01
71%
Mobile visitors do not reach the first price.
71% of mobile visitors leave the home page before the first price. On mobile the header has no booking button; it sits inside the menu.
Scroll maps · 32,780 mobile sessions
01 Case · Conversion optimisation
More consultation bookings from the same traffic.
Névé is an aesthetic medicine clinic on Rothschild Boulevard in Tel Aviv. Visits to the site were steady; bookings were not. Over sixteen weeks we measured where visitors left, tested three changes, and kept the two that worked.
Illustrative data. Névé is a concept clinic.
- Booking rate
- 2.10%2.74%
- Bookings per four weeks
- 1,0121,319
- Sessions per week
- ≈ 12,100 unchanged
- Experiments
- 3 run · 2 shipped
The brief
- Client
- Névé, aesthetic medicine. One clinic, one doctor, a site in Hebrew and English.
- Goal
- More completed consultation bookings per session, with no change to advertising spend.
- Primary metric
- Completed bookings divided by sessions.
- Guardrails
- Cancellations within 48 hours, page speed (LCP), and bounce rate must not get worse.
- Period
- Sixteen weeks: three weeks of research, thirteen weeks of testing.
- Tools
- GA4 with server-side events, a heatmap and recording tool, a split-testing tool.
02 Research
Where visitors leave
Four weeks of sessions before any change. Each step shows how many visitors continued and how many left.
Session-scoped funnel. Steps in order: any page, a treatment page, the booking form opened, a booking completed.
Sessions to booking 2.10%
What the sessions showed
02
78%
Almost every booking starts in the header.
78% of booking starts come from the header button on desktop. The “Book this treatment” button in the treatment panels starts 9%.
Click maps · 2,820 booking starts
03
6.8%
People click the “From” price, which is not a link.
6.8% of sessions that reach the treatment panels include a click on the price. In recordings, visitors then scroll looking for the cost of a course and what it includes.
Dead-click report · 240 recordings reviewed
04
38%
The form loses people at the time picker.
38% of abandoned bookings stop at the step where a time is chosen. The picker opens on a full month with no free slots marked.
Form analytics · 1,808 abandoned starts
03 Heatmaps
Clicks and scroll depth on the current site
The Névé home page as it stands, with four weeks of clicks and scroll depth laid over it. Scroll inside the frame; the ruler shows the share of visitors who reached each point.
Page
Layer
Legend
FewerMore
Illustrative data drawn over a real capture of the Névé demo site.
04 Experiments
Three hypotheses, three A/B tests
Each test came from one finding in the research. Traffic was split 50/50, the sample size was fixed before launch, and each test ran in whole weeks.
EXP-01
Hypothesis
| Variant | Visitors | Bookings | Rate |
|---|
How the numbers are calculated
Significance. Two-sided two-proportion z-test with a pooled standard error; a result counts at p < 0.05. The 95% interval for the difference uses the unpooled standard error.
Chance B beats A. Each rate gets a Beta(1 + bookings, 1 + visitors − bookings) posterior; the probability that B’s rate is higher is computed with a normal approximation to the two posteriors.
Sample size. Set before launch for 80% power at α = 0.05, for the smallest relative change worth detecting (MDE). Tests were not stopped early.
Every figure on this page is computed in the browser from the visitor and booking counts shown.
05 Results
Sixteen weeks, week by week
Booking rate against sessions. Traffic stayed flat; the rate moved when each winning variant went to all visitors.
- Booking rate
- Sessions
- Test running
- Winner goes to all visitors
Experiment log
| ID | Change | Page | Weeks | Result | Decision |
|---|---|---|---|---|---|
| EXP-01 | Booking bar fixed to the bottom of the screen | All pages · mobile | 4–8 | +30.6% | Shipped |
| EXP-02 | Full price and what it includes, beside the button | Treatment panels · all | 10–13 | +20.1% | Shipped |
| EXP-03 | Doctor’s credentials in the hero | Home · all | 10–13 | +4.3% | Not shipped |
| EXP-04 | Time picker shows the next five free slots | Booking form · all | 15– | — | Running |
| EXP-05 | Credentials beside the price, not in the hero | Treatment panels · all | — | — | Planned |
Rollout checklist
What happens before a winning variant replaces the original.
- Built into the site’s own code, test script removed
- Hebrew and English, right-to-left checked
- Tested on iOS Safari, Android Chrome and desktop
- Booking events verified end to end in GA4
- Page speed within budget: LCP under 2.5 s
- Guardrails unchanged: cancellations, bounce
- 5% holdout kept on the original for four weeks
- Result written up in the monthly report
06 Scope
What the service includes
- 01
Audit
The funnel, the pages and the booking form, measured and written up.
- 02
Analytics setup
Events for every step of the booking, checked end to end.
- 03
Heatmaps & recordings
Clicks, scroll depth and sessions, reviewed every week.
- 04
A/B testing
Hypotheses, sample sizes and tests run to a fixed plan.
- 05
Design & build of winners
The winning version, designed and built into the site.
- 06
Monthly report
What was tested, what changed, and what runs next.
