"The quiz works" and "the quiz brings in cheaper leads than a form" are two different claims, and the second one has to be proven with numbers. The catch is that a quiz has no classic "thank you" page, which is how conversions are usually tracked: everything happens on one screen. So measurement starts with setting up events, not with opening reports.
Here is what to set up, which metrics to watch, and how to read session recordings so that changes to your quiz actually make sense.
Four events the quiz sends
LeadForms sends four JavaScript events to your analytics on its own:
- leadforms_open — the quiz was opened;
- leadforms_start — the visitor clicked the button on the start screen;
- leadforms_stepN — question number N was completed;
- leadforms_finish — the contact form was submitted.
You don't need to write any code: the events fire automatically, and your job is to register them in analytics. In Google Analytics 4 they arrive as custom events with these exact names, and you should mark leadforms_finish as a key event (conversion) so it shows up in conversion and acquisition reports. Type the event name exactly as shown above, with no changes. The step-by-step setup is in the guide to connecting Google Analytics.
Another common stumbling block: if the quiz opens via a direct link, it lives on a different domain from your website. In that case you need to set up cross-domain measurement, otherwise analytics will count those visits as external traffic and the data won't connect.
Analytics conversions and ad platform events are not the same thing
This is where confusion always shows up, because the events are the same but they do different jobs.
Conversions in analytics are for you: viewing reports, comparing sources, understanding where people drop off. It's an analysis tool. It doesn't change how your ads run.
Events in the ad platform are for the algorithm: they tell the campaign who to look for. If you don't set a target event, the platform optimizes for clicks and brings you people who happily open the quiz and don't answer a single question.
The practical takeaway: setting up analytics without setting up conversions in the ad platform means seeing the problem very clearly and doing nothing about it. You need both. In Meta this is done with the pixel, which you connect by ID without pasting any code — see the guide to connecting the Meta pixel. What to choose as the target event and how to build audiences from quiz events is covered in the article on advertising a quiz in Meta Ads.
Metrics that actually tell you something
These four events give you the whole picture. What matters is not the absolute numbers but the transitions from one step to the next.
Start rate: start / open
How many of the people who opened the quiz clicked the button on the start screen. This is your grade for the first screen: the headline, the promise, the image. A drop here means the traffic is wrong or the ad's promise doesn't match what people see.
Completion rate: finish / start
How many of those who started made it to the contact form. This grades the quiz itself: length, wording, question order.
Lead conversion: finish / open
The bottom-line number, the one you compare with a regular form. The market benchmark: a website form converts at 2–5%, a quiz at 15–25%. If your quiz is well below that range, the problem is almost always in one of two places: the start screen or too many questions.
Cost per lead
For paid traffic this is the key number: campaign spend divided by the number of leadforms_finish events. Calculating it per click or per open makes no sense: a cheap click and an expensive lead coexist just fine.
The step-by-step funnel: exactly where people drop off
The leadforms_stepN event gives you what a regular form can't: a breakdown by question. In quiz analytics you can see how many people reached step two, three, four, and where the flow breaks.
How to read the drop-offs:
- drop at the first question — people didn't find their situation among the options. The wording is too general or, the opposite, too narrow;
- drop in the middle — one specific question creates friction. It's almost always a blunt "budget" question, a free-text field instead of options, or a question people simply don't know the answer to;
- steady attrition at every step — the quiz is too long. The 4–6 question benchmark comes from exactly this: every extra step costs you part of your audience;
- reached the form and didn't submit — too many fields, or it's unclear what people get in return.
One key principle: fix one thing at a time. If you change the headline, the question order and the form copy all at once, you won't know what worked.
Session recordings: what the numbers don't show
Reports show where people left. Recordings show why. These are real sessions: cursor movement, scrolling, clicks, pauses, typing. The simplest tool for this is Microsoft Clarity: it's free and gives you session recordings and click maps.
What to look for in quiz recordings:
- long pauses on a question — the wording is unclear and people reread it;
- the cursor going back and forth between options — the options overlap in meaning, so it's impossible to pick one;
- going back — people change their answer because they only understood the question after seeing the next one;
- clicks on inactive elements — an image or text looks like a button but can't be clicked;
- started typing a phone number and deleted it — the most valuable signal: they reached the form, but something stopped them. Usually it's the lack of a clear promise next to the field.
You don't need to watch every recording. It's more useful to filter: sessions where leadforms_start fired but leadforms_finish didn't, meaning people who started and abandoned. Ten of those recordings are usually enough to spot a recurring pattern.
Look at mobile sessions separately. On a phone, different things break than on a desktop: the question doesn't fit on the screen, a finger misses the option, the keyboard pops up and covers half the quiz. In the overall report this gets averaged out and looks like "mediocre conversion".
Sources: which traffic brings leads
UTM tags arrive with each lead, so in LeadForms you can see which campaign and which ad every contact came from. That shifts the conversation from "we're getting more leads" to "this combination pays for itself and that one doesn't".
The breakdown by source also explains odd drops. The same quiz can show very different completion rates on search and on social, simply because in the first case people were looking for you and in the second they were scrolling their feed. It makes sense to compare a source with itself over time, not sources with each other.
Common mistake: counting leads instead of sales
A quiz is optimized for the number of leadforms_finish events, and that's fine, up to a point. After that, things get out of balance: the simpler the quiz, the more leads, but the lower their quality.
So alongside conversion, watch what happens next: the share of contacts who pick up, the share of inquiries that are actually a fit, the share that end up paying. If leads doubled and sales stayed flat, the quiz is collecting the wrong people, usually because the questions that filtered out poor fits were removed.
A CRM connection shows this: LeadForms sends leads to Kommo (formerly amoCRM) and Bitrix24, and via webhook to any other system, and the quiz answers land in the contact record. From there on, you compare by sales, not by contacts.
Where to start if nothing is set up
The whole sequence takes one afternoon:
- connect Google Analytics 4 and Microsoft Clarity in the quiz settings and, if the quiz lives on a direct link, set up cross-domain measurement;
- register the four events and mark the form submission as a key event;
- if you run paid traffic, connect the ad platform's pixel separately and choose the form submission as the target event;
- go through the quiz yourself with blockers turned off and confirm the events were recorded;
- give it a week to collect data, and only then look at the reports;
- find the biggest drop in the funnel and fix one thing.
Testing without blockers isn't a formality: browser extensions stop analytics from sending the event, and you'll conclude the integration is broken even though it's set up correctly.
FAQ
Why aren't the events showing up in analytics?
Three common causes: the event name isn't spelled exactly the same as in the quiz, the quiz opened via a direct link is on another domain and cross-domain measurement isn't set up, or the test was done in a browser with an ad blocker. Test by going through the whole quiz and submitting the form: leadforms_finish fires at exactly that moment.
What's the difference between analytics conversions and ad platform events?
Analytics conversions are for your analysis; ad platform events are for the algorithm to learn. Having analytics set up without conversions means you see the problem, but the campaign keeps optimizing for clicks. You need both.
What quiz conversion rate is considered normal?
The benchmark is 15–25% versus 2–5% for a regular form. But everyone has their own starting point: how your quiz changes after each fix matters more than comparing with other people's numbers, because the niche and the traffic source have a huge impact on conversion.
How much data do I need before drawing conclusions?
Looking at reports with a dozen visits is pointless: random variance hides any effect. Wait until you have a meaningful number of quiz sessions, and compare full weeks, not individual days.
What should I look at first, the funnel or the recordings?
The funnel first: it shows which step has the problem. Then the recordings for that step: they explain the cause. The other way around, you'll just burn time watching sessions.
Do I need all this if I'm not running ads?
Yes. Even with organic traffic, the step-by-step funnel shows which questions scare people off. And when you do launch ads, you'll already have a baseline to compare against; otherwise you'll be judging the first weeks of the campaign blind.
