Quiz drop-off analysis: find the question that loses visitors
Quiz drop-off analysis shows which question people quit on. Learn to read the per-question view, tell bad questions from hard ones, and fix them.
By Bhumi Avaiya. 9 min read. .
To find the question that loses your visitors, open the per-question drop-off view for your quiz and look for the step where the number of people falls furthest compared with the step before it. That step is your suspect. This is quiz drop-off analysis in one sentence, and the rest of this post is about not fooling yourself when you do it.
A high drop is a clue, not a verdict. The question might be badly written, or it might just be long, hard or badly placed. The fix is different each time.
What the drop-off view actually tells you
Quizably records analytics for each quiz: submissions, completion rate, average score, result distribution, answer distribution and per-question drop-off. You can export the data, and there is a site-wide overview as well as a per-quiz view.
The per-question drop-off view answers one narrow question: out of the people who reached question 3, how many went on to question 4? It does not tell you why they left. You supply the why.
Open it in your own dashboard and check how it is laid out there. Then read it the way you would read a queue: start at question 1, walk forward, and note where the line gets noticeably shorter.
Look at the step, not the total
The overall completion rate tells you that something is wrong. It cannot tell you where. Two quizzes can have the same completion rate and completely different problems: one loses everyone at question 2, the other loses a few people at every step.
A steady, small loss at each step is normal. Nobody finishes everything they start. A sharp cliff at one step is what you are hunting.
Watch the start of the quiz too
If many people see the quiz but few answer question 1, the problem is usually the title, intro or page around it, not a question. The drop-off view starts counting when a visitor begins the quiz, not when the page loads, so people who never press start do not appear in it. Page views sit in the analytics CSV as total_views, which you can set against your starts. One caveat: a visitor who resumes or starts over after a reload creates a new submission, and the earlier one shows as drop-off at the question where they left. That can lower your completion rate without anyone quitting for good. See how to improve your quiz completion rate for that.
A bad question versus a hard or long one
This is the part most people skip. Before you rewrite anything, sort the suspect question into one of these groups.
| What you see | Likely cause | First move |
|---|---|---|
| Cliff at a question with a confusing answer list | Bad wording | Rewrite the question and answers |
| Cliff at a question that needs thought or research | Hard question | Make it easier or move it later |
| Cliff at a short-text question | Effort | Consider a choice question instead |
| Cliff at the email form | Ask, not question | Change timing or fields |
| Cliff at the last question | Fatigue or a dull ending | Shorten the quiz or lift the pace |
| Slow leak everywhere | Quiz is too long | Cut questions |
Signs of a bad question
A bad question is unclear, or it offers no answer that fits the person reading it. Watch for these:
- Two answers that mean almost the same thing
- No answer that matches what a real visitor would say
- Jargon your visitors do not use
- A question that asks two things at once
- An answer list so long people have to scroll to see it
Read the question as if you know nothing about your business. If you pause, your visitors paused too.
Signs of a hard question
A hard question is clear but demanding. “How much did you spend on marketing last year?” is clear, and it also makes people leave to check a spreadsheet. Some do not come back.
Hard questions are not always bad. In a trivia quiz, difficulty is the point. The trade-off is that a hard question placed early can chase people away before they are invested. The same question placed later, after a few easy wins, is often fine.
Signs of a long question
A long question is one where the reading load is heavy. A paragraph of setup and four paragraph-sized answers will lose people on a phone, even if every word is reasonable. Cut the setup, shorten the answers, and see whether the drop eases.
Quiz drop-off analysis: four patterns and what they mean
Think of these as starting hypotheses. Your data may disagree, and that is fine.
Early drop: questions 1 and 2
People who leave this early were not yet convinced. Either the promise on the page did not match the first question, or the first question asked for too much.
Start with a question that is fast and a little fun. Save the demanding ones for later. If the title promises “What kind of traveler are you?” and question 1 is about your budget, the mismatch will show up as a cliff.
Middle drop: one question in the middle of the quiz
A cliff at one middle step, with smooth numbers on either side, almost always points at that single question. This is the easiest pattern to fix because the evidence is clean. Read the question, compare it with the ones around it, and look for the signs above.
A drop that spreads across the whole middle is different. It suggests the quiz is long or repetitive, and you should look at the number of questions. The post on how many questions a quiz should have covers that trade-off.
Drop at the lead form
If the form appears before the result, the question “where did they leave?” becomes “did they leave at the form?” A cliff right there means people like the quiz but do not want to hand over their email on these terms.
Try the form in a different position, cut the fields down to what you really need, or make it skippable. The form can be required or skippable, before the quiz or right before the result. The trade-offs of each position are in quiz email form placement.
Drop at the last question
Losing people at the final question feels especially bad because they were so close. The causes are usually one of three: the question is the hardest one, the quiz has simply gone on too long, or the person knows the form is coming and is reluctant.
Check whether the last question is also the one that feels most like homework. If so, swap it with an easier one, and keep your strongest, most curious question near the end so people want to see where it leads.
A diagnosis checklist
Run through this list for the question that drops the most people. Write the answers down. Five minutes here saves a lot of guessing.
- Is there a real cliff? Compare the suspect step with the steps before and after it. If they look alike, there is no cliff.
- Is there enough data? If a handful of people explain the whole drop, wait.
- Is it the first question? If so, check the page and intro first.
- Is it the form step? If so, it is about the ask, not the wording.
- Can you answer it in five seconds? If not, why not?
- Does every visitor have an answer that fits? If not, add one or reword.
- Is it longer than its neighbors? Count the words of the question and its answers.
- Is it a different type from its neighbors? A short-text question in a run of single-choice questions is a speed bump.
- Does it need information people do not have handy? Rephrase it as a range or a feeling.
- Does it show well on a phone? Open it on a real phone, not just a narrow browser window.
If you get to the end and still cannot explain the drop, that is useful too. It tells you to test a change instead of arguing with the data.
Four fixes, in the order I would try them
Rewrite
Start here. It is cheap, reversible, and fixes the largest share of problems. Make the question shorter, use words your visitors use, and make each answer clearly different from the others. If you want a method for this, how to write quiz questions that point to a result walks through it.
Reorder
If the question is fine but demanding, move it later. The builder lets you drag questions into a new order, so this takes a minute. Put quick, easy questions first to build momentum, then place the heavier ones once the visitor is invested.
One caution: if you turned on question randomization, the order is shuffled per visitor, so a question’s position in your drop-off view may not be the same for everyone. Check how your own data behaves if you use that setting.
Split
If one question asks two things, turn it into two easier questions. It feels like making the quiz longer, but the total effort for the visitor often goes down. Two quick taps beat one confusing one.
Remove
Remove a question when the quiz still works without it. Ask yourself what result or follow-up decision this question feeds. If the answer is “nothing much”, cut it. Keep it if a result depends on it, and try the other three fixes first.
How long to wait for enough data
I will not give you a number, because the right one depends on your traffic. Here is a test instead.
Ask: if three more people had left at this question, would my conclusion change? If yes, you do not have enough data. Wait for more visitors, or promote the quiz to more people before you decide.
A clear cliff at a plainly confusing question is worth fixing now. A blurry pattern is worth waiting on.
If traffic to the quiz is thin, the bigger job is getting more people to it. The ideas in quiz metrics to track and how to judge them will help you decide what to look at while you wait.
Record every change and re-check
Change one thing at a time and write it down. A simple note is enough:
- Date of the change
- Which question
- What you changed and why
- What you expect to happen
Then wait for fresh traffic and look at the same step again. If you changed three questions and the drop improved, you will not know which change did it. Worse, you will not know what to undo if it gets worse.
Quizably can export quizzes as JSON, which gives you a simple backup before you make big edits. Take one if you are about to restructure the quiz. The exported analytics are a handy way to keep a before and after record.
If the drop did not move, your guess was wrong. Go back to the checklist and pick the next suspect.
Where to go next
Open your quiz analytics today and find the single biggest step down. Sort it with the table above, pick one fix, and write the change down before you make it. Drop-off is one of several numbers worth watching, and the quiz marketing guide shows how they connect to leads and follow-up.
If you do not have a quiz yet, start with the free plugin and follow the steps in how to create a quiz in WordPress. Publish it, give it some traffic, and come back to this checklist.
Frequently asked questions
What is quiz drop-off analysis?
It is the habit of checking, question by question, where people stop answering your quiz. The goal is to find the one or two places where visitors leave and fix those first.
How long should I wait before I trust the drop-off data?
Wait until each question has been answered by enough people that one odd visitor cannot change the picture. There is no magic number. If a single person leaving would move a question from fine to alarming, you need more traffic first.
Is a drop at the email form a question problem?
Usually not. A drop at the form is about what you ask for, when you ask, and what people expect in return. Test the form position and the fields before you rewrite any question.
Should I delete a question that loses a lot of people?
Only after you have tried to rewrite, reorder or split it. Sometimes a question is hard because it is important. Remove it if the quiz still works without it and the drop does not fade after a rewrite.
How often should I check drop-off?
Check after a new quiz has had real traffic, then again after every change you make. Once the quiz is stable, a quick look once a month is plenty.