Moving from a trainer to eLearning. Do you try to replace the trainer?

I guess the question is whether the avatar is a good enough replacement
for a live instructor that learners.
1. Can you interrupt it?
2. Can you ask it a question?
3. Can it understand the real reason for the question? Usually people
ask a question that shows a symptom of their misunderstanding, rather
than the fundamental reason for the question. If only the symptom is
answered, the fundamental misunderstanding usually remains.
4. Can it answer your question?

With electronically delivered training the different learning needs
pose challenges and offer possibilities, especially when you remove the
live instructor. I have never believed that a pre-recorded instructor
was a good substitute, however I see many trainers and content
developers sticking to the old paradigms. That is, "lets see how close
we can get to instructor-based training while delivering remotely":
1. Record the instructor on video or audio.
2. Expect that the course will only be accessed once (so provide only
linear navigation)
3. Provide only the instructor's presentation (both what they gave as
material, and only showing their answers to the questions posed during
video taping).
4. If they can't record the instructor, drop back to an avatar.
5. Remain in PPT "instructional design" which means breaking the content
so that it fits 3 points per slide, rather than organizing content to
coherent chunks with natural divisions. Here the fundamental issue is that the trainer is trying to replace the form of face-to-face instruction while delivering electronically.

These short-cuts result in courses with low value and low attention
retention. Some solutions:
1. If you record a video or audio, keep it under 2 minutes. If it is
longer than that, the user will "float" away to check e-mail or play
freecell. Make sure the video downloads and plays in less than 20
seconds. Justify its value: Does it show a process or is it just a
talking head? If a talking head, are there any visual queues that are
absolutely essential that could not be given with lower bandwidth
requirements (e.g. "this is important"). If audio doesn't match the
text, the user will have to follow one or the other, resulting in
cognitive overload.
2. Make the course following web practices: lots of navigation, let the
user choose their path, let the user reach any content within 3 logical
clicks, make it easy to get back to the content when the user needs to
look something up. Many LMSs block the student from viewing content
after they have completed the course. This would remove re-usability
for the end-user.
3. Show the same material in multiple forms. This is where the web (and
creativity) can really shine. Provide a screen snapshot animation, a
broken down step-by-step procedure, a longer article, a try-this
example, a quiz, a tracked test, links to other sites (vendors,
SecondLife, FaceBook, Wikipedia). Users will naturally gravitate to the
presentation forms that are easiest for them. (Don't try to control the
user's navigation...you will just lose them.)
4. People can read much faster than spoken language. People can also
understand content narrated much faster than we normally speak. Audio
content speed should be adjustable. It is amazing to watch a blind
person using a blind reader. They often set the audio to go really
quickly. Keep in mind that audio delivery is very linear, so it needs
to be sped up for people who can only access from audio.
5. Use the multiple navigation abilities (links to drill-downs) to break
the content based on what will give the best organization for learning,
rather than what will fit on one slide without scroll bars. Remember
scroll bars? They work fine if vertical. Horizontal scrolling,
however, should be avoided.