Learning Experience Platform (LXP) Market Grows Up: Now Too Big To Ignore

The learning experience platform (LXP) market is growing up fast.  Only a few years ago startup companies like Pathgather, Degreed, and EdCast pioneered the idea of a platform to make corporate learning content easy to find. These next-generation portals took off and now thousands of companies are looking to put their Learning Management System in the basement.

Since then this market has exploded (it’s over $300M and growing at 50%+ per year) and it now shows signs of age: vendors are getting bigger; products are becoming complex; the players are moving in different directions.

Will the LXP market replace the $4 Billion+ LMS market?  I think this is starting to happen, which is why every Learning Management System (LMS) vendor has jumped into this space.

Why Do We Have Learning Experience Platforms?

The original problem these products solved was what I would call discovery.

I want to learn something or take a course and I simply cannot find it in the course catalog (LMS).

LMS systems were never designed to be employee-centric. They were developed as “Management” systems for learning, focused on business rules, compliance, and catalog management for courses.

The LXP, which looks more like YouTube or Netflix, is a true content delivery system, which makes modern content easy to find and and consume.

Fig 1:  What an LXP Typically Looks Like

Both systems are needed (LXP and LMS), but over time these lines have blurred. To make it simple to understand, consider the picture below.

Fig 2: LMS vs. LXP

In the early days of the LXP, the green layer was very thin. All these products did was serve as aggregation portals.

As the product category grew, however, they became more functional. Not only does the LXP need intelligent methods to recommend content (discussed below), but they can also be used to recommend third-party articles, find people who are experts, and potentially index documents, videos, and other digital assets. In a sense, they are content management, knowledge management, and learning systems all in one – which is why the market is growing so quickly.

The Problem of Discovery Is Complex

The core problem, that of discovery, is very complex. (This is what made Google a multi-billion dollar company.) Inside a business, there are thousands of courses to find, and as you build more content (using tools like 360Learning), the library explodes. And the LXP reaches the internet, where every video, article, podcast, and webpage is possible a way to learn.

As the founder of EdCast puts it, the real LXP mission is to “index the world of learning,” to make it easy to find just what you need.  (Sounds like “The Google of Learning.”) Much harder to do than it is to say.

I’ve talked with many of the early users of LXP platforms and they love the systems at first … but as they become full of content the problem of “intelligent discovery” becomes hard. And that’s where the market is going next.

How Will Content Discovery Evolve?

Let me suggest this market is moving in six directions at the same time.

Fig 3: How LXP Products Recommend Content

1. Skills-Based Learning.

Vendors like Degreed, LinkedIn, EdCast, Percipio, and IBM have started to add skills-based discovery tools into their systems. While all LXP systems have skills-based tagging, these vendors are starting to build skills assessments, skills inferences, and skills-based learning paths. Companies have always wanted a skills-based approach to learning. It’s just much harder than it looks.

IBM’s Watson Talent Frameworks, for example, one of the most robust skills frameworks in the world, has over 3,000 job profiles and 2,000 skills integrated. An LXP that tries to map all these skills to content has a lot of work to do. Degreed has launched its own internal skills assessment engine, and I believe others will as well.

Pluralsight, one of the leading providers of software technical skills programs, has its own learning portal and has built a very innovative skills assessment engine called Skill IQ. This offering came through the acquisition of Smarterer, which was a very innovative system to crowd-source assessment based on the skills of others. While Pluralsight is not in the LXP market, one may ask “how can the LXP use the skills data from a content provider?”

2. Usage-Based Recommendations.

Vendors like EdCast, SkillSoft, Cornerstone, LinkedIn, and Fuse, aggregate massive amounts of data to recommend learning based on the usage of others. While this sounds like a good idea (this is the idea of Google PageRank or Facebook EdgeRank), it creates problems. When one program is widely used it becomes widely recommended, and it starts to “crowd out” other content that might have more value and credibility. And what if I”m in the Paris sales office, should I see the most popular content in China?

Vendors will build smarter recommendation engines over time, but I’ve heard stories of buyers who purchased an LXP and found it recommending incorrect content on the public internet because it had high rates of traffic. I’m not saying this approach is not a good idea, it just takes lots of R&D to make it work well.

When EdCast started working with NASSCOM and other major institutions its goal was to create faster and better machine learning recommendations. They now do it for each client, as do Degreed, Valamis, and CrossKnowledge. Degreed’s new OEM relationship with Harvard Publishing promises to help them better understand patterns of usage for Harvard’s content. Wiley, the owners of CrossKnowledge and one of the biggest book publishers in the world, is embarking on a data lake project to better recommend content.

And there are some important subtleties to this. Fuse lets teams segment themselves into communities, so the system can recommend the most popular content within that group. Their clients found this to be far more relevant than any type of enterprise-wide usage tracking.  (ie. If I work in a certain hotel, I may want to see content most useful among my peers, not necessarily across an entire global network.)

3. AI-Based Content Analysis.

The third approach to discovery is far more innovative and powerful. These vendors (the leader here is a company called Volley.com, followed by Valamis, IBM, and Docebo) actually injest instructional content (text, video, audio) and then identify the instruction within. 

In other words, they read the content and figure out “what is this trying to teach people?”

Volley.com is widely used on Wall Street, for example, and the company’s system can crawl through CyberSecurity documentation and “create” training, micro-learning, and assessments on security procedures for the bank. This approach has enormous potential as these systems identify “level of expertise” and “credibility” of content through pedagogical analysis.

IBM and Valamis’s new learning platform does this as well. Valamis, which is used by Boeing, for example, can fast-forward to a video segment and show you precisely what to watch based on instructional needs. The domain of content analysis is going to be huge.

4. Chat With and Understand The Learner.

The fourth way to recommend content is to ask the learner.

In the best of all worlds, the Learning platform should know your role, your experience in that role, what content you have consumed, your learning preferences, and of course what aspirations and goals you have. 

Most of the LXP products let the user define their interests when they log in. But that’s not enough. The system may need to know your tenure, aspirations, levels of expertise, and even your “way of learning.” In my case, I like to read a lot so I want more books; others may want podcasts or videos.

LinkedIn Learning has a lot of potential in this area, but companies like Filtered, Degreed, and Valamis are working on this too. Degreed, for example, has worked with clients like Bank of America to map specific job roles to learning recommendations. IBM’s YourLearning platform does this using IBM’s Watson Talent Frameworks.  Filtered’s product Magpie, asks you lots of questions when you get started to make sure the engine knows your role, interests, learning needs, and skills profile.

Fig 4:  Filtered’s Magpie Recommendations

5. Expand The Learning Business Rules.

The fourth approach to discovery is the boring, well-known, but badly needed model of “business rules for learning.” Every LMS has lots of this.

Every quarter we need people to take sexual harassment training; salespeople have to finish their new hire training by the end of the quarter; engineers in a pharma company need to be recertified on equipment; and on and on.

All these business rules, which we have heretofore left in the LMS, are creeping into the LXP. And why not?  If we are going to put hundreds of content objects into the LXP, why can’t we get it to formalize, recommend, track, and manage curricula and programs too?   This is a heavy lift for the LXP players, but hey whoever said being a learning platform company was going to be easy.

(PS the conversational interface is a new way of managing this, look at Valamis’s Digital Learning Assistant.)

Fig 5:  Digital Learning Assistant from Valamis

6. Store and Manage Data and Analytics.

The sixth new dimension of the LXP market is data. Where are all the utilization, history, tracking, assessment, and compliance data going to be stored?  Right now most companies put their LXP in front of their LMS. It’s only a matter of time before companies start saying “why do we have two platforms anyway?”  Can’t this LXP store the data in our LMS?  Or can we just buy one system?

Right now LXP systems don’t store much learning history data, but they are collecting a lot. You as a buyer have to decide where you put all this information, and how the machine learning can best use it all.

What Happens To The Business Rules of Learning?

There’s still a big open question in the market:  where do the business rules go?

A Few of the Business Rules for Learning:

  • Compliance: Some content is mandatory, so we have to track completion and manage compliance. When the content changes we may need to “recertify” people.
  • Career programs: Onboarding, new hire training, and career development are learning tracks. They may have pre-requisites, branching, and tests at the end.
  • Customer education: Customers may have to pay for certain courses; they may only see content which is externally licensed; we need e-commerce, training credits, and discounts.
  • Performance support: Sometimes employees just want to search for the answer to a question, they want to search within the content.
  • Development plans: after a performance review and an employee puts together a custom development plan, where do they store it?
  • Manager approval: do managers have to approve this course? Do they need to know someone has completed it?
  • Ad-Hoc learning: Sometimes employees don’t know what they need to learn: they just want the system to nudge them to read or learn something that’s important or adjacent to their current jobs.
  • There’s the whole new world of “learning in the flow of work.” How do I take learning content and make it relevant to me from my job?

One could argue that most of this functionality is in the LMS, so let’s leave it there. I think this will change. Much of this functionality will migrate into the core HR system over time (this is certainly Workday’s strategy), leaving the LXP as a more and more important system in the architecture.

Will the LXP Become an LMS?

The LMS vendors would have you believe that the LXP is a simple and easy system to build. Companies like Cornerstone want to give the LXP away.

I believe the opposite may be true. The LXP is the learning delivery platform of the future, and the level of investment in that platform is growing exponentially. Vendors with legacy LMS systems are trying to catch up, but they’re finding it’s harder than they thought. 

And now this market is too big to ignore. Most LMS vendors are building this type of functionality, including companies like LinkedIn, SkillSoft, Saba, and even Microsoft. 

Long ago when I was at IBM I learned something important from one of the most senior leaders in IBM research. It’s hard to make old software do new tricks.

Software is hard, but hardware is soft.

In other words, it’s not easy for legacy LMS vendors to turn their old systems into new platforms. In most cases, they have to start over (this is certainly what Skillsoft has done with Percipio), and I could expect to see acquisitions start this year.  (Including LXP vendors buying LMS vendors.)

The Economics Really Matter

The final point I want to make is about the economics of this market.

As the LXP space grows in size and value, training departments have to decide where they want to spend their money. One of our clients, a pioneering user of an LXP, has a plan to replace their LXP with the planned solution from their new LMS vendor. Why?  They just don’t want to pay for two systems.

In the other direction, there’s a lot of demand to reduce spending on the LMS. I know of two global enterprises who plan on “turning off” their LMS in the next few years.

The market is dynamic, exciting, and filled with innovation. If you are searching for an LXP or need help figuring this all out, just let me know. I’m here to help.