The Ultimate Guide to xAPI

Like This Article? Share It!

Throughout the years, we’ve spoken with clients and prospects who have come from a variety of backgrounds when it comes to their understanding of the Experience API (xAPI). The knowledge we’ve seen ranges from knowing almost nothing about xAPI, but wanting to learn more, all the way to experts in the field who need our technical expertise for a particular integration or consultation service. Our team at Riptide are part of the workgroup that created the xAPI specification and we have been serving enterprise customers with a functioning enterprise learning record store (Storepoints LRS) since 2014, so we know a thing or two about the spec! This publication is intended to first, help you conceptually understand this powerful data specification to address the question: “What is xAPI?” and second, show how you can immediately begin applying xAPI to your business or educational use-case.

Concepts in this guide will be covered in a fairly non-technical manner so everything can be communicated as clearly as possible!

What You’ll Learn:

1) A Brief History Behind the Experience API (xAPI)
2) What is xAPI?
3) The Anatomy of an xAPI Statement
4) What is a Learning Record Store?
5) The Importance of xAPI Conformance
6) The “So What” of xAPI
7) Selecting an xAPI Vendor

A Brief History Behind the Experience API (xAPI)

With the adoption and growth of the internet in the late 20th century, a digital revolution began. Training was no longer bound to the confines of instructor-led or a correspondence course, where one had to be in the classroom to learn or communicate with the instructor through mail. The internet was a technological disruption which radically began changing the ways that we think about education and training. With the adoption of the internet, we find that instead of learning simply happening in a controlled (synchronous learning) classroom environment, with controlled assessment strategies, learning is happening more and more in many uncontrolled (asynchronous learning) digital environments.

Along with widespread internet adoption, the mid-late nineties introduced Flash (shockwave) multimedia technology. Flash technology brought an amazing learning experience (UI/UX) into the digital learning mix and helped to further revolutionize distance learning for the next decade. Multimedia technology was also expanding to include content management systems, youtube videos, social platforms, and many other mediums.

The cognitive science of learning dates back to the early 20th Century, during wartime when it was critical to get large amounts of diverse people trained to a minimum standard (ex. Given X Conditions, can the learner perform X Action to X Standard?). This led to the documenting and hypothesis of findings starting in the 1940’s and 50’s. From the beginning of the cognitive science of learning (and especially digital distance learning), it has always been important to gain insights and evaluation about the effectiveness of the instruction. We are all familiar with the ADDIE model, and the “E” stands for formative and summative “evaluation.” The Shareable Content Object Reference Model, or SCORM specification, was created to provide this evaluation data, but unfortunately, was not created for the interconnected, multi-device world we live in today. The SCORM specification was developed in the late 90’s early 2000’s and it built upon the work of the Aviation Industry Computer-Based Training Committee (AICC), originally to make courseware content interoperability between various LMS’s. These specifications are also dependent upon maintaining a constant connection with the LMS. This early SCORM and AICC did not forsee the interconnected world we have today.

Today, digital learning happens in all types of environments (not just the LMS). Due to SCORM’s various limitations, only a few points of evaluation data are available in spite of its wide adoption. Today’s SCORM reporting provides barely more than a “completion” and very little insight about the learners activity. There is a collection of white papers and research for SCORM 2.0 that eventually helped inspire the development of something more appropriate for today and tomorrow – xAPI. xAPI does not rely upon an LMS, anything digital can report contextual xAPI data streams to a Learning Record Store.

Get Learning Analytics With xAPI

Yes – yes you can.

In 2010, the governing body for distributed learning initiatives in the government, the Advanced Distributed Learning Initiative (ADL), began looking for a new, standardized, experience tracking model that would be able to future-proof organizations who adopted it. What this means is having xAPI built in a way of being able to support current as well as emerging learning technologies. There was already considerable work being done in the high technology sector around Activity Streams. The thought behind xAPI is that it would provide any type of digital learning experience data that could be analyzed and used to correlate productivity and performance metrics. In addition to this work, the ADL is constantly striving towards continuing development efforts as cited below:

“ADL current focus around xAPI Data Visualization and Associated Learning Analytics
The xAPI specification provides the opportunity to conduct micro-level, semantically-defined behavior tracking. However, to date, organizations typically measure and store primarily macro-behaviors, such as test scores and course completion data. We believe the next step to widespread adoption of micro-level behavior-driven learning systems is the development of models, prototypes, and specifications for analyzing, interpreting, and visualizing this data.

ADL current focus around xAPI Content and Pedagogical (Andragogical) Models
…capabilities such as competency management, learning profile stores, learning analytics, and meta-adaptation…will enable personalized (adaptive), career-long, ubiquitous learning (i.e., truly anytime, anywhere, and integrated across any platform/device).”

(Source:https://adlnet.gov/collaboration/baa-fy2018)

So, xAPI isn’t a specification that just appeared out of thin air. There has been considerable planning and research throughout the years that serve as a solid foundation for xAPI. The initial requirements for xAPI were funded with Research and Development (R&D) money from the U.S. Government and the effort has remained completely open-sourced, community-based, and built from a passion to bring big data and activity stream technology to education and training. Since the beginning, xAPI has always been about understanding effectiveness and evaluating the effect of the learning content on the learner.

What is xAPI?

So, what is xAPI? Simply put, the Experience API (xAPI) is a technical data specification that, when implemented correctly, is agnostic about any type of digital learning content being delivered and consumed. It allows completely flexible and customizable tracking of behavioral learning activities through activity streams. These activity streams break down into Verbs and Activities. The xAPI statements follow a very simple format: Actor, Verb, Object or someone (or something), did (doing), this. This pattern, repeated, can create a contextual activity stream which produces a clear recording of the learning experience. This data can also be cross referenced with performance data so that you can map the training to performance.

xAPI can track moment-by-moment contextual actions such as: what is happening in a simulation or game, geographic location, browser connection, answering a question, the text of the question, the weight of the answer, credit earned, watching a training video, pausing or seeking video at xx:xx time, and so much more. xAPI was designed to address gaps in the existing SCORM standard, such as offline training, mobile training, AR/VR, and any digital training within or outside of a web browser. xAPI data is easily readable by humans and computers, and allows for simple integration and interoperability between disparate systems. The Experience API is supported and recommended by the US Department of Defense (DoDI 1322.26) and the US Department of Education (Ed Tech Developer’s Guide).

Example: Activity Streaming within Web Browsers

Social Media

Social media analytics data

To clearly articulate the power of activity-streaming technology, think about social media. These powerful social platforms gathers so much data from all of the users of their product, that they are able to create demographic categories or “personas.” This has a direct application in learning. Formal education uses techniques to diagnose and categorize the ways that learners may be struggling or excelling at concepts. Activity Stream technology is a way to provide the data needed to understand the learner and adapt for the learner’s benefit or in the case of social media, adapt for the advertisers and consumers benefit, in real-time.

As you can see, social media captures a seemingly endless amount of information around their users through big data and activity streaming technology.

These personas cover a wide range of details about people (including you!) using the Facebook platform which may include: what you like/don’t like, who your friends are, where you live, your education level, your job title, and much more.

Scary, huh? It’s true that marketers use this kind of data every day to understand what efforts resulted in sales and learn more about their customers to help businesses best serve them. The good news is, learning and development is using and harnessing the power of xAPI for similar purposes. Visualizing who your learners are, how they’re interacting with content, and what digital content they’re interacting with, can help you gain insights to serve them (and the business) in the best ways possible.

Example: Activity Streaming within Apps or Other Platforms

Activity streams are not limited to simply web browser activities (like we saw above). They can also be found and captured in platforms and software such as mobile phones, AR/VR devices, and video games.

Video Games

The screenshots below show metrics from a popular multiplayer video game, Overwatch. This game can be played on the PC or on other platforms like XBOX and Playstation 4 with players from around the world. There are a ton of metrics and data that the Overwatch team captures from their players. These metrics are not only useful to the company behind Overwatch, but also to individual players to see how they are performing, what characters (roles) they do best with, and areas they may need to improve on.

The screenshot below is from a player who played Overwatch as the character “Hanzo.” As you can easily see from the data collected, this player has played this character for a total of 13 hours, had 49 wins, and a 15% weapon accuracy. The player can use this information to derive insights on what they may need to improve on in their next games.

xAPI data can be gathered from anything digital

Activity stream data can be captured from anything digital as we’ve seen with our web-browser and non-web browser examples. It is possible for all of these types of data to be captured as xAPI.

So, why xAPI? Why don’t we just use one of the above proprietary technologies? Using proprietary technology is an option. But if we do that we are now locked into the solution and it is difficult to transition to another. xAPI is open and non-proprietary. With xAPI you are not locked into a proprietary struggle; the vendors must comply with the specification and this should be encouraging to the learning technology consumer.

The typical LMS business case can provide a good example here. For many companies, choosing and implementing an LMS is very costly and time consuming. It is also difficult when converting from one LMS to another and very often data is lost with that conversion. With xAPI and an LRS you should be able to easily swap out the old and bring in the new. If the important data is in a non proprietary format (xAPI), you can future-protect your technology and your training program. Right now, very few training organizations have good evaluation data nor do they have historical data and it is largely due to limited reporting technology (SCORM) and these proprietary roadblocks.

The Anatomy of an xAPI Statement

Remember the activity streams we just spoke about? These behavioral learning activities may happen within an e-learning course, performance systems, social learning, and informal learning.

The simplest way to understand the format of an xAPI statement is “Person”, “Did”, “This” or Actor, Verb, Object. This is the structure that can provide information about all activities. An example of this structure put to work might be:

“Nick Answered Question 1”
Actor: “Nick”
Verb: “Answered”
Object: “Question 1”

If you look at the first xAPI statement below, you’ll see that you can also get more context around Question 1 if you so desire. In this example, you can see that Nick answered with the “response” of “b.” You can also get even more data, including how many points the question was worth. In this case, 0.04. Another tangible example might be “Nick discovered a new pizza recipe.” This simple activity could lead to other xAPI statements documenting Nick’s experience making this pizza. These might include activities such as all of the steps in preparing the pizza, the technological tools that Nick used in making the pizza (scale, stove, etc), the results of informal judging by his family, all the way to Nick snapping a photo of the pizza and posting it on Instagram.

This may be more data than you can get now, but the power of xAPI goes far beyond this rudimentary description. You can easily track the weight of the question, the name of the course, the name of the test, the id of the question, metadata about where the question came from, the question and answer text itself, time of day, device(s) being used, and much more within the xAPI activity stream.

xAPI statement and activity stream example

If you continue to track these activity streams for Nick and start adding some more context to your activities, such as these, you can easily string along all of the things that Nick did as an activity stream. The resulting data from what Nick experienced and how he performed during the course may verify competencies if you have information about performance or behavior after the training. If your objective is to map training to performance, xAPI is how you get there!

Don’t be intimidated by the screenshot above. This is simply an example of raw xAPI data, and can easily create charts and graphs of exactly the types of data and analytics you want to see.

Now, every behavioral action on anything digital can be captured with xAPI. Imagine you have a bunch of learners that you’re wanting to capture data from (that could potentially be millions of activity stream statements!). So, you might be wondering where this xAPI data is stored and how you can actually begin using it? That brings us to our next topic of discussion: The Learning Record Store.

What is a Learning Record Store (LRS)?

Simply put, a Learning Record Store (LRS) is a database and an implementation according to the xAPI Specification. There is an LRS Conformance test to verify that an LRS is, in fact, an LRS. It is not important for you to understand all of the technical jargon, but it is important to know that if what someone is calling an “LRS” does not conform to the xAPI specification, it is not an LRS.

The LRS and xAPI are intrinsic to each other. An LRS is where Experience API (xAPI) activity stream data is stored, and it is where you go to get the data. The LRS can connect to any of the systems in your organization that you want to get data from. Examples of these systems might be a Learning Management System (LMS), Sharepoint, Youtube, Social channels, learning portals, performance systems, HR systems, or anything digital where you learners might go to get information or take assessments. With today’s technology, it is relatively simple to get data from systems to report to an LRS endpoint and, if it is done correctly, the work is done once and you’ll be receiving xAPI data without constantly requesting data from the system or department where the system resides. The immediate benefit to working this way is that you limit data transformation. With xAPI you can use single data transformation on multiple systems to get all of the data into one format. This can give you the power over all of the data in your ecosystem in a way that has not been available before.

Once you have your important systems reporting to an LRS, all of your learning activities will be speaking the same “language” making it easier for you to view and analyze this data in one place (see graphic below).

xAPI is big data for learning but we don’t have a magic xAPI wand you can wave, or a big data for learning switch we can turn on. We often use a crawl, walk, run strategy when engaging with our clients. This strategy is about starting simple and adding complexity. It sets them up for success whether they are looking for something simple like getting more data from a SCORM package, or something more complex like connecting all of their critical business systems to an LRS and adapting to the data through analytics and correlations.

The Importance of xAPI Conformance

Not all learning record stores are created equal. In April of 2017, the governing body for the xAPI specification and learning record stores, the Advanced Distributed Learning Initiative (ADL), released a conformance test for the Experience API.

Here’s a simple way to think of the LRS Conformance Test:

Web browsers (Internet Explorer, Firefox, Google Chrome, etc) have to support the HTML5 specification. The technology used to create these browsers must be conformant to this specification in order for users to make content that can work on any HTML5 conformant web browser. An example of this would be creating content, such as a blog post, through Google Chrome. This piece of content should work and show up exactly the same on Firefox, Internet Explorer, etc.

An LRS must support the xAPI specification in the same way that your web browser has to support the HTML5 specification to function properly. If you’re collecting xAPI data in one LRS, these statements should show up and work exactly the same way if you switched LRS’s.

The xAPI-Conformance test for the LRS covers 1,389 specific tests that an LRS must pass in order to be deemed “xAPI-conformant.” The current list of xAPI-Conformant Learning Record Stores can be found here: ADL Conformance Test.

When you start considering an LRS vendor, or an LMS vendor who claims to have an LRS, you should verify that they are conformant or can provide a verifiable date that they will be conformant.

If you’re interested in reading more about conformance, check out our full-length article on the topic: LRS Conformance – A Call to Action for Industry and Academia

The “So What” of xAPI

What is xAPI

So, what’s the big deal about xAPI? In addition to xAPI being a unifying concept, in that you can gain data and insight from anything digital that your learners/customers might be accessing, you’ll also be able to find trends in your data. Once you implement xAPI and start getting data from learning/training activities, you’ll be able to start setting quantifiable business and learning goals for your L&D team. This quantifiable goal-setting can be accomplished since you’ll have access to analytics that you’ll be able to tie to business goals.

The business value of xAPI presents itself in a variety of ways. Here is some of the “So What” of xAPI, quoted from a draft of the IEEE LTSC TAG xAPI 2018 Technical Report on xAPI:

1) “Understand how learning resources are being used: Tracking usage of learning resources and categories of resources to plan future investments, processes and intuitives.

2) Better understand the impact of learning: Capturing existing training and learning that is taking place in an organization and correlate to employee behavior and business metrics. For example, sales data could be mapped to your learning intervention data (xAPI statements) to show the return on investment.

3) Improve the quality of learning interventions: By capturing all learning taking place, the value of specific learning activities can be evaluated to understand their impact and improve the commissioning of future learning and overall quality.

4) Facilitate integrated learning ecosystems: xAPI is an interoperability specification and as such, data from any number of learning activities can be shared between systems both internally and externally to your organization. It can be used to help facilitate more complex learning ecosystems than might otherwise have been practical.

5) Gain valuable insights about learners and learner behavior: xAPI data from different demographics, cultures, industries, age groups, may be analyzed to better identify and more effectively teach to learner needs, types, trends, audiences, and preferences.”

Need a visualization? Check out the diagram below. You don’t have to get this complex right off the bat. But this is a generic example of how you might connect different systems within your learning “ecosystem” to gather data, build reports/data visualizations, and distribute them to a variety of audiences. The awesome part about this? You’ll have enough data to map your training to performance, and change/react your approach depending on the results you find!

xAPI and a Learning Record Store

Selecting an xAPI Vendor

As long as the LRS conforms to the specification, the differences in vendors is going to be found in two areas – Product and Services:

Product

The LRS can be differentiated through the product itself, the features of the product, the technology stack, the way the vendor implements and deploys their solution, how it is hosted, and where it is hosted. Most vendors today will have a reporting position as well which is the ways for you to report, visualize, analyze, and distribute dashboards, etc. Business software comes with some or all of these: a Service Level Agreement (SLA), Quality of Service (QOS), and Master Services Agreement (MSA). It is important for you to understand exactly what you need and what the prospective LRS does. Very often the acquisition process in businesses takes care of this, but it is not difficult to navigate if you understand your requirements (the way you intend to use xAPI now and the near future).

Services

The LRS can also be differentiated through the professional services or support the vendor offers and these will often appear as a Statement of Work (SOW) in the contract. The MSA will often have language that essentially puts the vendor at your service to do work upon your approval. This work might be to get a system or training package reporting xAPI to the LRS. At this time, xAPI/LRS is a fairly new endeavour and the LRS vendors are trusted technology partners with their customers. Some of them roll up their sleeves and get into the Instructional Design process with you. All of our customers of our Storepoints LRS are the subject matter experts and instructional designers. It is our philosophy in all of our learning technology products to provide the tools for the educators and trainers. We are not the instructional designers, rather, we serve the practitioners of pedagogy and andragogy. It is very much a relationship of scientist and engineer. We are the engineers serving the learning scientists. You have to figure out what level of service you want from the vendors because you are going to need to trust them as you modernize your approach.

Conclusion

We trust that you enjoyed and learned something from our Ultimate Beginner’s Guide to xAPI! If you have any questions around concepts covered in this guide, or thoughts about anything xAPI-related not covered here, our team is always ready to help!

If you’re ready to begin with xAPI in your organization and gather more meaningful insights for your team or customers, check out our 1 or 2 day onsite xAPI workshop. We’ll travel to you and give you all of the information and strategies you need to make xAPI a holistic part of your learning strategy.

Contact Riptide Here to get started correlating learning to business outcomes with xAPI.

7 Comments

  1. louis k.

    Probably the most comprehensive xAPI guide I’ve read…thanks for sharing this! Downloaded a copy for my team too.

    • Christy Puller

      Glad to hear it louis – Thanks for checking it out and let us know if you have any questions on it!

  2. Conformance guru

    “but it is important to know that if what someone is calling an “LRS” does not conform to the xAPI specification, it is not an LRS.” amen! we ran into trouble when implementing something that a vendor called an LRS, but wasn’t conformant. so very important..

    • Christy Puller

      Thanks for sharing your story. As you found out, LRS Conformance is so important!

  3. Manuel

    I appreciate the explanation, it is very thorough and gives me a better perspective on how to approach the subject with future clients.

  4. Sasha Rodriguez

    Definitely worth the read – I like the breakdown and helps to gain some insight on using the data-points collected to address KPI’s and overall organizational goals. Looking forward to reading the eBook.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

This site uses Akismet to reduce spam. Learn how your comment data is processed.