If everyone passes your training quiz first time, the usual interpretation is that the training worked. More often it means the questions were answerable without having learned anything.
Recognition versus retrieval
There's a large difference between recognising a correct answer and producing one.
Recognition is easy and feels like knowing. Shown four options, one of which is obviously the sensible corporate answer, most people select it without engaging with the material at all. The score is high and nothing was learned.
Retrieval — pulling an answer out of memory — is harder, feels worse in the moment, and is what actually consolidates the material. This is one of the more robust findings in learning research, and it's routinely ignored in workplace training because retrieval questions produce lower scores and lower scores look like a problem.
The obvious-answer trap
Most workplace quiz questions have a giveaway option. "What should you do if you receive a suspicious email?" with options including "report it to the security team" and "forward it to everyone in your department" isn't testing anything.
The fix is making the distractors plausible. Every option should be something a reasonable person might actually do. If a question can be answered correctly by someone who skipped the module — because one option is obviously the responsible-sounding one — it isn't measuring anything.
Test the decision, not the text
The weakest questions ask what the module said. The strongest present a situation and ask what to do.
Compare "According to the policy, what is the expense approval threshold?" with "A supplier sends an invoice slightly above your usual approval limit and asks for same-day payment. What do you do?" The first tests whether someone can find a number. The second tests whether they'll behave correctly under mild pressure, which is the thing you actually care about.
Scenario questions are also more interesting to answer, which matters for completion.
Keep it short
Two to four questions per short module is right.
Longer assessments push people into optimising rather than thinking — guessing, working backwards, or clicking through to get to the end. A brief check at the end of a six-minute module reads as a natural close. A twenty-question exam reads as an obstacle, and behaves like one.
Let people be wrong
A wrong answer is the most useful moment in the whole module, and most training wastes it.
If getting it wrong produces "Incorrect. Try again," nothing happens except irritation. If it produces a short explanation of why that answer is wrong and what the right reasoning is, the person learns more from the mistake than they would have from getting it right.
This also means it's fine — good, even — for people to fail questions. A quiz where everyone scores 100% has told you nothing about anyone.
Spacing beats cramming
If material genuinely matters, ask about it again later. A question in module five that revisits something from module two costs nothing and does more for retention than any amount of emphasis in the original module.
This is straightforward when modules are delivered over days rather than completed in one sitting — the spacing happens naturally as a property of the delivery.
What to do with the results
Look at the questions everyone gets wrong. That's usually a content problem rather than a people problem — the module was unclear, or assumed something the audience didn't know.
Look also at questions everyone gets right immediately. Those are candidates for rewriting, because they aren't measuring anything.
Generating quizzes
Writing good questions is slow, which is why most training quizzes are weak.
Poplearn writes the quizzes automatically as part of converting a document into modules — /courses new handles both. Reviewing the generated questions against the principles above is worth a few minutes: check that the distractors are plausible and that the question tests a decision rather than a sentence.
Modules and their quizzes arrive as Slack direct messages, spread over days, which gives you the spacing effect without planning for it. Free under 20 seats.
