The flipped classroom has a well-known failure mode, and every teacher who has tried it has met it. You assign the video. You plan the lesson on the assumption it was watched. Roughly a third of the room did not watch it, another third watched it while doing something else, and you find this out eleven minutes into a discussion that is not happening.
The fix is not a lecture about responsibility. The fix is a five-minute check at the start of class that tells you, before you commit to a plan, what the room actually knows.
The loop
- Assign the video as homework.
- Paste its URL at xyzyoutube.com and choose Quiz.
- Give the questions in the first five minutes of the next class.
- Teach against the results.
Step two takes about a minute. You get multiple-choice questions, five options each, one correct, with an explanation — generated from the video's transcript, so they test what was in the assigned material rather than what you assumed was in it.
Pick the level to match the class: Beginner checks that students watched and caught the stated facts, Intermediate tests whether they followed the reasoning, and Expert pushes on implications and edge cases for a seminar or an advanced group.
That last point matters more than it sounds. Questions written from memory a week after you found the video tend to test what you remember being important. Questions generated from the transcript test what the video said.
What the results tell you
The score is the least interesting output. The distribution is where the value is.
If everyone misses question four, question four is not a student problem. It is a signal that a specific ninety seconds of the video was unclear, and you now have a precise target for the next fifteen minutes of class. That is a better use of a lesson than a general review of everything.
If the class splits cleanly — half right, half wrong — you are usually looking at a genuine conceptual fork, which is often the best possible thing to build a discussion around.
If nearly everyone gets nearly everything, the video worked. Move on, and use the time you just saved.
Beyond the quiz
A few other formats fit teaching workflows:
Mind Map turns a video into something close to a lesson outline. The branches are your sections. Useful when you are building a unit around someone else's talk.
Q&A pairs give you open-ended discussion questions rather than multiple choice. Better for seminar-style teaching and for revision packs students take away.
Lecture Notes produce a structured handout from the video, which is worth giving to students who need the material in writing — including anyone who processes text more easily than audio.
Translation across 70+ languages opens up source material you could not otherwise assign, and gives students working in a second language a version they can actually study from.
Review before you grade
Worth being direct about this, because it is the one place the workflow can embarrass you.
Generated questions come from auto-generated captions, and captions garble numbers more reliably than they garble words. A question built on a misheard figure will look completely reasonable and be wrong. Before anything counts toward a grade:
- Check any question containing a number against the video.
- Read the three wrong options on each question. If one is obviously silly, the question is easier than you think. If one is arguably correct, remove the question.
A minute of this per quiz. It is the same review you would give a colleague's question bank, and for the same reasons.
Where it fits in a week
The teachers who get the most out of this tend not to use it for high-stakes assessment at all. They use it as a fast, low-stakes, almost disposable instrument: a two-minute entry ticket, an exit check, a starter while attendance is taken. Cheap enough to run often, which is precisely what makes retrieval practice work.
Try it on the video you have already assigned this week — the quiz will take less time to generate than reading this took. Start here.