Workday Skills Cloud: A Big Idea With Much More To Come

In the Fall of 2018 Workday introduced the Skills Cloud, a new offering designed to help companies create a “skills ontology” that discovers the skills (and skill gaps) in your workforce. While it sounded like a fascinating concept, we hadn’t heard much about it until Workday Rising this Fall.

Well, I just spent a day going into details with Workday and I want to give you an update.

Let me summarize by saying this is a very big deal, and it delivers on one of the biggest promises in human capital systems:  the move from “systems of record” to “systems of capability.” And I think the term “Skills Cloud” understates what this will become.

Creating a Job and Skills Ontology

First, let’s start at the core. The Skills Cloud (and related products) are trying to identify thousands of skills and experiences in your workforce and arrange them in an Ontology.

What is an “ontology?” If you look at the word in the dictionary, it is described as a “categorization scheme” for large amounts of data.

In Human Resources, we use ontologies all the time. Why? Because we have to arrange, document, and categorize lots of information about people.

Look at the whole concept of a job. The reason we have “jobs” is because we need to get work done (ie. designing, building, marketing, or selling something), and we want to decompose the work into tasks. Jobs were designed to “bundle” these tasks, and we ended up with what we now call a “job description.”

Whenever you write a job description, you identify the functional area (sales, marketing, engineering finance), the job role (tech, managerial, support), the level, reporting structure, and details of the job itself. As you write all this down, you end up describing the job tasks and responsibilities, skills needed, certifications required, experience desired, and more. This becomes an ontology.

If it’s a job in engineering, for example, you may specify knowledge of the C++ language, an understanding of data structures, and the ability to write and debug code. If it’s a job in sales, you may look for candidates that can develop rapport, ask penetrating questions, persuasively propose solutions, and ask for the order. If it’s a job as a manager you’d specify abilities to interview people, select candidates, set goals, and so on.

You get the picture. The problem is quite complex. Every job is a little bit different, and the way we describe it can vary.

In accounting, this kind of categorization is easy: there are Generally Accepted Accounting Principles. In HR and with people, however, there is no real standard – so companies build many types of job descriptions and then use many forms of assessment, testing, and interviewing to evaluate people. (The pre-hire assessment industry is over $2Billion in size and tremendously fragmented with tools and types of assessments.)

Some companies use criteria like education, prior work experience, and prior employer; others use technical or professional skills; others evaluate people based on soft-skills (PowerSkills), potential and other attributes. Whatever scheme you use, it’s yours to define and you end up creating your own “Ontology” over time.

Competency Models: They’ve Come and Gone

To try to make this easy, the pioneers of organizational design came up with the idea of competency models:  a detailed breakdown of job tasks, each of which demands skills, competencies, and behavioral strengths. (Often called KSA’s, Knowledge, Skills, and Abilities).

In my early days as an analyst, I read books about competency models and found them fascinating. They’re developed by evaluating jobs, and then studying what successful people in these jobs do. They are based on the idea that a “job” can be designed by a designer, and we can specify what work each person will do.

In the real world, however, work is much messier. While we usually have a job title, the work we do varies widely from day to day, project to project, and year to year. As the company grows and changes, every job tends to adapt. And today, technology is eliminating more routine work than ever.

So a competency model developed five years ago may be quite out of date today, which is why companies like LinkedIn, Indeed, EMSI, and BurningGlass constantly study the latest skills in demand, by looking at real-world job requisitions being created every day.

To make this even trickier, the hierarchical nature of a job has changed. Rather than work directly for a “boss,” many of us now work in teams, on projects, and hold a variety of roles. In service-related industries (almost all industries are now becoming service industries) your job adjusts based on need. So while you may have been hired to “design” or “build” something, you’re likely to be listening, adapting, and adjusting your work output all the time. And every year there are new automation and digital tools to learn.

There is also enormous variation in jobs between companies. A sales role at SAP is very different from a sales role at Michelin Tire, for example. And even well-defined jobs in engineering, finance, HR, and IT vary widely based on company culture.

Some companies, for example, value speed over quality – so they expect quick results and look for people with deep domain experience. Others value innovation and creativity, so they want people who can think outside of the box. Still others base success on quality over speed (Boeing is facing up to this right now), so they expect you to focus on ponderous detail. So while an off-the-shelf competency model may seem like a fit, the true drivers of success in one company are quite different in another.

And thanks to technological change, the most basic elements of work keep changing.  If you’re an engineer and don’t keep up on your trade, you fall behind quickly. Studies show that “the half-life of skills is five years” which is a frightening thought if you think about it. (Half of your years of expertise go out of date in five years?)

My experience shows that today your “ability to learn” is one of the most important elements of a job. Almost every day I find something I need to learn more about, so I feel like I”m “reskilling myself” on a constant basis. So my “job description” has to adapt, and we need HR systems that automatically keep up.

Then there’s the word “Skills,” which I find misleading. We tend to use it in a sloppy way, often ignoring the granularity of “what a skill really is.” Is “creating a pivot table in Excel” a skill? Or is “building a predictive model” a skill? 

I like to use the word “capability,” which describes how people use skills to solve problems at work. A skill is an atomic item, and skills plus experience create capabilities. So while Workday calls it system the Skills Cloud, I hope it goes well beyond “building a pivot table” and can understand these higher level ideas.

Skills plus experience create capabilities.

In the model we’ve built for the Josh Bersin Academy, we describe what I call Full-Stack capabilities – solutions you know how to build and deliver. These capabilities are dependent on many skills, which in turn are related to each other. And among these skills, some are hard and some are soft (which I call Power Skills), so technical skills are never enough.

We also have to remember that experience is one of the most important skills of all. Nobody learns a skill until you have to use it, so it’s your projects, assignments, and roles that define your success. So implicit in a discussion of skills is the fact that we have to measure experience and include this in the ontology too. 

When you interview a candidate, for example, it’s likely you’ll ask the person to “tell you about your experience in this job” or “share a situation where you’ve done this in the past.” This is because real-world experience IS a skill, so we have to consider job history, successes and failures, and the people you’ve met and known as part of your “ontology” too.

Your Company As A Collection of Capabilities, not Individuals

I’ve been thinking about this for a long time, and it’s a very important topic. If you think about your organization or team – it’s not a collection of people, it’s a collection of skills, all of which translate into capabilities. It doesn’t really matter how many people you have, it matters what they’re capable of doing. The Skills Cloud, I would argue, must expand to understand this bigger picture.

I once asked the head of recruiting for one of the leading energy companies “you do a lot of recruiting, what are the key drivers of success in your candidates?” His answer was surprising: “the single most predictive factor in a new hire’s success is the recruiter, nothing else.” I had to scratch my head.

“What we’ve found is that our top recruiters really understand the skills, background, experience, and culture of our company – and they know how to select the right person that will thrive.” In other words, the top recruiters are “human Skills Clouds.”

Can you imagine a human capital system that understands all of this? That’s the whole idea.

What if your HR software could identify the true skills and experiences in your company, identifying the characteristics of the highest performers? This information could be used by managers, individuals, and executives to select people, develop people, and create powerful career paths for everyone.

It’s not a new idea: many vendors have gone after this market. Not only have LMS companies tried to build these tools, more than ten years ago I talked with a company that built its own “skills cloud.” The platform was a fancy assessment platform that let you self-assess your skills, and then let managers and peers assess them as well. It was elegant and magnificent, but nobody wanted to buy it. It was too much work to implement, and companies just didn’t focus on this area. Today self-assessment tools are built into many systems, but most companies tell me they just aren’t used enough.

What if we ask managers to create “individual development plans” and tell leaders it’s their job to assess and improve skills in their teams. This makes sense too (many books have been written on this), but to be honest, it shows mixed success. Most top managers get promoted for “doing the job well” and they may or may not know why and how they succeed. A sense of self-awareness and the ability to develop others is itself a rare skill, so most companies tell us that fewer than a quarter of their managers are good at developing people.

How about putting the burden on L&D?  Let’s ask the Chief Learning Officer to assess the skills in the company and build a set of Capability Academies to push the company’s skills forward. This is what most good companies do (there is almost always a Sales Academy and Leadership Academy). But this is a messy, complex business … can’t we just buy a piece of software that does this automatically?  After all, AI is so powerful it can tell us which route to drive to work, can’t it tell us what skills we need?

Enter Workday Skills Cloud

Enter Workday Skills Cloud, one of the industry’s most ambitious attempts to assess and categorize skills through software. Workday, through its acquisition of Identified (in 2014) got interested in Ontologies long ago.  After all, every Workday customer has to build the job catalog, so this problem of “creating a job and skills ontology” is going on in every single Workday customer.  So why wouldn’t Workday try to do this in an intelligent, automated, data-driven way?

If you think about the problem, there are essentially two ways to solve it. 

Approach 1:  Define an organization, job, and skills model and then use software to “apply it to your company.”

This is the approach IBM takes with its Talent Frameworks (an exhaustive set of job descriptions and capabilities developed by IBM). You sit down with your leaders and more or less “design” the skills, experiences, capabilities, and job levels in your company. And organizations do this all the time.

While this is exciting and fun to do, it’s quite difficult and takes enormous amounts of time. And as you proceed you realize that the instant you’re done it starts to become out of date. So a second approach is starting to take hold.

Approach 2:  Develop a system that “self-describes” the work, skills, and capabilities you need to succeed.

What if the system automatically figures out what skills you need in real-time? By reading new job descriptions and understanding feedback and results, this data is floating around in your company right now.

Think about “Waze” vs. “Google Maps.”  Google Maps navigates the world based on a lot of hard work measuring and photographing roads. It’s similar to sitting in the conference room designing jobs and competency models. 

Waze, by contrast, “watches” where people go, and can “find out” what roads are open, closed, fast, and slow. Google Maps tells you how the world is supposed to work. Waze tells you how it really is working.

So what companies really want is the Waze of jobs and careers. Just tell me what successful people are doing in the company TODAY, and show me how to get there from here.

And this is what Workday Skills Cloud is setting out to do. 

(By the way, companies like Eightfold.ai, PhenomPeople, Degreed, Edcast, and others are working on this too.)

How Does This Work?

The Workday team has been working on this for several years now, and they’ve built a very interesting system. And you have to think about it as a whole system. It’s not an application sitting on top of Workday, it’s whole new infrastructure within Workday that considers all aspects of work in the context of skills, capabilities, experiences, and relationships.

Architecturally, the Skills Cloud is a set of powerful search, matching, and AI-driven prediction algorithms. It doesn’t really know what a skill is, but it knows that people with “software engineering” roles also have “C++ and Java” associated with their job descriptions, feedback, and other communications. It’s somewhat similar to Google Search – it clusters words together based on patterns, relationships, job history, and other measures of “proximity.”

Consider, for example, a successful salesperson in your company who over-exceeds their sales quota, manages large accounts, and successfully moves into management. The Skills Cloud may discover that this person went to certain schools, studied certain topics, and spent a lot of time learning, discussing, or writing about account management, consulting, and marketing. This person’s “cloud” would identify these skills at a deep level, and others could use this information to improve their own levels of success.

It’s a complex problem, one that has been going on in Workday since the acquisition of Identified in 2014.  Google for Jobs and Indeed have been doing this for job search (“find me an engineering job for my level of experience within 10 miles of my house), but the Workday problem is even harder. Not only does Workday want to “fit people to jobs,” they want to help people find the new skills they need, the new assignments they should consider, and even the people they should meet.

Once turned on it has amazing potential. With more than 275 customers live, the system is collecting data now. It will soon be used to help search for critical skills, help people improve their capabilities, and help executives understand skill gaps in their organizations.

The system has three main components:

1/ Skills Inference

The first is the engine that finds, categorizes, matches, and evaluates skills. This part of the system reads job descriptions, feedback, and many other sources. It’s a little bit like what happens under the covers of LinkedIn. Without you knowing it, LinkedIn recommends you jobs, associates, and learning based on your own experience, conversations, and resume.

Where does the data come from? It looks at current jobs and job transitions, courses completed, projects and assignments completed, talent and performance reviews, public feedback, and resume and job history. Of course, companies have to “opt-in” to the Skills Cloud, so this won’t happen unless you want it turned on.

Degreed, for example, infers your skills from the content you click on. You can also be “tagged” by others (EdCast, LinkedIn, Gloat and other systems do this).

2/ Skills Verification, where users and others can verify the skills on your profile. This includes features that let you and others verify your skills through experiences and feedback.

Users have the ability to view and edit this profile, so while the system may infer that you know a lot about machine learning because you just worked on a machine learning project, you can always “dial down” that it discovered. It will build this evidence through your job history, evaluations, and other data in the system.

While the system does not let you endorse your own skills, it prompts managers and relevant indivisuals (project owners, instructors) to endorse your skills in the context of work (gigs, projects, training, jobs). When I worked at a consulting firm we “endorsed” people informally – with a system like this we could quickly see who the experts really are.  (EdCast, Degreed, and other systems do this.)

3/ Skills Strength. If you haven’t demonstrated the use of a particular skill in some time, it may decline in value in the system. I don’t know how Workday decides the “aging” of skills (in my case I seem to get better at things after I come back to them later, but who knows), but this is an important signal as well.

How Will This All Work?

The Workday Skills Cloud is already live with more than 275 customers. Later this year it will be used by a variety of applications, including the Workday Talent Marketplace, Recruiting, Learning, Career Hub (a career recommendation platform) Planning and Skills Insights (a system similar to LinkedIn Talent Insights), and apps for project skill identification and skill scheduling.

And if this works well, it can be useful for many operational issues. Suppose an employee quits or takes leave and you need a replacement? A manager could quickly search for someone with the required skills, experience, credentials, or interests. This is the real future of a dynamic talent management platform.

This Is A Big Deal: And Competition Is Coming

As I mentioned above, this is a big and important project. Not only could Skills Cloud analyze and categorize vast amounts of data, it will change the fabric of the HCM system. One could say that if Skills Cloud works, Workday is no longer an ERP system, but now a “system of capabilities” – one that can constantly evaluate, measure, and improve the capabilities in your company.

As Google has learned over the last twenty years, categorizing billions of words into an ontology is a serious effort. Workday is essentially doing something similar – finding ways to group and understand skills without any formal direction. 

As I talked with the product team it occurred to me that the most useful part of the system may be within a functional area.  In sales, for example, are there a set of deep and unknown skills we have in our company that drive high levels of success? How do our manufacturing engineers and supervisors compare to others in our industry?  These types of functional models will be extraordinarily valuable over time.

While Workday is a big and amazing company, there is some serious competition work on this too. There may be a “war of the skills clouds” ahead.

First, companies like Degreed, EdCast, Valamis, and others in the LXP space are all working on skills inference. They have millions of data records from their learners’ activity, so they can infer skills and skills by looking at skills demand.  As Workday Learning becomes more popular the company will be able to do the same – but today these systems are “skills systems of record” too. They don’t have the range of data Workday can access, but they’re smart companies so they’re going to try the same things.

Second, Microsoft and LinkedIn are working on this too. LinkedIn doesn’t have internal data for employees, but today with LinkedIn Talent Insights you can already see your company’s skills vs. your competitors. Granted their skills data is all through recommendations, but the company is now offering its skills assessments so its database will improve. Products like Microsoft Project Cortex will index this data inside your company also, so I could see Microsoft offering something similar to Workday in the next year or two.

Third, many recruiting and AI companies are building similar products as well. These include Gloat (an AI engine for internal talent matching), Eightfold.ai (an vendor that has more than a billion employee records and has built a “Google for Jobs” you can buy for your own company), Fuel50, PhenomPeople, and Avature. Remember that recruiting vendors have been working on job matching algorithms for years.  Each of these companies sees skills matching, internal mobility, and various forms of skills development in their future – and they have been successfully matching people to jobs for a long time. ADP is doing this also, through their embedded AI engine which powers its entire Next-Gen HCM platform.

Fourth, there are a few dark horses like IBM and Google. IBM, through its Watson Career Assistant and other skills inference products, could focus in this area. Today IBM sells its AI tools as recruiting enhancements, but the IBM Watson Career Assistant is essentially a skills cloud of its own – if the company decided to partner with Workday it could be quite interesting. And Google, now that they’re trying to “become Oracle” and sell enterprise solutions, could start selling its ontology-building engine to corporations as well.

Ultimately all these “ontology builders” have to work together. I could see a world a few years ago where core systems like the Workday Skills Cloud collect and aggregate ontology information from the recruiting, learning, and other HR systems in the company. But right now each vendor is doing their own thing.

A New World Is Arriving

While Workday Skills Cloud is still new and just getting started, the potential for this technology is game-changing. Imagine a system that intelligently analyzes the capabilities, experiences, and skills of all your people – and what power this could give to your leaders. Yes, there are many ethical and privacy issues to think about here (which I detail in this article), but to me this cat is out of the bag. 

Just as Twitter, Google, and Facebook know hundreds of things about your consumer behavior, why wouldn’t your company want similar information to help you perform better, find the next job, and decide how to progress in your career.

We’re working closely with Workday on this technology and I will keep you informed. And in 2020 we will be launching a new program in the Josh Bersin Academy to help you understand this topic.

Stay tuned, because this new area of HR tech will be an exciting space in the year ahead.