Our Courses Can’t Keep Up With the Skills They Teach. What Should Change?
If you’ve built a technical course recently, this may sound familiar. The team spends most of a year on it, including round after round of SME review. It launches. Then, around the six-month mark, a field supervisor mentions that the procedure changed, and the course is now teaching a slightly older version of the job.
That usually isn’t anyone’s fault; the timelines simply don’t line up anymore. Research from IBM and Boston Consulting Group puts the half-life of a technical skill at about two and a half years, down from roughly seven a decade ago. Most organizations review curriculum every three to five years, while software and process changes often arrive every 12 to 18 months. By the time a new course clears review, the skill it covers has usually started to age.
The instinct is to respond with volume: more modules, more video, more microlearning. That helps with information, not skill. Completion rates for large-scale online courses have hovered between 5 and 15 percent across years of MOOC data, and completion only tells us who reached the end, not who can do the work. This matters because the skills holding their value as AI takes on routine cognitive tasks are largely hands-on ones: sizing up an unfamiliar physical space, calming an upset customer, deciding what to do when the machine in front of you doesn’t match the manual. People get good at those by practicing them.
Three changes worth making:
Sort your catalogue by shelf life
Take your 20 most important courses and ask how quickly each skill goes out of date. A compliance course that holds for five years can stay on its current production schedule. A procedural or technical course that holds for 18 months needs a different approach, one where an update takes days instead of a quarter and reaches every learner the next time they log in. If a course takes longer to update than its skill takes to decay, it will always be behind.
Measure practice, not attendance
In PwC’s study of VR soft-skills training, learners finished up to four times faster than classroom learners and were 275 percent more confident applying what they learned. PwC also found that costs fell as programs grew: VR matched classroom costs at around 375 learners and became 52 percent more cost-effective at larger scale.
Those results came from headset-based programs, so treat them as a direction rather than a guarantee for other formats. Still, much of that practice no longer requires a headset. A 360-degree walkthrough of your own site, a 3D model of equipment that learners can take apart, or procedure steps that appear on a real unit through a phone camera can all run in an ordinary browser. When learners get stuck, a remote expert can see what they see and talk them through it.
Pilot one course, and pick a hard one
Choose the course that generates the most retakes or “can someone just show me” requests. Run it with one cohort for one quarter, alongside your existing version, and track a metric leadership already follows, such as time to competency, error rate, or first-time fix rate. You probably don’t need a production crew for the first lesson; footage from a phone or a 360-degree camera is often enough.
A pilot like this also helps with budget pressure. Finance rarely asks whether training is good. They ask what it changed. Practice-based programs tend to produce operational numbers, like faster ramp-up and fewer repeat service visits, and those hold up far better in a budget review than completion rates do.
At Reality Ready Training, we build immersive lessons from real job sites and equipment. In my session, “Field-Ready, Faster,” at the virtual ATD L&D Managers Conference on Tuesday, November 17, I’ll go through the cost math and learning research and show how these lessons get built.
Citations for this post are in our 2026 white paper, The Future of Workforce Training, at rrtraining.org. And if you already know which course is your hardest to teach, I’d like to hear about it at ATD.