Funnel analytics
You will read the report on your funnel and see at which step people leave without paying.
Before you start
- A published funnel with visitors. Until then the section shows “No data yet”. Publish a funnel needs the Pro or Business plan, so on the free plan there is nothing to count.
- Any role. A project member sees numbers only for their own quizzes and paywalls; others see all.
Step 1. Set the period and the slice
Open “Analytics” in the project menu. It has four tabs: “Statistics”, “Answers”, “Leads”, “Cohorts”. Start with “Statistics”: it has the funnel and the table.
The default period is the last 30 days; there are also “Today”, “Last 7 days”, “All time” and custom dates. We keep data for a year, so “All time” means the last twelve months.
Other panel fields narrow the selection and combine: pick a country and a campaign to see those matching both.
Step 2. Read the funnel
Below the filters are six cards, one per step. A card counts sessions: someone who came twice in an evening counts twice; reloading the page five times in one visit counts once.
| Step | Counted when |
|---|---|
| Funnel opened | the quiz page opened |
| Quiz started | they pressed the button on the first screen |
| Quiz completed | they reached the end of the quiz by any path |
| Paywall shown | the paywall opened; the thank-you screen after payment does not count |
| Checkout started | they pressed buy, and we sent them to Stripe |
| Payment completed | Stripe confirmed the money went through |
Under each card is its share of the previous step. The lowest share is where you lose the most people.
If the project has “On your own” selected, the three money steps are replaced by one — “Left for checkout”, when we sent the person to your page. From there you count. Such a project has no “Cohorts” tab either: cohorts are counted by buyers.
Step 3. Find where the quiz breaks off
The funnel does not say at which question people left. Dig two ways.
The “Quiz 50%” and “Quiz 90%” columns in the table below: sessions that reached the middle and nearly the end of their path. A gap between “Quiz started” and “Quiz 50%” means people leave in the first half.
The “Answers” tab shows each question, how many answered it and the split by options. Where the count drops sharply is the break-off point. Free-text questions are left out entirely — they may hold personal data.
The report “how many people on which screen” is in “Audit”, “Traffic” tab — along with the path of a specific visitor.
Step 4. Compare sources
Below the funnel is a table. The “Group by” field sets its first column: country, campaign tag, quiz, domain, A/B test. The “By day” switch at the top splits it by days. “Columns” turns metrics on and off; “Export CSV” exports exactly what is on screen.
The campaign tag comes from the ad address. Add tags to the link of any published funnel — no A/B test needed, we read them from any of our public pages. Visits without tags go into one row, “Direct traffic / no UTM”.
| Column | What it holds |
|---|---|
| Views | openings of quiz pages and the paywall, repeats included |
| Uniques | how many different people came |
| Quiz 50%, Quiz 90% | sessions that reached the middle and nearly the end of the quiz |
| Paywall | how many times the paywall opened |
| Registrations | sessions in which the person left an email |
| Purchases | sessions with a successful payment: two purchases in different visits count twice |
| Revenue | the sum of those payments |
| Installs | how many times the app reported its first opening after payment |
| Traffback | sessions where the person pressed “back” in the browser and we sent them to the app store instead of off the site (setup) |
The same button turns on hidden ones: “Sessions” (visits instead of people), “Purchase conversion” (purchases per unique visitor), ARPU (revenue per unique visitor) and ARPPU (revenue per buyer).
Your AppsFlyer or Adjust tracker can also report installs to us, into “Installs”. You set this up in the tracker's account, not the developer: the address is in “Settings”, “App connection” tab, option “Through an ad tracker”.
Why the numbers differ
- Test payments do not count. The “Show test payments” checkbox under the filters brings them back into “Purchases” and “Revenue”.
- Preview is not counted. Clicking through the funnel from the editor leaves the numbers unchanged — check on the published address.
- The answer can be a minute old. The “Refresh” button recounts.
- Views, sessions and uniques are three different numbers. One person in an evening gives one unique visitor, two visits and five page openings.
- The ad account counts its own. Our “Views” are openings of our pages, not ad clicks: whoever clicked and left before loading never reached us.
- The network does not see purchases by Europeans without consent. A European visitor who did not press “Accept” on the banner is not in the ad account, but we count their payment — Ad networks and pixels, section “Tracking consent”.
- “Total” at the bottom of the table is not the sum of the rows. Someone who came on two days is in both “By day” rows, but once in “Total”.
- “Revenue” in the table rows holds only first payments. Renewals and refunds come without a country or campaign tag, so they appear only in “Total”.
Two more tabs
“Leads” — a named list of those who left an email: which quiz they came from, when they last visited, which ad brought them. “Export CSV” above the list exports it, including the utm_term and utm_content tags missing in “Statistics”.
“Cohorts” — ad payback: people are grouped by the date of their first visit, showing what each group brought on day one and by day 7, 30, 60, 90, 180 and 360. Spend is typed by hand into the “Spend” column and kept only in this browser: another computer will not show it.
Next
- Emails to customers — how to bring back those who reached payment and did not pay.
- Store — who pays you and how much they bring.
- A/B tests — how to compare two funnel variants on one ad address.
- Ad networks and pixels — so that Meta and TikTok see sign-ups and payments and learn from them.