From experience I can tell you that making trainings is multi stage process. First you have to envision one and construct the timeline and events for it. It is very easy to fall into a trap of showing user something which requires previous knowledge which simply isn't there or is falsely assumed. Spend your time wisely and do a detailed breakdown of whole training. Write down each lesson, revisit and refine to a point where you have whole training literally on paper. Make absolutely sure that you don't convey not even a shred of incorrect information to viewer and be precise. If you will use an explanation which is not technically correct but you assume your interpretation will help viewer to understand the concept better make sure to point it out. This can take enormous amount of time.
Second is to practice narration and vocalization. This is quite demanding and will take time to get comfortable with it. Watch your breathing, pops and crackles.
Explaining something and building it at the same time is exponentially more difficult and it gets worse if subject is technically challenging. One more thing to consider in your (and my) case is to ensure what you are conveying is in the spirit of language and topic covered.
Third is editing. The better you do the ones above, less time will be spent editing. Depending on how refined you want to go this stage can take most of the production time.
Fourth, but not least important is to keep checks and checkpoints. Save often, remember where your cursor was in UI once you paused, recording app can crash, double check your mic every time, mute your cellphone etc. Make a list which you have to go through before recording. Don't hesitate to be detailed regarding scene files to a point such as "Saved_at_5:32_prior_to_when_I_extrude_a_polygon_cursor is over_this_node" style.
And good luck mate, you are taking quite a challenge now 🙂