Episode Transcript
Speaker 1 00:00:10 Hello everyone. Today I want to talk about a very important topic that's on the mind of every HR professional, every training leader and every vendor, <laugh> in hr, and also a lot of public policy makers. And that is the issue of skills and skills taxonomy. This is a massive problem and opportunity and confusion. And so what I'd like to do for the next 30 to 40 minutes to explain what it all is, first of all, the word taxonomy, or it's sometimes called ontology, is a way of organizing information. So the Dewey Decimal system is essentially an ontology that creates a taxonomy for finding books. Every database, every financial system, every marketing system, every research library, every magazine has some form of an ontology. And it because it helps you find the thing you're looking for, I suppose even a grocery store you could say, has an ontology.
Speaker 1 00:01:13 In the case of skills, we have this enormous problem of defining what people know how to do. And the word skills refers to the capabilities that an individual has to do something fairly granular or small. Now, on my other podcast where I talked about the Global HR Capability Project, I talked about the difference between a skill and a capability. In my opinion, and you can argue this as much as you'd like, a skill refers to a relatively fine grained technical or operational skill that can be validated and measured and probably certified A capability is a broader set of skills that come together to solve a business problem. For example, in sales objection handling is a capability. It's a very complex set of skills that you need and experiences and information and context that might involve skills like listening, communication, note taking, perhaps using industry data, doing research and so forth.
Speaker 1 00:02:20 So what we have in companies is basically tens of thousands or hundreds of thousands of skills that people use every day to do their jobs. Everything from turning on your computer and doing your email and using Excel and doing a financial model and creating a budget, or writing a piece of code or designing something in PowerPoint or designing something in a design system or operating a piece of equipment. There's just hundreds of thousands of these things. Now, what's happened over the last decade since businesses have become digitized, there's been a very clear understanding that there are new skills needed now that we may not have needed a few years ago. By the way, this is not a new problem. When I graduated from college in 1978, I went into the oil industry and we were working on oil refining and the hydrocracker and different chemical processes, which were actually skills that were needed at the time.
Speaker 1 00:03:18 And I guarantee you, every year or two, the oil and gas industry has been upgrading those technologies and, and teaching people how to operate new equipment and creating needs for new skills. Software is not the first time this has happened. It's happened for probably hundreds of years, to be honest, but it's more acute now because we have all this data and all these systems to categorize these skills. And we have these tools that infer skills or try to figure out what skills you have by looking at the jobs you've had, the experiences you've had, the tests you've taken, the courses you've completed, the people you know, all of those things come together to create a consolation of skills. Now, why is skills taxonomy or skills ontology such a big topic? Well, right now we're in a world where companies have a lot of very high powered human capital software.
Speaker 1 00:04:10 They've bought Workday or Success Factors or Oracle or another system, and we have lots and lots of data about our people. And every time we make a decision about a person who to hire, who to promote, how much pay to give somebody, what kind of a raise somebody should get, whether somebody should get a promotion or not, whether we have somebody in the wrong job, why somebody's succeeding, why somebody's failing, something to do with their skills. And many, many articles have been written about the fact that we need to move to a business model where people are rewarded for their skills and their capabilities, not their tenure or their level or their political power or their political connections. I'm not saying that's gonna happen overnight, but it's starting to happen more and more. There are lots of companies where one software engineer will make $500,000 a year sitting next to another software engineer who makes $75,000 a year because the $500,000 a year software engineer just is a lot better at their job.
Speaker 1 00:05:13 In fact, it was Bill Gates who once said that the top five software engineers at Microsoft at the time probably were writing code that was more important than the rest of the thousands of people who worked there. And I believe that's true. There is this idea of a very unequal distribution of skills in most roles, and we wanna know who's really good at what. So we can put them into the role where they can most take advantage of the skills that they have. And of course, as the company changes, as the marketplace changes, as the technologies changes, as the, as the products you offer change, we have to reskill people on a regular basis. And that goes for me too. We're all in a world of continuous reskilling or upskilling, and sometimes we need a lot of new skills because the job we're in is going away, or maybe the career we're in is going away, and we really wanna learn how to do something different.
Speaker 1 00:06:03 So what a skills ontology is intended to do is organize these skills into a hierarchy. In sales, we have lead generation, we have opportunity identification, we have qualification, we might have objection handling. I mean, we might have 20 things that somebody would decide are the capabilities or skills needed in sales. And each of those are gonna have related skills underneath them. And in fact, if you really look at the way the skills ontologies work, they're really not that hierarchical. They're more like what is called a graph database. The skill of objection handling in sales involves a lot of the same skills of communication and teamwork in an internal meeting, which means it's also a leadership skill. So we can't create a hierarchy because if we create a hierarchy, we're gonna have at the edges of the hierarchy all sorts of things that are replicated from job to job.
Speaker 1 00:07:01 Everybody in the company needs to know how to use their email. So do we wanna put email skills into every job description? Of course not. That would be sort of absurd to think about it, but we need an ontology that essentially connects skills to each other so that as you move from role to role, we have a relatively intelligent view that, oh, well, you know, you have about 75% of the skills you need for this new job. There's only a few things different here that you need to become familiar with. Now, given the complexity of this problem, we're at a very early stage of off-the-shelf solutions, and there's essentially three parts of the market. The first is software vendors. Every software vendor from Degreed to EdCast to Cornerstone to Workday to Oracle. Soon SAP Eightfold, on and on and on has a piece of software that tries to figure out what your skills are.
Speaker 1 00:07:57 How does it do that? It reads through words in all of the artifacts you have online and tries to find matches. It's not, that's different from what Google does. So if they read your resume and they read your emails and they read your performance review and they see the word software engineer, many, many times the system might conclude that you have some skills or some experience in software engineering. And the more sophisticated ones do more. They might say, because you worked on this project and this project was about those things, you probably picked up these skills. And because at that point in time, this project was focused on this technology, or if you worked for Google in 1998, you probably used this form of machine learning because that's what was going on at Google at that time. And by the way, if you're connected on LinkedIn to this person and you have a strong connection to that person, well they have skills in this area.
Speaker 1 00:08:50 So you probably understand those areas too. So there's a lot of very interesting algorithmic technology that tries to infer skills. And the more sophisticated tools that do this, I think one of the most advanced one is eightfold actually do a pretty good job of figuring out what you are most likely to be good at. Because while you would expect your skills assessment to be a testing problem, it actually isn't a testing problem. It's more of a context problem. If you've worked in this area and solved this type of problem successfully in the past, that's proof that you probably do have these skills. And certainly they've been validated through real world testing. There's also, of course, operational skills where you're certified to operate a machine, perform a healthcare procedure, conduct an audit, or do other pieces of work. And those kinds of databases are also indexed by these systems.
Speaker 1 00:09:42 So that's the first category of the market software. Now, unfortunately, there are no standards for skills taxonomy software. Everybody talks about how great their tools are. The reality of it is they seem to fall into three categories. The software vendors focused on different areas. One is skills related to learning tools like LinkedIn Learning, EdCast and Degreed Cornerstone. And most of the learning experience platforms try to infer and understand your skills and the skills in a piece of content to match you to the right content. So if you're searching around in the LX P and you type Java 15 different times in different phrases, the system's gonna get pretty smart about the fact that you're either interested in Java or you know something about Java or maybe you're an expert in Java. The second type of software is matching software, which is the what the recruiting tools use.
Speaker 1 00:10:38 So products like Eightfold and Avature and Phenom people and iSims will take the skills in a job description. They'll read it and they'll look at your background and your bio and they'll do a reasonably good job of saying, most likely you would be a good candidate for this job. Or among the candidates in the pool, here are the people that you probably should call back first, cuz they seem to have the skills that are most interesting for this job. And if you look at tools like Gloat and Fuel 50, they're actually using that same technology for internal mobility so that an internal candidate can get referrals or recommendations for jobs based on skills that they might be able to use in another part of the company, which is a massive new industry actually in hr. So those are software tools. The third type of software tools are what I call this hybrid skills inference tools, which I think today are really two vendors primarily, and that's EdCast and eightfold.
Speaker 1 00:11:37 And I think this will soon be the domain of LinkedIn where you look at recruiting and sourcing data and learning and career data together and create a very holistic view of what somebody's skills may be in a more advanced level, not in a position to tell you which of these vendors is better at what, but their design points are different. So that's the software world. By the way, you already have a lot of this software. You're learning management system, your applicant tracking system, your recruiting system, your learning experience platform. If you have Workday or Oracle, your HCM system is doing this, it might be doing it well, it might not be doing it well. Does it understand your industry and your domain? Maybe it does, maybe it doesn't. But the software vendors really aren't trying to get into content so much, which I'll explain in a minute.
Speaker 1 00:12:25 They're really just trying to create great algorithms. And by the way, humans are needed to. It's interesting, if you really look at what goes on behind the scenes at Facebook and Google, there are human beings correcting and improving the search results from algorithms. So while the vendors are gonna tell you their tools do great things, you really do want to have humans involved. I've talked to Ericsson about this most recently. And in the area of 5g, which is a massive new set of technologies and business models and economic models, they're manually going into tools like Degreed and organizing information to make sure that it meets their needs for capabilities. And that's the whole conversation about capabilities versus skills. Okay? So that's the software part of this. The second big broad part of the skills taxonomy industry is data. You can buy skills databases, you can go to IBM and buy something called Talent frameworks, which is a little bit old, but it's out there.
Speaker 1 00:13:24 You can go to Emsi and you can get a database of 20,000 or 30,000 or 40,000 skills, which they collect on a regular basis by crowdsourcing and scraping job descriptions and job postings from all over the United States and Canada. You can get the same thing from Burning Glass, but you can go to companies like nasscom, which is the IT training consortium for most of India and many of the countries in the far east. And you can get a massive database of skills for it. Every possible granular IT skill that you need. You can go to Oracle and get a database of skills for Oracle DBAs, Microsoft and GitHub hub skills for Microsoft Technologies, on and on and on. So if you wanna really make a business out of this, you can collect all these skills databases and you could load them into your skills system.
Speaker 1 00:14:15 And theoretically the software should try to make some sense of it. And many of the vendors have done this. I mean, I give EdCast a lot of credit for their work in working with country governments, industry groups, and a lot of skills taxonomies to bring them together so that their software does try to rationalize them. Because what happens when you look at all these databases is the same skill has different titles and different systems. A skill that says collaboration in one ontology might say teamwork in another one, and it might say getting along with others in a third one, but they're all kind of talking about the same thing. So it does you no good to have all three of those in your system, each pointing different pieces of content or different characteristics. You really wanna bring them together. So what the software vendors are starting to do, and Degreed is doing this and EdCast is doing this through their new skills DNA technology, they're starting to give you tools to rationalize and simplify the ontologies that you have.
Speaker 1 00:15:17 I would venture to say if you turned on one of these skills indexing systems into your human capital system and looked at the results, you're probably gonna end up with a hundred thousand skills, many of which are the same thing in different places, though the software vendors are getting better and better in doing that. By the way, one of the biggest data providers of skills is your company. Your job descriptions, the information you already have about how you do business. We had a call the other day with an oil and gas company and they talked about the need to build a skills model for the future of the energy industry, which will involve solar power and electric distribution systems and different forms of chemical power and hydrogen power and on and on and on. Where are they gonna figure out what those skills are? Well, they already have a massive database of skills and exploration production, refining distribution, probably driving ships and other things like that.
Speaker 1 00:16:08 They're gonna have to find these skills somewhere. So they're gonna go out and they're gonna look at job descriptions. We're gonna help them get some databases, we're gonna help them figure out what those skills are. And once you build that kind of infrastructure, that's gonna be an ongoing thing because all of these domains are constantly changing. Okay, so there's the technology part of this, there's the data part of it, and the third part is the business part. And that's where you come in. Let's assume you find the Cadillac end-to-end system that brings and categorizes the skills you need for your company. And you've decided there's 25,000 skills you're gonna run your company on. What good is that? If the c e O says, Hey, we need to get into this new business, go find me a bunch of people that know how to do this.
Speaker 1 00:16:54 Do you go into your skills database and quickly try to figure out what the top 300 things are that are gonna help you find these people? Probably not. What I think most companies need to do, and we're helping quite a few companies with this right now, is strategically decide from the business level first, what are the capabilities you need in your company to be successful in your market? And let me give you an example of what this means. One of our clients is a software company. And software companies are all in highly competitive markets. It is a a very, very competitive industry in general. And this company is a very successful company that sells software for design engineering in different parts of construction and different parts of the manufacturing industry. Air products were developed for individual buyers and individual users in all of these different industries.
Speaker 1 00:17:47 And they're very, very well known. So they have lots and lots of customers. Well, it turns out that their customer set is changing. The individual buyers of their software that used to be in different companies or different domains of a given company are now working in groups. Project teams are working with, design teams are working with manufacturing teams. There's pre-fabricated construction being built. Companies are building end-to-end systems that include HVAC and architecture and electronics. There's smart lighting systems, there's smart heating systems. So this idea of having individual software products that teach you and manufacture and automate different parts of these individual design areas is becoming less useful. So they need to change their product set so that their products are really more like integrated suites or end-to-end solutions. Well, it turns out that they have plenty of technology people that know how to do this, but the entire company was not designed that way.
Speaker 1 00:18:44 They were really designed to optimize one customer set, understand the needs of that customer set and build end-to-end solutions for that customer demand. Now they need people to understand the cross domain applications of every piece of software they buy. Well, that's a capability, that's not a skill, that's a capability in systems thinking, in understanding an entire industry in having relationships across the company. So this new category of skills, which is a combination of soft skills and technical and professional skills, is forcing the company to move out of the domain of deep technology skills and broadened the horizontal part of the T-shaped career in these system thinking skills. Now, this didn't come from a piece of software. No, no, no piece of software figured this out. This was the CEO and the heads of the business observing the market they're in and making a very strategic decision that they needed to go to market in a different way.
Speaker 1 00:19:40 So that's the third part of this problem. And this essentially third part is going on all the time. When I first became an analyst many years ago and started studying a lot of companies, I came to the conclusion that in every single industry I looked at, there are three types of companies going after the same market. There are innovators who try to pioneer new ideas and break through and differentiate themselves with something that nobody's ever done before. There are executors that try to create the lowest cost and the highest value per dollar. And they may do it in a commodity way, but they always win business because they're cheaper than everybody else. And then there are the customer intimate companies that may not be the most innovative and they may not be the cheapest, but they really know the end-to-end needs of the customer. And they have broader, more consultative solutions to sell.
Speaker 1 00:20:30 And I mean, this is in every industry. This is in software, this is in manufacturing, this is in supply chain, this is in healthcare, this is in financial services. You can always find somebody who's doing it cheaply. You can always find somebody who's doing it innovatively. And you can always find somebody who's kind of like the great solution provider. Well, even though you're in the same market selling the same types of things to the same customers, those three types of companies are going to need different capabilities. A pioneering innovative company is gonna have to know everything about the technology. A low cost solution provider is gonna have to know everything about the supply chain and the end-to-end solution. Customer intimacy company is going to have to know a lot about listening and consulting and agile problem solving. Those are not gonna be the same companies.
Speaker 1 00:21:14 So if you don't work on this third area of the business capabilities, you're not gonna get the best value of all the skills, taxonomy, investment. Now the final thing I wanna talk about a little bit, cause this is a very big topic, is the marketplace. And as an analyst, I talk to a lot of vendors and I love the vendor part of my job. It's fascinating. But the problem with the vendor market for you as a consumer or buyer is it's really confusing. Everybody has plastered the word skills onto their website, upskilling, reskilling, continuous skilling, and it's becoming meaningless. So let me just talk about what's going on in this space. There's essentially several different kinds of providers. The first is what I would call technologies, Workday, degreed, EdCast, Oracle, cornerstone. They sell software. So while they may call themselves upskilling companies, they actually don't do any upskilling.
Speaker 1 00:22:11 They give you tools that help you do upskilling and you need to evaluate them based on their level of domain experience in your industry, whether their functionality is easy to use, whether you like the vendor, whether they can handle your scale or size, whether they do business in the geographies that you're in and so forth. The second companies that are in the whole skills area are content companies. You to me, Skillsoft, Corra, hundreds of companies like this. They are really in the content business. They develop courses, programs, curricula, certifications, tests and various things around that. They understand the need for skills in in some sense a deeper way, but they all have to have a limited view because they can't build skills for everything. Like in our academy for example, we have 91 capabilities for hr, just for hr, and I'm not kidding, there are 91.
Speaker 1 00:23:09 And you will look through them and you will say, yep, they all fit. And so we've built now hundreds of hours of content aligned towards these capabilities just in hr. So you go to it, you go to finance, you go to sales, you go to marketing, you go to seo, cloud computing, on and on and on. There's a lot of content out there. Each one of those vendors would like to sell you a skills model, but the chances are it will be of limited value over time because your company's gonna need things that are gonna cross between these vendors. And that's where the software vendors come in. The third part of market is really to me, consultants and I spent almost seven years at a consulting firm. And I know that sometimes you just need help. And uh, the consulting firms have done a reasonably good job.
Speaker 1 00:23:56 I don't think a great job of helping companies figure this out. We happen to be one of them, by the way. We do, you know, a pretty good amount of consulting. And I think the problem consulting firms have generally speaking is they're really good at understanding problems and putting together teams to solve problems. But they're oftentimes not experts on the technology providers in the market. So you really as a corporate buyer, have to do your homework on the technology part of this, on the content part of this, and then use the consultants as best you can. My suggestion to consulting firms is that they go after vertical markets, become an expert in oil and gas, become an expert in retail, become an expert in healthcare, become an expert in supply chain, and year after year company after company, the consulting firms will get better in those domains.
Speaker 1 00:24:46 Finally, let me add one more thing and that is tools. One of the reasons that I think the skills taxonomy market and the skills ontology issue is challenging is we don't have enough tools yet. You could read the marketing on Workday Skills Cloud or Eightfold or Degreed or any of them, and you could sort of believe that they have everything, but actually they don't. They're really relatively new technologies. They're very, very quickly advancing in inference and in connections and creating smarter AI driven skills. Taxonomies automatically. They are not very mature yet for emerging skills and blending skills When they sound the same, they're not very good at matching skills to job history or career history. Some of them are, some of them aren't. They're not very good at helping you rationalize and decide which skills to keep and which skills to get rid of. And they're not necessarily that good yet at ingesting skills models from other companies.
Speaker 1 00:25:46 They're all working on that. And so I think the, for the market that's gonna be pretty exciting in the future is what I would call off the shelf tools for building your skills taxonomy. One of the companies that's really been working on this is EdCast EdCast, which is known for its L X P product, has built a set of tools called Skills dna, which will be introduced soon that I think you ought look at. Uh, LinkedIn is building a set of taxonomy tools for all the LinkedIn learning content and the LinkedIn recruiting content that I think will also be very interesting. Uh, I have to believe that Workday and Eightfold and the other companies are gonna give much more open tools for skills analysis and skills development and skills architecture as well. One more quick point, and that is the issue of assessing skills. For all of my time as an analyst, I've had debates with people about assessment, and you might believe that testing and assessment is critical to validating skills.
Speaker 1 00:26:45 Well, it certainly is for operational skills because if somebody can't demonstrate they know how to do a healthcare procedure, you probably don't want them sitting down in front of a patient. And that includes running a drilling rig or driving a crane or even delivering a package, frankly, if they don't know how to drive the truck. But for most skills, you're not gonna be able to do that. You might give somebody a test at the end of a course and you might get a reasonable answer, but you're going to have to learn the capability levels or the proficiencies of people through data. And my experience with this is that as your skills taxonomy advances and becomes more sophisticated, you are gonna get smarter and smarter at figuring out who the highly skilled people are. For example, if you go to LinkedIn, there was a system in LinkedIn where you could recommend people based on certain skills.
Speaker 1 00:27:37 In the early days, we kind of thought it was gonna be a useful tool for figuring out who knew what. Well, it turned out it really wasn't that useful because a lot of people were just recommending skills for people just because they felt like it. Now, what's happened in LinkedIn that the system has become so big, there's 740 million people in there. LinkedIn actually knows who the experts are because of the connections, the experiences, the level of domain expertise based on who they've worked with and what jobs at a pretty sophisticated level. Now, they haven't exposed all that data to you as a corporate buyer yet, but you can do that in your company, and this is what recruiters do every day. By the way, when a smart recruiter goes out there and looks for candidates, they don't just read their resume, they look at their history, they figure out who they worked with, they looked at what companies they worked on and where, where they worked, when they worked in each company, and who their references are and what their references said about them.
Speaker 1 00:28:31 And really good recruiters can do a pretty good job of figuring out who's a highly capable person in a given area. And that's really where this is going. So I, I would just encourage you, don't get too wrapped around the axle on validating or assessing skills. You absolutely need to do it for operational jobs where the skills must be validated, but I think you'll find that the technologies will be able to do this better and better on their own with your guidance. The final thing I would say just on this whole topic is this is not a simple problem coupled to this problem is the problem of job architecture and too many details being put into job descriptions, which is also being addressed. So I think if you're a reasonably good sized company, the way to solve this is assemble a team of three to five people that own the skills taxonomy for your company.
Speaker 1 00:29:20 We can help you get smart on this. Get to know the vendors, get to know the technologies, get to know the databases. Let's come together and talk about it. We have a whole big reset group talking about this right now, and I think you'll find it becomes one of the most valuable infrastructure groups you have in your company. We need infrastructure for security, we need infrastructure for data, and we need infrastructure for skills and the skills and capability taxonomy. I know this is just the introduction to a very complex topic, but I hope this was a good overview and please call us if you'd like any more assistance with this really, really important area.
