage, they can kind of use this against the competition, some companies will go hybrid, where, you know, as you start getting into regionalization, you have some of these platforms, which integrate really well with greenhouse and work there. And all of the usual suspects, which kind of come out of the box, but you start getting into the Asia regions, especially China and kind of outside there, where all of a sudden, they use different HR systems there. And some of these out of the box vendors don't integrate with that, they might need to have a custom solution. Right. So, you know, sort of globalization plays a big part size of company, what exactly they're trying to do with all of these different use cases, which is ultimately great, because, you know, one size fits all, in some ways, causes, you know, some sort of monopoly, which is not great for the end customer.

David Turetsky: 13:31

Sure. I mean, you know, you could definitely go that route. But one of the things that one size fits all does do is it might provide companies with an understanding of benchmarking, right, where they can get an understanding about what their experience is versus what other people have. Yeah, other people do. So there are some benefits. But

Tarush Aggarwal: 13:52

and, you know, I think, in general, right, when something gets commoditize becomes available to everyone, there's more information on the table, but also, the marginal utility of that value starts to diminish, because it's everyone kind of has access to it now. So it's a good, you know, it's a good baseline to kind of get set up. And if you're not doing that, then, you know, this is some very, very low hanging fruit to basically get there. But it's no longer a sort of competitive advantage. And, you know, in today's world, we all have fighting for that competitive advantage of what can we do, which is kind of better. And then then the next person

Dwight Brown: 14:29

One of the things that I've found in my past experience has been, anytime we looked at some sort of one size fits all vendor, oftentimes they did a lot of stuff. Well, kind of a mile wide and an inch deep concept, though, and it you know, on the surface, it always looked good, but it was when we started digging down into the details that we figured out, it really didn't fit the need that that we had. And so I would say that that really can be a barrier. as well, with everything, you've got to make sure that what looks good on the surface is actually going to serve what your needs are. And I've seen, I saw over and over that scenario play out and think, you know, probably what keeps us where we are in terms of commoditizing everything.

Tarush Aggarwal: 15:20

It's, it's a weird kind of example. But yeah, it's a little bit like Facebook analytics, right? Like, when you're kind of getting started Facebook Analytics, you know, on the ad manager makes a lot of sense. Yours, how much money you spend years, how many views, clicks, conversions, you kind of gotten, it all looks really, really good from the outside. And all of a sudden, now you get Google Analytics, because you want also want to use Google AdWords. And now you realize, they are not comparing things apples to apples, they have their own definitions on how they want to do things to in order to make their platform looks good. And once you can't compare apples to apples, you sort of realize that, hey, actually, if I really want to do this, I need the raw data, I need to get the raw data and I need to add the insights layer on top of it. Sure. And it's kind of similar, right? Where, you know, a lot of these tools optimizer, how quickly can you go get started, you know, we'll just integrate with everything a year, you had your dashboard, that's amazing to get started. But you could you sort of get to this size where you have a bespoke question and a bespoke question. It's very hard to model on top of a prebuilt dashboard, you have to get into the modeling layers yourself to basically get more sophisticated.

David Turetsky: 16:33

And that's certainly true. Definitely true of certain industries, like the investment banking, the commercial banking industry, where they're trying to solve very specific questions or solve, answer very specific questions and solve very specific problems. And their level of sophistication around their measurement is probably higher than other industries. And so they do have a need, and they can probably pay for those, those questions and the answers to those questions, I should say. So there are going to be some companies, though, that are much more big, just that the beginning stages, and totally may not ever get out of those beginning stages, because the kind of the balance between the risk and reward and the return on that investment, isn't there? And so yeah, they may stay in the world of candor, in a one size fits all, and that's okay. But it doesn't get to your point, it doesn't get you the utility that getting the bespoke answer would would give to those industries.

Tarush Aggarwal: 17:37

Yeah, totally. I think majority of companies will always live in that space, right? Like, how many companies end up becoming 1000? Person companies? Right, like, very, very few. Right. So, you know, those one size fits all? There's a reason why, you know, the big players all talk about ease of use and getting set up and exactly you, you know, 80%? Out of the box. Yeah, that's the biggest piece of the pie.

David Turetsky: 18:01

Right. Right. And I think one of the things that prevents other companies from coming in, to challenge them is that there is incredible amounts of logic, and incredible amounts of I don't want to say middleware, but configurations that you need to get to, in order for people to find value in the data, that, that the economic realities of trying to build, whether it's the metrics, or the visualizations, yes, it's not for the faint of heart, it's very expensive to get going,

Tarush Aggarwal: 18:33

it is very expensive to get going, I think, you know, five years ago, what all would be custom generated, not say, not customer, I'd be like, written down data modeling, you know, logic, you know, what's going to start to happen is, as, you know, decision making, which is this kind of whole area of analytics, which is, you know, having, you know, computer generated storytelling on looking at data sets, seeing how they're changing, and, you know, getting really, really sophisticated around that, as you know, we see more inventions in just a modern data stack, and more and more layers over there, you know, some of that competitive advantage of years, 10 years of experience starts to get a little bit easier. So you're probably seeing right now, new players coming in, which are probably challenging the sort of visitors in the world and being able to do those integrations, not in 678 months, but to be able to go do it in one or two months, right, because there's just so much you know, I'm kind of super bullish on you know, this is going to become more of a level playing field. And, you know, when that kind of happens, more and more companies will go into more niche kind of use cases. Then you you still might have someone who's got 50 60% of the market but it gets really interesting because you have all of these other new these are the niches cases come up

David Turetsky: 20:03

So let's talk a little more about that.

Dwight Brown: 20:05

So since there's so much consolidation, since, you know, inevitably we've got these little companies that are popping up getting bought by the bigger companies, and consolidating towards a one size fits all offering, do you think we'll ever get to the point where the one size fits all is the predominant vendor,

Tarush Aggarwal: 20:27

I think, you know, it doesn't even have to be a black and white answer it in reality will probably be hybrid, where, you know, you get 60 70% of what you want out of the box with this vendor and the remaining 20 30%, you can, you know, have more customizable, like, even if you look at workday, right, like, workday is a solution. But at the same time, workday is highly configurable, where you can build stuff, you know, on top of Workday, and, you know, add your own kind of logic over there. So probably, as some of these platforms, you know, some of these one size fits all solutions will, you know, have a sort of application platform layer, just like Salesforce, you know, it's got a store, which you can build an application on top of Salesforce and surface that out what sort of Workday has the same thing, you are going to see that in the HR analytics space, where you're 60% of the box, and go belly or 40%.

David Turetsky: 21:19

And I think that is definitely true of the HR. Well, what could be called the human capital management providers like SAP, PeopleSoft, ADP, and workday, were they they built a platform of data management. And, you know, even in the ERP will tell you this, too, they built us total data management platform. And they have a data model, which is very clear about how the data model works on the HR side. And then they're leveraging a common infrastructure to do analytics, but then they allow for the extension, or API's to Tableau or some other toolset that enable you to take the data further.

Tarush Aggarwal: 22:08

And in some ways, that works, right. And when you're big enough, you kind of have to make it work, what tends to happen, and we saw this with Salesforce, too, is that, you know, all of these, as you said, they have a very clear data model. And, you know, you kind of have to play well with it. And even if you want to push data, and you have to define these objects upfront and play into this kind of world, which in some ways is pretty limiting, because you have to, you know, you you can't think outside the box, because it's not built to think outside the box, right, where, you know, traditionally, if you use something like a data warehouse, you have all the flexibility in the world. And you're probably you're probably building and analyzing using sort of standard SQL in Python, whereas you get custom objects, and, you know, the, these, these inferred languages on top of his platforms, like Workday, or Salesforce, where, you know, they kind of have their own languages, and you have to kind of do it over there, because they're dealing with not only their own application, but now they have to think about, you know, performance of your applications and data there, and data sharing and security and all of this. So, you know, that route isn't really as open and flexible as you would like it, you know, obviously, these vendors are going to tell you, you know, we give you 60% Of the remaining 40%, you can burn, but it's not as as many rainbows and butterflies as they make it out to be

David Turetsky: 23:32

No, you're definitely right. I mean, if you want to go beyond that, 40% you really need a stack, to be able to take the data out, put it into something that enables you to ask different questions. And, and you're not going to be able to do that inside of these platforms. It just it just they are they are what they are. And taking the data out the moment you take the data out. Either you're building something around it to keep that security and to keep everything in tune. Or you're gonna have to be okay with it being out. Which is Yeah, exactly. Exactly secure, unless you make it secure. And it's not exactly gonna be up to date unless you make it up to date.

Tarush Aggarwal: 24:17

Yep. And then you have to worry about data lineage, and, you know, keep track of the metadata where it comes from, right, you don't want to have multiple sources of truth, because you can't you take it out and edit it, and it's not going to go back talk to the main source. So, you know, these are examples of like, you know, going back to kind of the first question of why do companies sort of struggle with some of the infrastructure pieces because they don't want to go manage all of this. This is like, pretty complex, right? Like, getting this sort of setup or getting this set up correctly. And so, you know, a lot of these companies ended up going with these vendors, or, you know, just completely doing things themselves. And they, I think we're gonna see a lot of innovation on you know, some of these things kind of coming together and playing Well as the HR and experience kind of gets more and more mature, and part of this is going to come with different sorts of vendors, making some of these processes a lot easier, like, how do you work with both platforms and a hybrid solution. And, you know, you need to worry about the syncing of security, your lineage or some of these kind of pieces.

David Turetsky: 25:28

Tarush, thank you very much. We appreciate your time. And just to summarize, we've been talking about how HR can leverage data stack that enables them to do more than what the one size fits all might be able to provide for them, and how they can ask better questions by being able to take a set of architecture and a set of application and a set of data logic and layers, and not have to worry about just getting it from one vendor. And actually, just not having to worry about it at all. Anything else that you want to talk about before we wrap?

Tarush Aggarwal: 26:08

Yeah, you know, I think, thank thank you, again, so much for having me on the show, you know, excited to see companies get more and more on the, you know, AI layers and, you know, more and more upstream, because that's really where some of these sort of sort of some of the competitive advantages gonna come. And I think the next few years are going to be sort of super interesting around sort of, in terms of those areas.

David Turetsky: 26:30

Well, thank you very much Tarush. Thank you, Dwight.

Dwight Brown: 26:33

Thank you. Thank you for being with us today. Tarush, we appreciate it.

Tarush Aggarwal: 26:36

Thank you, David and Dwight, thanks so much for having me it was a blast being on the show.

David Turetsky: 26:40

Our pleasure. And thank you for listening. And if you have any questions, please don't hesitate to reach out to the show. If you like it, either. Give us five stars or refer us or follow us. Thank you so much for listening to HR Data Labs podcast, take care and stay safe.

Announcer: 26:56

That was the HR Data Labs podcast. If you liked the episode, please subscribe. And if you know anyone that might like to hear it, please send it their way. Thank you for joining us this week and stay tuned for our next episode