Episode Transcript
Speaker 1 00:00:07 Okay, today I'm gonna talk about a really, really big, complicated, important topic, the changing and evolving role of data in hr. And before I get into all the details, let, let me kind of give you a high level concept. And that is that in most companies, certainly software, IP services related companies, retail companies, most distribution companies, the largest cost of the business is people. It's payroll and benefits. About 60 to 61 to 62 to 63% of payroll goes to the employee. About a third of that goes to insurance and benefits. It's a massive amount of money. It's trillions of dollars around the world. So you would expect that the cfo, F o had good numbers on that spent, no, they don't. In 99% of the companies I talk to, they have much better data about the customers, the products, the parts in the warehouse, the supply chain, the financials, the account receivable, the accounts payable than they do the people.
Speaker 1 00:01:12 By far, it's very common for me to talk to a company that doesn't know how many employees they have that doesn't know how many contractors they have that doesn't know how much money they're spending on overtime. That doesn't know very well why people are quitting or why people are performing poorly. They don't have that data. It's hard to get it. Now, why don't they have it? Because there are hundred and hundreds of data elements about people. There's the operational data, obviously how many people we have and who's doing what job, and what hours they worked and how much they're getting paid. Then there's all of the HR data. What is their performance rating? What is their tenure? Are they ready for promotion? What's their employee engagement scores? Are they trained on this? Are they trained on that? When was the last time they went to a course on, on, on compliance, et cetera.
Speaker 1 00:02:01 Have they had an accident? Has there been a harassment claim, et cetera. And then there's performance data. Who's performing at a high level, who's performing at a low level, not just through performance ratings, but through real information about performance. How many lines of code have they generated? How many things have they sold? How many widgets have they produced? How many customers have they taken care of? That data's out there. It's not in hr. So we're dealing with a very, very messy problem. And not only is the data in many places, it's in many systems, but the definitions of the data are unclear. So most of the big HR vendors, the HCM vendors who have tried to build and sell integrated analytics tools, have basically been not that successful because the exception of success factors, who bought a company that did this, that data isn't easy to find.
Speaker 1 00:02:53 It's located in lots and lots of systems. The training system, the recruiting system, the compliance system, the operating systems in the, in the business. But there's also data about their workday, their data in Microsoft, on the Microsoft graph, how many, how much time do they spend in meetings? Who are they communicating with? And that kind of social data, it's very, very valuable. There's organizational network data. What teams are they a member of? Who are they communicating with? More or less? There's data about the type of communications they're having, which is called passive listening. Is this person talking more to this person versus that? Are they angry? Are they upset? And, and basically this is gonna keep growing because now pretty soon we're gonna get data about their behavior in meetings and voice and video and all that stuff. So what are we gonna do about all this?
Speaker 1 00:03:42 Well, I, I would say there's essentially two categories of problems. First is the data is in many places. So regardless of whether you buy a big HCM platform and by say, Workday or whatever it is, that's maybe 15% of the data you need, 20 at the most. So that system isn't really the system of record for everything. It's a system of record for some things. So you need a federated tool or system that can pull data from many sources. But the second problem is metadata. What does it mean? The definitions? How are we gonna define turnover? What does this payroll number mean? Is it the hourly rate? Is it the annual rate? Is it the annual adjusted rate based on time of year? I mean, those metadata definitions of these data elements make or break these analytics systems because you can't trust it if you don't know how it computed.
Speaker 1 00:04:37 What it computed. And let me just assure you that this is very much an unsolved problem. Let me, let me just mention a podcast that I heard just yesterday of Mark Zuckerberg talking about the layoffs and the performance and productivity initiatives going on at Facebook at Meta. And he made the comment that in February of 2023, which was this year, he found out that the average manager had three people reporting to them. And by the way, they have a big people analytics team there, and he was surprised. He made all sorts of excuses for that. And they decided to do a big layoff. They've laid off 25% of the company since this layoff period started. They clearly did not have good data about performance, and they didn't have what we call systemic data. They didn't have integrated data to make decisions. They probably had lots of pieces of information, but they weren't brought together.
Speaker 1 00:05:32 So le let's just agree that this is a complicated problem and that buying a new platform or HCM platform is not necessarily a solution. Now we've been, um, doing a bunch of interviews of companies about people analytics because it's a fascinating area and most of you who know me know that I, this is one of the first things I did 20 years ago is get involved in this originally in learning analytics. And I wrote a book called the Training Measurement Book and then later did many, many studies of people analytics and all the different dimensions of it. And it is really not a simple problem. So where are we? Well, now that we're entering the age of AI and big data systems are easier and easier to get, there's some really interesting new things that have come to market and they've fallen into several categories.
Speaker 1 00:06:21 The first is the original concept of a data warehouse, which by the way, I was involved in designing a lot of these when I was at Sybase years ago, was a relational database with very strictly defined rown columns and fields that somebody designed for you or a vendor and you put data into it. And those are great systems in theory, but these days they don't keep up because the number of data elements and the volume of data changes so fast. And some of the data is not actually tabular data. So even though I know a lot of you have data warehouses and they're probably very well used, those are one piece of the solution. The more extens, extensible, expansive systems usually allow you to put many types of data into the system. And so what we need is we needed more flexible architecture. Now, it turned out, back in the days of relational database praise, there was a technology invented called a multi-dimensional database.
Speaker 1 00:07:28 Most of you don't know what this is, but when you open up Excel and you click on a pivot table, it's essentially a multi-dimensional database. What it does is it takes lots and lots of data elements and it cross computes or dynamically computes all of the cross products between all of them. Sounds nuts, but it works really, really well. And what it allows you to do is say, I'd like to look at turnover by gender. I'd like to look at turnover by tenure. I'd like, like to look at turnover by age, by gender, by location, by city, by manager. And rapidly, almost instantly give you those answers. You can do it on a relational database, but it's a lot of queries and it takes time and kind of costs a lot of money. So these multidimensional systems entered the world. They entered the world 20, 25 years ago, and today they have spawned some very specialized tools.
Speaker 1 00:08:19 Unfortunately, as most of you know, the general data analytics tools, Tableau, click, view, et cetera, are not really that good for hr. The reason is they don't know anything about the domain. They're good tools for managing data in general, but they'll do you no good if you don't have the right definitions, if you haven't cleaned the data up, if you haven't connected the data pieces to each other. So the real problem, as I said, is not collecting it, but it is making sense of it. And so the vendors that have really nailed this the best are really only a few. And the one that we've been spending a bunch of time with recently is a company called vizier that a lot of you know are also others. There's a company called Chart Hop. There's a company called One Model. There's a company called Cruncher that have similar technologies to Vizier, but Vizier is way ahead of the others.
Speaker 1 00:09:15 So I want to talk about Vizier as an example, and also to just give them a little bit of credit for what they did. So what these guys did, these are technologists that originally were involved in the development of Cognos and some of the early multidimensional databases. They decided they were going to build a specialized analytics system for people data for hr. And you know, we can call it analytics or we can call it data management, we can call it whatever we want. And so what they did is they built a technology that allows you to collect and aggregate and extract data from any HR transactional system. And there's a lot of vendors, there's hundreds of them and a tool to transform and load that data into a multi-dimensional layer with definitions. And they have built the definitions for virtually everything that you're going to analyze in hr.
Speaker 1 00:10:08 And you can customize those definitions. Now that sounds kind of clunky and kind of, you know, expensive and messy. And you might say to yourself, well, we just bought Workday, we just bought Oracle. You know, we have Tableau, we have ClickView, we have a data warehouse. Why do we need one more tool? And that's the problem is a lot of people think that way. Well, the reason is, if you wanna do this yourself and really build an integrated data set that's useful for hr, you're gonna have to do everything that vizier did by hand or you're gonna have to hire a consulting firm to do it, and they're not gonna do it very well. Plus they're not gonna build it as a product. They're gonna build it as a, as a one time shot. And it needs to be fast and it needs to be easy to use and it needs to have graphics and charting and so forth in it too.
Speaker 1 00:10:53 So, so these vendors like Vizier and Chart Hop and the others, they have done this. Now you may say to yourself, great, that's good for really big companies who maybe they really think they need this stuff. Let me go a little bit further here and argue that this is an essential tool set or platform for the next couple of years. Now, we've been talking a lot about systemic HR and the whole theme of systemic HR is what we call falling in love with the problem. Don't go out there and do a training program or an onboarding program or build a skills taxonomy or whatever as a standalone project. Do it in the context of understanding the problem. And if you don't have good data, you don't understand the problem. Let me just give you an example. One of the largest manufacturers of batteries has about, uh, nine or 10,000 employees.
Speaker 1 00:11:49 And they're growing. They're gonna double in size of the next few years as you can imagine. And they do a lot of manufacturing. And the woman who runs a lot of the HR technology and HR analytics for them who's really fairly brilliant, actually, she wanted to work with manufacturing on productivity. And she started to look at the data about who was working what hours, who is locking overtime, who is getting paid bonuses, who is quitting, who is coming in late and relate it all to the output of the factories of the manufacturing lines and the quality. She has the quality data cuz she works in the manufacturing organization. And she discovered some absolutely spectacular things. She found that a lot of the bonuses and overtime that was being paid had no relationship at all to the output or the quality of output. And so that managers who did not have good data either by the way, were making really poor decisions on how to spend their money.
Speaker 1 00:12:47 And she said, when I first showed the raw data to the managers, all they did was argue with me and tell me why. While that data's no good, let me think about it, come back later. She said, once I showed them the information in a tool set like vizier and I could dive in and drill up and down and show them the relationship between this data and other factors, she said there was no arguments in about 15 minutes. They believed everything she said. And this is the problem you're gonna have. And I've seen this problem for years in all the work I've done, is that if you don't have a quality trusted source of information and you go marching into some manager's office and tell him or her that he's got a turnover problem or this or that, and he has to look at the data or she, and you can't produce something that they understand or believe in or they find an error, you're gone.
Speaker 1 00:13:38 They're just gonna say no, they're not gonna listen to you. So quality and meaning and integrated data is critical. And the reason it's even more important now is we're entering, as I'm gonna be talking about at our conference next week, we're entering a period of time for the next maybe decade or longer where there's going to be a shortage of workers, there's going to be a shortage of skills, and you're going to have to make systemic decisions when you hire somebody. You're gonna have to trade that off against retraining somebody internally, changing the pay to improve retention, possibly redesigning the team and those strategic consulting roles that of course you wanna play as an HR person. You can't do them without information, without data. You're gonna be looking around for data and you're not gonna find it. Are you gonna spend months looking for it, trying to aggregate it together?
Speaker 1 00:14:29 And then you're going to come to some conclusion and realize, well, it's kind of right, but I'm not sure what I don't know. And so this is a massive issue. Now, you would think that after billions of dollars have been spent in eight cm and probably billions of dollars in r and d amongst all the big vendors that this problem would've been solved. It's not, Workday doesn't do all this. Success factors doesn't do all this. Oracle doesn't do all this. They have good tools for analyzing data. Yes, but you have to decide what the definitions are and you have to build the right data structures and then you have to put it in a form that people can find it and use it. And that's very, very hard work. And frankly, most IT departments frankly have other things to do. You go to them with a three year project to build an integrated data management system for people data and all these other extraneous new forms of data, they're gonna say, well, we're working on the new AI strategy or, or we're, you know, working on our website or car CRM or whatever.
Speaker 1 00:15:29 They got plenty of other things to think about it. It does get done occasionally, but very, very rarely. And another pause or rather problem with this situation is that I talk to a lot of companies that tell me, well, you're gonna replace our old HCM system with a new one because then we'll have integrated data. Well, no you won't. Maybe you'll have better data about the data in the hcm, but that doesn't mean you're gonna know everything about the employees in your company at all. So these tools like vizier and the others are really, really high value systems. Now let, let me fast forward to the next part of this and that is ai. We're gonna be, um, launching our AI deep dive white paper at the conference next week. So you'll be able to get it a couple weeks later. But if you think a little bit about AI and what it is and what it does, these large language models in these neural networks are essentially gigantic data management systems.
Speaker 1 00:16:28 The data might be a number, the data might be a token, the data might be an object, an image, a piece of a video, a piece of a sound bite. But basically they're doing the same thing. And if you look at the talent intelligence tools, eightfold seek out gloat, beamery, sky hive and others, what they're really doing is they're taking structured and unstructured data and they're using neural network type of technology to give you much, much, much better understanding of it through models. And what the AI does that the typical analytic systems don't do, is they learn from the data what is going on. So you can literally take a large dataset with no definitions because the system doesn't know if it's text or, or image or, or numbers in this system will predict based on queries or based on patterns, how to classify that data and what patterns are within it.
Speaker 1 00:17:28 In fact, there's a new branch of AI and some new tools coming out on something called graphic probabilistic analysis that can look at large sets of data and see the edges and the convergence and the patterns in the data automatically essentially from algorithms. So one of the reasons that it is really valuable to get this people analytics stuff cleaned up is that you will be able to use it for ai. Now, you could make an argument that the talent intelligence platforms will replace the people analytics systems. That's not necessarily true and I really don't believe that is going to be possible. These kinds of drill down queries are relationship between engagement and tenure and pay and who came in at what time and so forth. Those are very granular decisions that need to be made. Like what went on at the battery company, we also talked to the head of HR analytics at a large e-commerce tech company, one of the largest.
Speaker 1 00:18:26 And he said, until we had an integrated analytics platform, I couldn't get anybody in the company to even pay attention to things like employee engagement. No one, no one thought it was even meaningful until I could show them the the relationship between some of these engagement surveys and things we were doing and business measures and performance and pay, by the way. And then there's pay equity, which I haven't even touched on in this podcast, but how are you gonna do pay equity analysis and compensation analysis if you don't have good data? And once you do it, you're gonna want to analyze it and correlate it to many, many factors. Gender, race, time in the company, age, location, and other factors. And you're not gonna have a system to do that either. So this new world of what I call systemic analytics is sitting here waiting for you to do something about it.
Speaker 1 00:19:13 What I find in most companies is they don't understand the complexity of the problem. They aren't necessarily looking at the right vendors and they tend to believe that either a new E r P is gonna solve this problem or they're gonna hire a consulting firm, they're gonna build some big data warehouse and it's gonna get taken care of. I personally have found that those are not the right answers and I think this issue is more urgent than ever. I would encourage you to talk to these vendors I mentioned on the podcast, poke around, call us in some cases a talent intelligence tool will give you the information you need and you won't have to do a project like this. But frankly, if you're a big complicated, relatively distributed company and you have business partners out there making strategic decisions, they're gonna need this. Talk to the head of HR at LinkedIn last week about systemic hr.
Speaker 1 00:20:04 And we were talking about a lot of things and one of the things I asked her was, what do you think the number one skill is that's missing in the hr uh, professionals? And she said, analytics. And I said, well, what is it about analytics that don't understand? Is it data management? Is it statistics? Is it charting? Is it storytelling? Is it what she said? Well, not really. It's not really any of those things. They don't have access to good data and they have a hard time finding it. So as you can see, things that look like problems with analytic skills are really not problems of analytic skills. They're problems of data management in creating an analytic strategy. So as we dive into AI and get all gaga about all the things happening in ai, which are gonna be wonderful for hundreds of reasons, let's not forget the basics.
Speaker 1 00:20:52 And that is that a sound infrastructure with a tool set like vizier or chart hop or one model or cruncher or others, can pay for itself dozens and dozens and dozens of times over in many, many, many projects and totally empower your business partners to do things that they wanna do but can't do because they don't have the right information. So we are happy to talk to you more about this. This is an area of my personal interest and we know a lot about the technologies and a lot about what works in different situations, but I wanted to make that point before we enter into our conference. So one more thing. Next week, two big things. The summer solstice and the irresistible 2023 conference, for those of you that are coming and the conference is full, it's about 450 people, you're gonna have a spectacular time in Southern California.
Speaker 1 00:21:49 Not only are you gonna see some of the most beautiful sunsets in the world, in some beautiful spots all over the USC campus, including the Olympic Coliseum, but you're gonna meet some incredible people and we're gonna unleash some new research. I'm gonna show you some new things we're doing in our company and we're really gonna learn a lot together. So for those of you that are coming, I hope you're really excited and I look forward to seeing you all there. As always, if you have any questions about this whole issue of data and analytics, please reach out to us. Have a great week.
