Delivering Multi-Modal Training
Citrix Education has adopted DITA and a CMS to create dynamic, multi-modal training. Today we are delivering classroom-based and high-fidelity online training built from a single source of content. In this article, we discuss three recent DITA-based development projects to
- Evaluate the results against business objectives
- Identify clear challenges and benefits
- Identify lessons learned
- Discuss best practices
The Citrix Education courseware development team recently passed two milestones: two years developing courses in DITA and one year using a content management system. These milestones prompted us to reflect and re-assess whether we have achieved our business objectives for moving to DITA. We found that, generally, we have; however, our success has varied by project.
We moved to DITA to
- Improve the efficiency (not necessarily the cost) of translation
- Decrease courseware development time
- Treat content as a business asset to leverage, not create and forget
Our process has evolved as outlined in Figure 1.
Figure 1: Process Evolution
In this article, we review projects that map to each of these phases and evaluate our results against our business objectives. We discuss lessons we’ve learned and best practices.
Prior to DITA, we authored unstructured content using desktop publishing tools. An average project took 18 months from design through translation and followed this process:
- Develop the instructor-led training content: “the book.”
- Translate the book.
- Spend 3 months copying, pasting, and tweaking content in an eLearning tool.
- Translate the online version with basically the same content as the ILT.
Each translation project averaged 3 months. The outline for our projects is shown in Figure 2.
Figure 2: Project Outline
Counting translation as one resource, we estimate that the total work time for all resources was 48 months. In other words, it took us four years of effort to create the ILT and online versions of a course! Changes to our business made this process unsustainable. So we began looking at DITA.
Structured Content, Unstructured Mindset
Our early DITA efforts were ill-defined and fraught with questions. Authors were faced with learning new technologies, tools, and nebulous processes, while business drivers and deadlines loomed. Further, authors and managers found the book paradigm difficult to shake. These factors led to authors feeling tremendous pressure to “just fill pages,” rather than focusing on instructional design and effective reuse.
As a result, this type of project often gives inadequate attention to design in a misguided effort to save time. However, poor design has consistent negative impacts:
- Authors don’t understand the requirements for their assigned modules, resulting in a rush to fill pages with any available content.
- Authors don’t understand how the content they produce fits into a broader course context, resulting in redundancy between modules.
- Authors cannot plan for reuse across modalities, requiring redesign for online delivery.
- Authors feel even more time pressures because of the above problems, leading them to disregard authoring standards and best practices in another misguided effort to save time.
In short, applying an unstructured mindset to structured content combines the inefficiencies of desktop publishing with the complexities of component file management—a disastrous mix.
The timeline show in Figure 3 illustrates one such project.
Figure 3: Example Project Timeline, Project #1
From a calendar perspective, the project was successful: it was released in two months fewer than a typical unstructured project. But the project required 50 percent more hours of effort to save those two calendar months. To put it another way, the project took six years of development effort!
The primary cause for extra effort was redesigning and rewriting the content for online delivery after the ILT development was complete, a process that took 5 months. A thorough design and adherence to writing standards would have made the redesign unnecessary.
Additionally, because file maintenance standards were ignored, authors were not sure which files were used in which version of the course or if they were used at all. It took nearly 3 months to sort out the mess before the content could be sent for translation.
Projects like these clearly do not meet the objectives we set out to accomplish by moving to DITA. An evaluation of the business drivers is shown in Figure 4.
Figure 4: Evaluation of Business Drivers, Project #1
Although we had guessed that the following principles were important, projects such as this example show us that they are vital:
- Management must understand and enforce standards and best practices.
- Topic-level design and writing must account for online and offline delivery.
- Poor design impacts development, translation, and maintenance.
- Source files (numbering in the thousands) must be well-maintained by all team members.
Write for Online, Use Offline
As our DITA implementation and experience matured, we began to look for more efficient ways to develop content for multiple modalities. We changed four key aspects of our approach:
- Design courses for online delivery, with the assumption that they would “just work” offline.
- Build online and offline versions of the course throughout development.
- Write in a minimalist style.
- Collaborate on content development with other teams in Citrix.
These changes have made a tremendous impact, as shown by Figure 5.
We’ve shown a reduction from 72 months of effort to 18.5. We’ve released courses in multiple languages in just over 8 months, instead of 16. We saw the biggest savings in the reduction of online course development, which fell to 2 weeks of effort from 15 months. These changes are outlined in Figure 5.
Figure 5: Example Project Timeline, Project #2
Rather than designing an ILT manual and then converting it to online as an afterthought, a topic-level design takes multiple modalities into account. For example, consider topic length. A topic with sections that spans three pages is fine for print, but several separate, nested topics would be more appropriate for online presentation. Furthermore, the need for interactions, videos, and images can be identified during the design phase, saving time later.
Our second change, adopting minimalism, aligns well with instructional design principles, most notably that information should be action-oriented and task-focused. Adopting minimalism, which included training all team members, prompted us to discuss what we felt should be included in a course (concepts and tasks) and what should not (reference material). We also began avoiding transitional text, forward and backward references, and cross-references that were common in earlier courses. This change supported reuse and allowed developers to focus on the most useful content.
Third, we focused on collaboration. Although cooperation among content development groups was common, the process was informal. We began instituting cross-team content councils to structure how we approach, communicate, and collaborate. A successful content council looks at all content required for a product release and brings many groups together to most efficiently develop content. In reality, differences in deliverable types, timelines, and organizational commitment make cross-team reuse difficult, but we have seen an overall increase in efficiency.
These projects meet all our business objectives for moving to DITA and demonstrate what is possible when authors adhere to our best practices. In addition, we establish effective cross-team collaboration as shown in Figure 6.
Figure 6: Evaluation of Business Drivers, Project #2
Lessons learned from these projects were that
- We waste effort cutting and pasting content from Word to DITA, because Word is often the content collaboration tool of choice. We’ve since begun looking for an automated transform.
- A set of common standards across teams would help us avoid rewriting content developed by other teams to fit our style and standards.
- While a minimalist style makes it easier to convert content from DITA to interactive online items, we did not see significant efficiencies until we developed an automated transformation.
- It isn’t necessary to build the online version of a course in just 5 days. An extra week allows us to deliver higher quality.
Leveraging Another Team’s DITA
Having successfully achieved reuse between modalities, our next step was reusing topics between courses and teams. An opportunity presented itself in two courses with the following characteristics:
- Timelines were closely aligned.
- Both courses covered an enabling technology that comprised approximately 20 percent of the material in each course.
- XML content from our Technical Publications group was available as a starting point for courseware development.
- The two projects were the first to use a content management system and not merely version control.
The first project had a slightly less than normal duration. Team members felt that using source from tech pubs reduced development time, but we aren’t able to say by how much. The second project did not see a reduced timeline; six months with four developers is typical. Some delay resulted from product schedule changes and team turnover. It is possible (and unverifiable) that content sharing could have reduced the duration in the absence of these other factors. See Figure 7 for a timeline.
Figure 7: Example Project Timeline, Project #3
Sharing content was not, however, without difficulties. The first and easiest to address was that not all content was in DITA; some was in DocBook. We leveraged freely available transforms and in-house scripting to convert the content and divide the book-length file into topics and maps. The more difficult problem was the content itself: topics were generally longer and more detailed than appropriate for training material, and the tagging and writing styles differed from ours. It was obvious when an unrevised topic from tech pubs was inserted into a course manual—it simply didn’t fit.
Content sharing between courses was no less problematic. In some cases, modifying shared topics was beneficial; for example, we could resolve questions of technical accuracy, clarity, or adherence to standards for both courses in shared files. In other cases it was not; for example, issues arose when the change in topic scope appropriate for one course was not okay for the other. Managing these conflicts required constant communication between the project leads, and not all resolutions were mutually satisfactory.
Leveraging content provided mixed results as shown in Figure 8.
Figure 8: Evaluation of Business Drivers, Project #3
Lessons learned from these projects were that
- Project leads need to discuss shared topics and maintain open communication.
- XML is incredibly malleable if you have the skillset.
- Tools are the “easy” part; changing people, process, and practices is far more difficult.
Overall, DITA has helped us achieve a level of scalability and flexibility that would not be possible with desktop publishing tools. However, DITA also introduces complications around builds, automation, and file management. Managing these issues requires constant vigilance. To help, we have adopted the following best practices:
- The annotated topic list should evolve as the content is developed.
- The characteristics of quality online content–succinct, well-organized, focused on users’ goals—also make quality offline content.
- Authors need to build and evaluate multiple outputs during development to see how the content comes together. It is easier to make small corrections along the way, rather than major revisions at the end.
- File management, even with a CCMS, is critical to project success.
- Authors should understand that standards are a means of codifying collective experience, meant to prevent problems they may not foresee.
- While standards are vital, there will always be questions about specific applications. Senior team members can help junior developers resolve these questions effectively.
- A process to submit change requests and have them validated by other team members enables everyone to participate in the shaping of standards and improve the process.
These practices have enabled Citrix Education to meet business needs while improving our processes. With the foundation we have established, we are reaching new markets by supporting new languages and looking to expand our portfolio into new publication channels, none of which would have been conceivable three years ago.
Citrix Systems, Inc.
Ben Colborn is a Lead Courseware Developer for Citrix Education. In addition to developing instructor-led and online training material for Citrix virtualization technologies, he works to improve learning experience quality and simplify internal development processes. He holds a B.A. in English from the University of Idaho and an M.A. in TESOL from San José State University.
Citrix Systems, Inc.
Patrick directs the content management initiative at Citrix Education, where he led the implementation of DITA and a CMS to enable multi-channel publishing of training content. He has spoken at conferences on a variety of topics, including DITA-OT customization, training new authors, business cases and cost savings, and automating quality assurance checks. Recently, he has focused on simplifying his team’s authoring process by implementing a set of DITA 1.2 constraints.