Finding and fixing where visitors drop off

When fewer people finish your popup or quiz than you’d like, the Analytics dashboard shows you exactly where they quit, which step they gave up on, and which answers came right before they bailed. This is how you turn a leaky quiz into one that converts, by fixing the worst step first instead of guessing.

Before you start

The drop-off tools only appear when you look at one popup at a time.

Open Analytics from your admin.
In the Popup dropdown at the top, pick a single popup. The dropdown defaults to All popups, and while that’s selected you only see cross-popup summary cards. The funnel, hotspots, and timing sections stay hidden.
Check the date range buttons in the same bar: Today, Yesterday, 7d, 30d, 90d, All, or Custom. The default is the last 30 days. If you see a No data yet message, widen the range (try 90d or All).

Some sections only appear when there’s something to show. The Findings panel, Abandonment Hotspots, Time on Step, and the form-friction tables each render only when the popup has the relevant data, so a brand-new or very clean popup may show fewer sections than a busy one.

How to find where visitors drop off

Start with the Findings panel. Once you’ve picked a single popup, look near the top for the Findings panel. It only appears when there are issues worth flagging, so if you don’t see it, nothing crossed the alert thresholds for this popup and date range. When it’s there, its subtext reads “Top issues to investigate, ranked by severity and visitor impact.” It automatically surfaces the worst problems first, so this is your shortlist. It shows up to 8 findings. If there are more, you’ll see a note that the rest are “ranked lower (not shown). Address the top 8 first.”
See the cliff in the Step Funnel. Scroll to Step Funnel (“Drop-off at each step of your popup flow”). Each bar is one step, in editor order, showing how many different visitors reached it. The percentage under each bar compares it to your first step. A big gap between two bars is the cliff. If a step has dismissals, a translucent red overlay marks them, labeled Dismissed at step.
See which step they gave up on in Abandonment Hotspots. This section (“Where visitors drop off”) attributes each non-converting visitor to the last step they viewed. Rows are sorted with the biggest leak at the top, each showing the abandon count and its share of all abandoners. The top row is the step to fix first.
Read the top answer before bail. Under a hotspot row you may see a line like “Top answer before bail: ’…’ (X%).” This is the answer choice most of the bailing visitors picked right before quitting. It’s a strong hint that the answer, or what comes after it, is the problem.
Check timing in Time on Step. The Time on Step table shows the median and p90 time visitors spend on each step. A long time can mean a confusing question or too much to read. The Outliers column counts gaps over 10 minutes (someone left a tab open), which are dropped so the numbers reflect engaged visitors.
Drill into specific answers. Under each survey question you’ll see a cohort table. The Dropoff column shows the share of people who picked that answer but never advanced past the step, with the sample size in parentheses. The vs avg column shows how that answer’s cohort buys compared to the question’s average. This is observational: it tells you which answer cohort buys more or less, not that the answer caused it.
Catch form problems. For popups with forms, Form Field Friction lists fields where visitors clicked in, then clicked away while still empty. Form Validation Errors lists fields where a submission failed, grouped by error. Both point to fields people don’t want to fill in.

What to change

The dashboard tells you where and gives a hypothesis. Deciding the fix is up to you. Common moves:

A step with the biggest hotspot share: rewrite the question to be clearer, shorter, or lower-commitment.
A slow step in Time on Step: cut text, simplify the choices, or split it into two.
A field high in Form Field Friction: remove it or make it optional.
A repeated error in Form Validation Errors: check the field’s rules and make the expected input obvious.

You make these changes in the editor for that popup. Naming your steps there also gives every analytics section clear labels instead of generic ones.

Tips

Step Funnel and Abandonment Hotspots measure different things. The funnel shows how many visitors reached each step (where the cliff is). Hotspots show which step people gave up on. On a branching quiz the funnel counts can mislead, so trust hotspots for the branching-safe read.
Hotspots only count people who didn’t finish. Anyone who submitted is excluded. A busy step with high completion won’t show up, and that’s correct.
Some data only exists from April 29, 2026 onward. “Top answer before bail” and the per-answer Dropoff column come from answer-tracking that’s forward-only. Older or low-traffic popups will show no top-answer line and a dash in Dropoff.
Small samples are hidden on purpose. A Dropoff cell shows a dash until at least 5 people picked that answer. Low-volume popups will show many dashes. Widening the date range helps.
A high Outliers count next to a small Sample in Time on Step means that step’s timing is unreliable. Read it with caution.
Unnamed steps get generic labels like “3. Quiz Question”. Name your steps in the editor and the names flow into every analytics section.
A routing or branching step won’t appear in hotspots if it’s flagged as routing-only. That’s intentional. Passing through it doesn’t count as viewing it.

FAQ

I picked a popup but still see nothing. Check the date range. If it says No data yet, that popup had no activity in the window. Pick another popup or widen the range with 90d or All.

Why doesn’t a step show in Abandonment Hotspots? Either no non-converting visitor stopped there, or the step is set to routing-only, which is excluded from hotspots by design.

Why is the Dropoff column full of dashes? Each answer needs at least 5 people before a number shows, and the answer-tracking data only goes back to April 29, 2026.